CN103117975B - System of compensating MU-MAS communications and dynamically adapting communication characteristics of MU-MAS communication system - Google Patents

Abstract

A system for compensating for MU-MAS communication and dynamically adapting to communication characteristics of the MU-MAS communication system is described. The system for compensating MU-MAS communications includes one or more code modulation units for coding and modulating information bits for each of a plurality of wireless client devices to generate coded and modulated information bits; one or more mapping units for mapping the encoded and modulated information bits into complex symbols; and a MU‑MAS frequency/phase offset-aware precoding unit for adopting feedback from the wireless The channel condition information obtained by the client device is used to calculate MU-MAS frequency/phase offset-aware precoding weights, and the MU-MAS frequency/phase offset-aware precoding unit uses the weights to the complex symbols obtained from the mapping unit Precoding is performed to pre-cancel frequency/phase offset and/or inter-user interference.

Description

Compensating MU-MAS communication and dynamically adapting to the communication characteristics of MU-MAS communication system system

This application is a divisional application of an invention patent application with the filing date of August 20, 2008, the application number of 200880102933.4, and the title of "System and Method for Distributed Input and Distributed Output Wireless Communication".

priority claim

This application is a continuation of Application No. 10/902,978 filed on July 30, 2004.

technical field

The present invention generally relates to the field of communication systems. In particular, the present invention relates to systems and methods for distributed-input and distributed-output wireless communications using space-time coding techniques.

Background technique

Space-time encoding of communication signals

Space multiplexing and space-time coding are relatively recent developments known in wireless technology. Since several antennas are used at each terminal, a special type of space-time coding is called "multiple-input multiple-output" (MIMO). By using multiple antennas for transmission and reception, multiple independent radio waves can be transmitted simultaneously in the same frequency range. The following article provides an overview of MIMO.

IEEE JOURNAL ON SELECTED AREAS INCOMMUNICATIONS, VOL.21, NO.3, APRIL 2003 by IEEE Member David Gesbert, IEEE Member Mansoor Shafi, IEEE Member Da-shanShiu, IEEE Member Peter J. Smith and IEEE Senior Member Ayman Naguib to Practice: An Overview of MIMO Space-Time Coded Wireless Systems".

IEEE Member David Gesbert, IEEE Member Helmut Bolcskei, Dhananijay A. Gore, and IEEE Member Arogyaswami J. Paulraj IEEE TRANSCTIONS ONCOMMUNICATIONS, VOL.50, NO.12, DECEMBER 2000: "Outdoor MIMO Wireless Channels: Models and Performance Prediction".

Basically, MIMO technology is based on the application of spatially distributed antennas for generating parallel spatial data streams within a common frequency band. Radio waves propagate in such a way that individual signals can be separated and demodulated at a receiver even though they are transmitted in the same frequency band, which can result in multiple statistically independent (that is, effectively separated) communication channels. Thus, in contrast to standard wireless communication systems that strive to suppress multipath signals (i.e., multiple time-delayed signals at the same frequency with modifications in amplitude and phase), MIMO can rely on uncorrelated or weakly correlated multipath signals, at Higher throughput and improved signal-to-noise ratio are achieved within a given frequency band. Examples show that under the condition of equivalent power and signal-to-noise ratio (SNR), MIMO technology achieves higher throughput (throughput), while traditional non-MIMO systems can only achieve lower throughput. The feature is described on a page titled "What MIMO Delivers" on the Qualcomm (QCOM is the largest supplier of wireless technology) website http://www.cdmatech.com/products/what mimo delivers.jsp: "MIMO is the only multiple antenna technique that increases spectral capacity by delivering two or more times the peak data rate of a system per channel or per MHz of spectrum. To be more specific, for wireless LAN or applicationsQUALCOMM's fourth generation MIMO technology delivers speeds of 315Mbps in36MHz ofspectrum or 8.8Mbps/MHz.Compare this to the peak capacity of 802.11a/g(even with beam-forming or diversity techniques)which delivers only 1Mbps ofspectrum/MHz.17 ".

In general, MIMO systems face a practical limit of less than 10 antennas per device (and thus less than 10× throughput improvement in the network) for several reasons:

1. Physical constraints: There must be sufficient separation between MIMO antennas on a given device so that each receives a statistically independent signal. Although improvements in MIMO throughput can still be seen even at antenna spacing of wavelength fractions, the efficiency deteriorates rapidly when the antennas are closer together, which results in a lower MIMO throughput multiplier.

See for example the following references:

[1] D.-S.Shiu, G.J.Foschini, M.J.Gans, and J.M.Kahn, "Fading correlation and its effect on the capacity of multielement antenna systems," IEEE Trans, Comm., vol.48, no.3, pp.502 -513, Mar. 2000.

[2] V.Pohl, V.Jungnickel, T.Haustein, and C.von Helmolt, "Antennaspacing in MIMO indoor channels," Proc.IEEE Veh.Technol.Conf., vol.2, pp.749-753, May2002.

[3] M. Stoytchev, H. Safar, A.L. Moustakas, and S. Simon, "Compactantenna arrays for MIMO applications," Proc.IEEE Antennas and Prop.Symp., vol.3, pp.708-711, July 2001.

[4] A.Forenza and R.W.Heath Jr., "Impact of antenna geometry on MIMOcommunication in indoor clustered channels," Proc.IEEE Antennas and Prop.Symp., vol.2, pp.1700-1703, June 2004.

In addition, for small antenna spacing, the coupling effect between each other may degrade the performance of the MIMO system.

See for example the following references:

[5] M.J.Fakhereddin and K.R.Dandekar, “Combined effect of polarization diversity and mutual coupling on MIMO capacity,” Proc.IEEE Antennas and Prop.Symp., vol.2, pp.495-498, June 2003.

[7] P.N.Fletcher, M.Dean, and A.R.Nix, "Mutual coupling in multi-elementarray antennas and its influence on MIMO channel capacity," IEEE Electronics Letters, vol.39, pp.342-344, Feb.2003.

[8] V. Jungnickel, V. Pohl, and C. Von Hel molt, "Capacity of MIMO systems with closely spaced antennas," IEEE Comm. Lett., vol.7, pp.361-363, Aug.2003.

[10] J.W.Wallace and M.A.Jensen, "Termination-dependent diversity performance of coupled antennas: Network theory analysis," IEEE Trans. Antennas Propagat., vol.52, pp.98-105, Jan.2004.

[13] C. Waldsch midt, S. Schulteis, and W. Wiesbeck, "Complete RFsystemmodel for analysis of compact MIMO arrays," IEEE Trans. on Veh. Technol., vol.53, pp.579-586, May 2004.

[14] M.L.Morris and M.A.Jensen, "Network model for MIMO systemswithcoupled antennas and noisy amplifiers," IEEE Trans.AntennasPropagat., vol.53, pp.545-552, Jan.2005.

Also, when antennas are crowded together, the antennas usually have to be made smaller, which can also affect antenna efficiency.

See for example the following references:

[15] H.A. Wheeler, "Small antennas," IEEE Trans. Antennas Propagat., vol. AP-23, n.4, pp.462-469, July 1975.

[16] J.S.McLean, "A re-examination of the fundamental limits ontheradiation Q of electrically small antennas," IEEE Trans. Antennas Propagat., vol.44, n.5, pp.672-676, May 1996.

Finally, with lower frequencies and longer wavelengths, the physical size of MIMO devices becomes intractable. An extreme example is at HF bands, where MIMO device antennas must be separated from each other by 10 meters or more.

2. Noise limitation. Every MIMO receiver/transmitter subsystem generates some level of noise. When more and more of these subsystems are placed close to each other, the background noise increases. At the same time, when it is necessary to identify more different signals from a multi-antenna MIMO system, lower background noise is required.

3. Cost and power constraints. Although cost and power consumption are not the focus in some MIMO applications, in a typical wireless product, both cost and power consumption are critical constraints when developing a successful product. For each MIMO antenna, a separate RF subsystem is required, including separate analog-to-digital (A/D) and digital-to-analog (D/A) converters. Unlike many aspects of digital systems that are scaled by Moore's Law (an empirical observation made by Intel co-founder Gordon Moore that the number of transistors on an integrated circuit for a tiny device doubles approximately every 24 months twice; source: http://www.intel.com/technology/mooreslaw/), such dense analog subsystems typically have physical size and power requirements that scale linearly with cost and power. Therefore, a multi-antenna MIMO setup would be extremely expensive and consume a staggering amount of power compared to a single-antenna setup.

As a result of the above, most MIMO systems expected today are on the order of 2 to 4 antennas, resulting in a 2 to 4 times increase in throughput and some SNR (Signal to Noise Ratio) due to the diversity benefits of multi-antenna systems rise. MIMO systems with 10 antennas have been anticipated (especially at higher microwave frequencies due to shorter wavelengths and closer antenna spacing), but except for some special and cost-insensitive applications, more than 10 Antenna is very impractical.

virtual antenna array

A particular application of MIMO-type techniques is virtual antenna arrays. Such a system is proposed in a research paper presented by the European Research Collaboration in the Field of Science and Technology, EURO, Barcelona, Spain, 15-17 January 2003: Center for Telecommunication Research, King's College London, UK: "A step towards MIMO: Virtual Antenna Arrays”, Mischa Dohler & Hamid Aghvami.

As stated in the document, a virtual antenna array is a system of cooperating wireless devices (such as cellular phones) that communicate with each other on separate communication channels (provided they are in close enough proximity to each other), rather than on their primary communication channel with their The base stations communicate so as to work cooperatively (for example, if they are GSM cellular phones in the UHF band, this could be the Industrial Scientific Medical (ISM) wireless band at 5GHz). By forwarding information between several devices within mutual relay range (except within range of a base station) as if they were physically operating as one device with multiple antennas, single-antenna devices potentially implement MIMO-like Same throughput improvement.

In practice, however, such a system is extremely difficult to implement and of limited use. First, each device must now have a minimum of two distinct communication paths to achieve increased throughput, and the availability of its second relay link is often uncertain. Also, since they have at least a second communication subsystem and greater computing needs, the devices are more expensive, physically larger, and consume more power. Furthermore, the system relies on a very complex real-time collaboration of all systems, potentially through multiple communication links. Finally, due to increased simultaneous channel utilization (e.g., simultaneous telephone call transmissions using MIMO techniques), the computational burden on each device increases (usually exponentially with linear increase in channel utilization), which has a negative effect on Portable devices with severe power and size constraints are far from practical.

Contents of the invention

The present invention provides a system for compensating the frequency and phase offset of multi-user multi-antenna system MU-MAS communication, the system includes: one or more coding and modulation units for encoding and modulating the information bits for each wireless client device to generate encoded and modulated information bits; one or more mapping units for mapping the encoded and modulated information bits into complex symbols; and MU-MAS A frequency/phase offset-aware precoding unit, configured to calculate MU-MAS frequency/phase offset-aware precoding weights by using channel condition information obtained from the wireless client device through feedback, the MU-MAS frequency/phase offset The shift-aware precoding unit uses the weights to precode the complex symbols obtained from the mapping unit to pre-eliminate frequency/phase offset and/or inter-user interference.

The present invention also provides a system for compensating the in-phase quadrature (I/Q) imbalance of multi-user multi-antenna system MU-MAS communication, the system includes: one or more coding and modulation units for The information bits of each wireless client device in a wireless client device are encoded and modulated to generate encoded and modulated information bits; one or more mapping units are used to map the encoded and modulated information bits into complex numbers symbol; and a MU-MAS IQ perceptual precoding unit configured to calculate MU-MAS IQ perceptual precoding weights using channel condition information obtained from said wireless client device through feedback, said MU-MAS IQ perceptual precoding unit using The weights precode the complex symbols obtained from the mapping unit to pre-cancel interference and/or inter-user interference due to I/Q gain and phase imbalance.

The present invention also provides a system for dynamically adapting to the communication characteristics of a multi-user multi-antenna system MU-MAS communication system, the system includes: one or more coding and modulation units for encoding and modulating the information bits for each wireless client device to generate encoded and modulated information bits; one or more mapping units for mapping the encoded and modulated information bits into complex symbols; and MU-MAS A configurator unit for determining a subset of users and a MU-MAS transmission pattern based on channel characteristic data obtained by feedback from the wireless client device, and responsively controlling the code modulation unit and the mapping unit.

A system and method for compensating for frequency and phase offsets in a multiple antenna system (MAS) with multi-user (MU) transmission ("MU-MAS") is described. For example, a method according to one embodiment of the present invention includes: sending a training signal from each antenna of a base station to one or each of a plurality of wireless client devices, one or each of the client devices analyzing Each training signal to generate frequency offset compensation data, and receive the frequency offset compensation data at the base station; calculate MU-MAS precoder weights based on the frequency offset compensation data to pre-cancel the frequency offset at the transmitter ; use the MU-MAS precoder weights to precode the training signal to generate a precoded training signal for each antenna of the base station; send the precoded training signal from each antenna of the base station to the Each wireless client device in the plurality of wireless client devices, each client device analyzes each training signal to generate channel characteristic data, and receives the channel characteristic data at the base station; based on the channel characteristic data to calculate a plurality of MU-MAS precoding weights, the MU-MAS precoder weights are calculated to pre-cancel frequency and phase offsets and/or interference between users; use the MU-MAS precoder weights to precode data , to generate a precoded data signal for each antenna of the base station; and send the precoded precoded data signal to each client device through each antenna of the base station.

Description of drawings

In conjunction with the accompanying drawings, the following detailed description can obtain a better understanding of the present invention, wherein:

Figure 1 shows a prior art MIMO system.

Figure 2 shows an N-antenna base station in communication with multiple single-antenna client devices.

Figure 3 shows a three antenna base station communicating with three single antenna client devices.

Figure 4 shows the training signal technique used in one embodiment of the invention.

FIG. 5 shows channel characteristic data transmitted from a client device to a base station according to one embodiment of the present invention.

Figure 6 illustrates a multiple-input distributed-output ("MIDO") downstream transmission according to one embodiment of the present invention.

Figure 7 illustrates a multiple-input multiple-output ("MIMO") uplink transmission according to one embodiment of the present invention.

Fig. 8 shows a base station cycling through different client groups to distribute throughput according to one embodiment of the present invention.

FIG. 9 shows proximity based customer grouping according to one embodiment of the present invention.

Figure 10 shows an embodiment of the invention used in an NVIS system.

Figure 11 shows an embodiment of a DIDO transmitter with an I/Q compensation functional unit.

Figure 12 shows a DIDO receiver with an I/Q compensation functional unit.

Figure 13 shows an embodiment of a DIDO-OFDM system with I/Q compensation.

Figure 14 shows one embodiment of DIDO 2x2 performance with and without I/Q compensation.

Figure 15 shows one embodiment of DIDO 2x2 performance with and without I/Q compensation.

Figure 16 shows one embodiment of SER (Symbol Error Rate) for different QAM constellations with and without I/Q compensation.

Figure 17 shows one embodiment of DIDO2x2 performance with and without I/Q compensation at different UE locations.

Fig. 18 shows one implementation of SER in an ideal (i.i.d. (independent and identically distributed)) channel with and without I/Q compensation.

Figure 19 shows one embodiment of a transmitter architecture for an adaptive DIDO system.

Figure 20 shows one embodiment of a receiver architecture for an adaptive DIDO system.

Fig. 21 shows an implementation manner of the adaptive DIDO-OFDM method.

Figure 22 shows one embodiment of an antenna layout for DIDO measurements.

Figure 23 shows an embodiment of an array configuration for different orders of DIDO systems.

Figure 24 shows the performance of different levels of DIDO systems.

Figure 25 shows one embodiment of an antenna array for DIDO measurements.

Figure 26 shows one embodiment of DIDO 2x2 performance with 4-QAM and 1/2 FEC rate as a function of UE location.

Figure 27 shows one embodiment of an antenna layout for DIDO measurements.

Figure 28 shows how in one embodiment DIDO 8x8 produces a larger SE than DIDO 2x2 with low TX power requirements.

Figure 29 shows one embodiment of DIDO 2x2 performance with antenna selection.

Figure 30 shows the average bit error rate (BER) performance of different DIDO precoding schemes in the i.i.d. channel.

Figure 31 shows the SNR gain of ASel as a function of the number of additional transmit antennas in the i.i.d. channel.

Figure 32 shows the SNR threshold as a function of the number of users (M) for block diagonalization (BD) and ASel with 1 and 2 external antennas in the i.i.d. channel.

Figure 33 shows the BER versus per-user average SNR for two users located in the same angular direction and with different angular spread (AS) values.

Figure 34 shows similar results to Figure 33, but with a higher angular separation between users.

Figure 35 plots AS as a function of the SNR threshold for different values of the average Angle of Arrival (AOA) for users.

Figure 36 shows the SNR thresholds for an exemplary case of 5 users.

Figure 37 provides a comparison of the SNR threshold BD versus ASel with 1 and 2 additional antennas for the 2-user case.

Figure 38 shows similar results to Figure 37, but for 5 users.

Figure 39 shows the SNR thresholds for BD schemes with different AS values.

Figure 40 shows the SNR thresholds in spatially correlated channels with AS = 0.1° for BD and ASel with 1 and 2 extra antennas.

Figure 41 shows the calculation of the SNR thresholds for two other channel scenarios with AS=5°.

Figure 42 shows the calculation of the SNR thresholds for two other channel scenarios with AS=10°.

Figures 43-44 show the SNR threshold as a function of the number of users (M) and the angular spread (AS) of the BD and ASel schemes for the case of 1 and 2 additional antennas, respectively.

Figure 45 shows a receiver equipped with a frequency offset estimator/compensator;

Figure 46 shows a DIDO 2x2 system model according to one embodiment of the present invention.

Figure 47 shows a method according to one embodiment of the invention.

Figure 48 shows the SER results for the DIDO 2×2 system with and without frequency offset.

Figure 49 compares the SNR threshold performance of different DIDO schemes.

Figure 50 compares the amount of overhead required for different method implementations.

Fig. 51 shows the simulation with a small frequency offset of _fmax = 2 Hz and no integer offset correction.

Figure 52 shows the results when the integer offset estimator is turned off.

detailed description

In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the invention. It will be apparent, however, to one of ordinary skill in the art that the present invention can be practiced without some of the specific details. In addition, well-known structures and devices are shown in block diagram form in order to avoid obscuring the underlying principles of the invention.

FIG. 1 shows a prior art MIMO system with a transmit antenna 104 and a receive antenna 105 . The throughput of such a system can achieve 3 times the throughput that is usually achieved in the available channels. There are a number of different ways to implement the details of such a MIMO system, which are described in published literature on the subject, the following explanation will describe one such way.

Before the data is transmitted in the MIMO system of Figure 1, the channel is "characterized". This is achieved by initially transmitting a "training signal" from each transmit antenna 104 to each receiver 105 . The training signal is generated by the encoding and modulation subsystem 102 and converted to an analog signal by a D/A converter (not shown) and then converted from baseband to RF by each transmitter 103 . Each receive antenna 105 coupled to its RF receiver 106 receives and converts each training signal to a baseband signal. The baseband signal is converted to a digital signal by a D/A converter (not shown), and then the signal processing subsystem 107 characterizes the training signal. The characteristics of each signal may include many factors including, for example, phase and amplitude relative to a reference signal internal to the receiver, an absolute reference signal, a relative reference signal, characteristic noise, or other factors. The characteristics of each signal are usually defined as a vector representing the phase and amplitude changes of several aspects of the signal as it travels through the channel. For example, in a quadrature amplitude modulated ("QAM") modulated signal, the feature may be a vector of phase and amplitude offsets of several multipath images of the signal. As another example, in an orthogonal frequency-division multiplexing ("OFDM") modulated signal, it may be a vector of phase and amplitude offsets for several or all of the individual sub-signals in the OFDM spectrum.

The signal processing subsystem 107 stores the channel characteristics received by each receive antenna 105 and corresponding receiver 106 . After all three transmit antennas 104 have completed their training signal transmissions, the signal processing subsystem 107 will have stored three channel signatures for each of the three receive antennas 105, which form a 3×3 matrix 108, which Denoted as the channel characteristic matrix "H". Each individual matrix element Hi,j is the channel characteristic of the training signal transmission from transmit antenna 104i received by receive antenna 105j.

At this point, the signal processing subsystem 107 inverts the matrix H 108 to produce H ⁻¹ , and awaits transmission of the actual data from the transmit antenna 104 . Note that a variety of existing MIMO techniques described in the available literature can be used to ensure that the H-matrix 108 is invertible.

In an implementation, the content (payload) of the data to be transmitted is sent to the data input subsystem 100 . It is then split into three parts by a splitter 101 before being sent to the encoding and modulation subsystem 102 . For example, if the content is ASCII bits of "abcdef", it can be split into three sub-contents "ad", "be" and "cf" by the allocator. Each sub-content is then sent to the encoding and modulation subsystem 102 individually.

Each sub-content is encoded individually by using an encoding system suitable for the statistical independence and error correction capabilities of each signal. These include, but are not limited to, Reed-Solomon coding, Viterbi coding, and Turbo Codes. Finally, each of the three encoded sub-contents is modulated using an appropriate modulation method for the channel. Exemplary modulation methods are Differential Phase Shift Keying ("DPSK"), 64-QAM modulation, and OFDM. It should be noted here that the diversity gain provided by MIMO allows higher orders of modulation constellations that are also feasible in SISO (Single Input Single Output) systems using the same channel. Each coded and modulated signal is then transmitted out through its own antenna 104 following D/A conversion by a D/A conversion unit (not shown) and RF generation by each transmitter 103 .

Assuming that sufficient spatial diversity exists between the transmit and receive antennas, each receive antenna 105 will receive a different combination of the three transmitted signals from antenna 104 . Each RF receiver 106 receives each signal and converts them to baseband signals before an A/D converter (not shown) digitizes the signals. If y _n is the signal received by the nth receive antenna 105, x _n is the signal transmitted by the nth transmit antenna 104, and N is noise, then this can be described by the following equation.

y ₁ =x ₁ H ₁₁ +x ₂ H ₁₂ +x ₃ H ₁₃ +N

y ₂ =x ₁ H ₁₂ +x ₂ H ₂₂ +x ₃ H ₂₃ +N

y ₃ =x ₁ H ₁₃ +x ₂ H ₃₂ +x ₃ H ₃₃ +N

Assuming this is a system of three equations with three unknowns, then it is a matter of the signal processing subsystem 107 deriving the linear algebra for x ₁ , x ₂ and x ₃ (assuming N is at a sufficiently low level, to allow decoding of the signal):

x ₁ =y ₁ H ^-1 ₁₁ +y ₂ H ^-1 ₁₂ +y ₃ H ^-1 ₁₃

x ₂ =y ₁ H ^-1 ₂₁ +y ₂ H ^-1 ₂₂ +y ₃ H ^-1 ₂₃

x ₃ =y ₁ H ^-1 ₃₁ +y ₂ H ^-1 ₃₂ +y ₃ H ^-1 ₃₃

Once the three transmitted signals x _n are derived, they are demodulated, decoded and error corrected by the signal processing subsystem 107 to recover the three bit streams originally separated by the splitter 101 . These bit streams are combined in a combiner unit 108 and output from a data output 109 as a single data stream. Assuming system robustness against noise impairments, the bitstream produced by data output 109 will be the same as the bitstream introduced into data input 100 .

Although the described prior art systems are generally effective up to four antennas, perhaps up to 10 antennas, for reasons described in the background section of this disclosure, it becomes difficult to Very unrealistic.

Typically, such prior art systems are bi-directional, with the return path implemented in exactly the same way, but reversed, with a transmit and receive subsystem on each side of the communication channel.

Figure 2 shows an embodiment of the invention, in which a Base Station (BS) 200 is configured with a Wide Area Network (WAN) interface (eg via T1 or other high speed connection) 201 and provided with a certain number (N) of antennas 202 . For the time being, we will use the term "base station" to refer to any wireless station that communicates wirelessly with a set of clients at fixed locations. Examples of base stations may be access points in a Wireless Local Area Network (WLAN), or WAN antennas or antenna arrays. There are a number of client devices 203-207, each having a single antenna, which are served wirelessly by the base station 200. Although for the purposes of this example it is quite easy to think of a base station located in an office environment in which it serves user devices 203-207 equipped with wireless networked personal computers, this configuration will find use in a wide variety of applications Situations, indoors and outdoors, where base stations serve wireless clients. For example, the base station may be located on a cellular telephone tower, or on a television broadcast tower. In one embodiment, the base station 200 is placed on the ground for uplink transmission at HF frequencies (such as a frequency of 24 MHz) to reflect signals from the ionosphere, as proposed on April 20, 2004, serial number No. .10/817,731, titled SYSTEM AND METHOD FORENHANCING NEAR VERTICALINCIDENTCE SKYWAVE ("NVIS") COMMUNICATION USINGSPACE-TIME CODING, as described in the co-pending application, assigned to the attorney of the present application, is hereby incorporated by reference.

Certain details associated with base station 200 and client devices illustrated are for illustrative purposes only and are not necessary in accordance with the underlying principles of the invention. For example, the base station may be connected via the WAN interface 201 to a number of different types of wide area networks, including dedicated wide area networks such as those used for digital video distribution. Similarly, a client device may be any kind of wireless data processing and/or communication device including, but not limited to, cellular telephones, personal digital assistants ("PDAs"), receivers, and wireless cameras.

In one embodiment, the n antennas 202 of the base station are spatially separated so that each transmits and receives non-spatially correlated signals as if the base station were a prior art MIMO transceiver. As described in the background, experiments with antennas placed at λ6 (i.e. 1/6 wavelength) intervals have been done, which successfully achieved a throughput boost from MIMO, but in general, the further apart these base station antennas are Placement, the better the performance of the system, λ2 is the minimum satisfactory distance. Of course, the underlying principles of the invention are not limited to any particular separation between the antennas.

Note that a single base station 200 can well place its antennas at great distances. For example, in the HF spectrum, the antennas can be 10 meters or more away (eg, in the NVIS implementation mentioned above). If 100 such antennas are used, the antenna array of the base station can occupy an area of several square kilometers.

In addition to the space diversity technique, in order to improve the effective throughput of the system, an embodiment of the present invention polarizes the signal. Improving channel capacity through polarization is a well-known technique that has been used by satellite television providers for many years. Using polarization techniques, multiple (eg three) base station or user antennas can be brought very close to each other and still be non-spatially correlated. While conventional RF systems typically only benefit from two-dimensional (eg, x and y) diversity of polarization, the structures described here can further benefit from three-dimensional (x, y, and z) diversity of polarization.

In addition to spatial and polarization diversity, one embodiment of the invention employs near-orthogonal radiation patterns to improve link performance via pattern diversity. Pattern diversity can improve the capacity and bit error rate performance of MIMO systems, and its advantages over other antenna diversity techniques can be found in the following articles:

[17] L. Dong, H. Ling, and R. W. Heath Jr., "Multiple-input multiple-output

wireless communication systems using antenna pattern diversity,"Proc.IEEE Glob.Telecom.Conf.,vol.1,pp.997-1001,Nov.2002.

[18] R. Vaughan, "Switched parasitic elements for antenna diversity," IEEE

Trans.Antennas Propagat., vol.47, pp.399-405, Feb.1999.

[19] P. Mattheijssen, M.H.A.J. Herben, G. Dolmans, and L. Leyten, "Antenna-pattern diversity versus space diversity of of use at handhelds," IEEE Trans. on Veh. Technol., vol.53, pp.1035-1042, July 2004.

[20] C.B.Dietrich Jr, K.Dietze, J.R.Nealy, and W.L.Stutzman, "Spatial, polarization, and pattern diversity for wireless handheld terminals," Proc.IEEEAntennas and Prop.Symp., vol.49, pp.1271-1281 , Sep.2001.

[21] A.Forenza and R.W.Heath, Jr, "Benefit of Pattern Diversity Via2-element Array of Circular Patch Antennas in Indoor Clustered MIMOChannels", IEEE Trans.on Communications, vol.54, no.5, pp.943-954 , May 2006.

By using pattern diversity, multiple base station or user antennas can be brought into close proximity to each other and nonetheless not be spatially correlated.

Figure 3 provides additional details of one embodiment of the base station 200 and client devices 203-207 shown in Figure 2 . For purposes of simplicity, the base station 300 is shown with only three antennas 305 and three client devices 306-308. It should be noted, however, that embodiments of the invention described herein may be implemented with an almost unlimited number of antennas 305 (ie, limited only by available space and noise) and client devices 306-308.

Figure 3 is similar to the prior art MIMO structure shown in Figure 1, in that both have three antennas at each end of the communication channel. The notable difference is that in prior art MIMO systems, the three antennas 105 on the right side of FIG. The signal processing subsystem 107 is processed. In contrast, in FIG. 3, the three antennas 309 on the right side of the figure are each coupled to a different client device 306-308, each of which may be distributed anywhere within range of the base station 305 . In this way, each client device's received signal can be processed in its coding, modulation, signal processing subsystem 311 independently of the other two received signals. Thus, Figure 3 shows a multiple-input (ie, antenna 305) distributed output (ie, antenna 305) system, hereinafter referred to as "MIMO "system.

Note that this application uses a different terminology usage than previous applications to better align with academic and industry practice. Copending Application No. 10/817,731, entitled "SYSTEMANDMETHOD FOR ENHANCING NEARVERTICAL INCIDENCE SKYWAVE ("NVIS") COMMUNICATION USING SPACE-TIME CODING," filed April 20, 2004, previously cited, and July 2004 In application No. 10/902,978 filed on the 30th (this application is a continuation of that application), "input" and "output" (in the context of SIMO, MISO, DIMO, and MIDO) have the same meaning as that term in this application The meaning in is the opposite. In the previous application, "input" refers to a wireless signal input to a receiving antenna (eg, antenna 309 in FIG. 3 ), and "output" refers to a wireless signal output from a transmitting antenna (eg, antenna 305 ). In academia and the wireless industry, the opposites of "input" and "output" are commonly used, where "input" refers to the wireless signal input to the channel (i.e., the wireless signal sent from antenna 305), and "output" refers to the wireless signal from The wireless signal output by the channel (that is, the wireless signal received by the antenna 309). This application adopts a usage of this term that is contrary to the usage in the applications cited earlier in this paragraph. Accordingly, the following depicts equivalents of term usage between several applications:

10/817,731 and 10/902,978 this application

SIMO = MISO

MISO = SIMO

DIMO = MIDO

MIDO = DIMO

The MIDO structure shown in Figure 3 achieves a capacity boost similar to that achieved by MIMO on SISO systems for a given number of transmit antennas. However, one difference between MIMO and the particular MIDO embodiment shown in FIG. 3 is that each MIDO client device 306-308 requires only a single receive antenna to realize the capacity boost provided by multiple base station antennas, whereas with MIMO, each client The device requires at least as many receive antennas as the desired capacity multiple. Given that there is usually an enforced limit on how many antennas can be placed at a client device (as explained in the background), this typically limits MIMO systems to between 4 and 10 antennas (4x to 10x capacity). Since the base station 300 typically serves many client devices from fixed and powered locations, it is practical to expand it to well beyond 10 antennas, and to separate the antennas by an appropriate distance to achieve space diversity. As mentioned, each antenna is equipped with a transceiver 304 and a portion of the processing power of the coding, modulation and signal processing section 303 . It is worth noting that in this embodiment, no matter how much the base station 300 is extended, each client device 306-308 will only require one antenna 309, so the cost for single-user client devices 306-308 will be very low, and the base station 300 Costs can be shared among a large base of users.

In FIGS. 4-6, examples of how MIDO transmissions from base station 300 to client devices 306-308 are accomplished are shown.

In one embodiment of the invention, the channel is characterized before MIDO transmission begins. For a MIMO system, each antenna 405 transmits training signals one by one. FIG. 4 only shows the transmission of the first training signal, but for three antennas 405 there are three separate transmissions. Each training signal is generated by the coding, modulation and signal processing subsystem 403 , converted into an analog signal through a D/A converter, and sent out through each RF transceiver 404 as an RF signal. A variety of different coding, modulation, and signal processing techniques may be used including, but not limited to, those described above (eg, Reed Solomon, Viterbi Coding; QAM, DPSK, QPSK modulation, etc.).

Each client device 406-408 receives the training signal through its antenna 409 and through the transceiver 410 converts the training signal into a baseband signal. An A/D converter (not shown) converts the signal to a digital signal where it is processed by encoding, modulation and signal processing subsystem 411 . The signal characteristics logic unit 320 then identifies characteristics of the resulting signal (eg, identifying the phase and amplitude distortions described above) and stores the characteristics in memory. This feature processing process is similar to the processing process of the MIMO system in the prior art, and a significant difference is that each client device only calculates the feature vectors of one antenna instead of n antennas. For example, the training signal of a known pattern initializes the coding, modulation and signal processing subsystem 420 of the client device 406 (either by receiving it in a transmitted message when generated, or by other initialization processes). When the antenna 405 transmits the training signal in a known pattern, the coding, modulation and signal processing subsystem 420 uses correlation to find the strongest training signal reception pattern, which saves the phase and amplitude offsets, which it then The mode is subtracted from the middle of the received signal. Next, it finds the second strongest received pattern associated with the training signal, saves the phase and amplitude offsets, then it subtracts the second strongest pattern from the received signal. This process continues until some fixed number of phase and amplitude shifts (eg, 8) are preserved or the detectable training signal pattern falls below a given background noise. This vector of phase/amplitude shifts becomes element H ₁₁ of vector 413 . At the same time, the encoding, modulation and signal processing subsystems of client devices 407 and 408 perform the same process, producing their vector elements _H21 and _H31 .

The memory for storing channel characteristics may be a non-volatile memory, such as a flash memory, or a hard disk, and/or a volatile memory, such as a random access memory (eg, SDRAM, RDAM). In addition, different user devices may simultaneously use different types of memory to store characteristic information (for example, a PDA may use a flash memory, while a notebook computer may use a hard disk). The principles underlying the present invention are not limited to any particular type of storage mechanism at the various client devices or base stations.

As noted above, since each client device 406-408 has only one antenna, each stores only 1x3 rows 413-415 of the H matrix, depending on the scheme used. FIG. 4 shows the stage after the transmission of the first training signal. Here, the first column of 1×3 rows 413-415 stores the channel characteristic information of the first antenna of the three base station antennas 405. The remaining two columns store the channel characteristics of the next two training signal transmissions from the remaining two base station antennas. Note that the three training signal patterns are transmitted at separate times for illustrative purposes. If the three training signal patterns are chosen so as to be independent of each other, they can be transmitted simultaneously, thus reducing the training time.

As shown in FIG. 5, after all three pilot transmissions are complete, each client device 506-508 sends back to the base station 500 the 1x3 rows 513-515 of matrix H that it has stored. For purposes of simplicity, only one client device 506 is shown in FIG. 5 transmitting its characteristic information. A suitable modulation method (eg DPSK, 64QAM, OFDM) can be used in conjunction with appropriate error correction coding (eg Reed Solomon, Viterbi Coding and/or TurboCodes) to ensure that the base station 500 receives the line 513 accurately Data in -515.

In Figure 5, although all three antennas 505 are shown receiving signals, the single antenna and single transceiver of the base station 500 is sufficient for receiving transmissions per 1x3 row 513-515. Under certain conditions, however, many or all of antennas 505 and transceivers 504 are used to receive each transmission (i.e., using prior art single-input multiple-output (“SIMO”) in coding, modulation, and signal processing subsystem 503 processing technology) can achieve better signal-to-noise ratio (SNR) than single antenna 505 and transceiver 504.

When the coding, modulation and signal processing subsystem 503 of the base station 500 receives the 1x3 rows 513-515 from each client device 507-508, it stores the 1x3 rows 513-515 in a 3x3 The H matrix 516. For client devices, the base station may store matrix 516 using a number of different storage technologies including, but not limited to, non-volatile mass memory (eg, hard disk) and/or volatile memory (eg, SDRAM). FIG. 5 shows the stage at which the base station has received and stored the 1×3 row 513 from the client device 509 . As the 1x3 rows 514 and 515 are transmitted from the remaining client devices, they may be transmitted and stored in the H-matrix 516 until the entire H-matrix 516 is stored.

Referring to FIG. 6, an embodiment of MIDO transmission from base station 600 to client devices 606-608 will now be described. Because each client device 606-608 is an independent device, each device receives a different data transmission. Thus, an embodiment of the base station 600 includes a router 602 positioned in communication between the WAN interface 601 and the coding, modulation and signal processing subsystem 603, which receives a plurality of data streams (in the form of a bit stream) from the WAN interface 601, The data streams are sent as separate data streams u1-u3 corresponding to each client device 606-608, respectively. To this end, the router 602 may use various known routing techniques.

As shown in Figure 6, the three bit streams, u ₁ -u ₃ , are routed into the coding, modulation and signal processing subsystem 603, which encodes them into statistically independent error-corrected streams (e.g., using ReedSolomon , Viterbi or enhanced coding) and modulate them with the appropriate modulation method for the channel (such as DPSK, 64QAM or OFDM). Furthermore, the embodiment shown in FIG. 6 includes a signal precoding logic unit 630 for uniquely encoding the signal transmitted from each antenna 605 based on the signal characteristic matrix 616 . Specifically, in this embodiment, the precoding logic unit 630 multiplies the three bit streams u ₁ -u ₃ in FIG. 6 by the inverse matrix of the H matrix 616 to generate three new bit streams u' ₁ -u ' ₃ instead of routing each coded and modulated bit stream to a separate antenna (as done in Figure 1). Then, a D/A converter (not shown) converts the three precoded bit streams into analog signals, which are sent out as RF signals by the transceiver 604 and the antenna 605 .

Before explaining how client devices 606-608 receive the particular stream, the operations performed by precoding module 630 will be described. Similar to the MIMO example in Figure 1 above, the encoded and modulated signal for each of the three original bitstreams will be denoted as u _n . In the embodiment shown in FIG. 6, each _ui contains data from three bitstreams routed by router 602, each such bitstream to be one of three user devices 606-608.

However, unlike the MIMO example in FIG. 1, where each _xi has each antenna 104 transmitting, in the embodiment of the invention shown in FIG. 6, each u _i is received at each client device antenna 609 (plus any noise in the channel N). To achieve this result, the output of each of the three antennas 605 (which we denote as v _i ) is a function of u _i and the H matrix characterizing each client device. In an embodiment, the precoding logic unit 630 in the encoding, modulation and signal processing subsystem calculates each v _i by implementing the following equation:

v ₁ ＝u ₁ H ^-1 ₁₁ +u ₂ H ^-1 ₁₂ +u ₃ H ^-1 ₁₃

v ₂ ＝u ₁ H ^-1 ₂₁ +u ₂ H ^-1 ₂₂ +u ₃ H ^-1 ₂₃

v ₃ =u ₁ H ^-1 ₃₁ +u ₂ H ^-1 ₃₂ +u ₃ H ^-1 ₃₃

Thus, unlike MIMO, where each _xi is computed at the receiver after the channel transforms the signal, the embodiments of the invention described here solve for each _vi at the transmitter before the channel transforms the signal. Each antenna 609 receives u _i which has been separated from the other un _-1 bit streams for the other antennas 609 . Each transceiver 610 converts each received signal to a baseband signal, where an A/D converter (not shown) digitizes it, and each encoding, modulation, and signal processing subsystem 611 decodes its x _i bit stream The reconciliation decodes and sends its bit stream to the data interface 612 for use by the client device (eg, an application on the client device).

Embodiments of the invention described herein can be implemented using a variety of different encoding and modulation methods. For example, in OFDM implementations where the frequency spectrum is divided into multiple sub-bands, the techniques described here can be used to characterize each individual sub-band. However, as mentioned above, the underlying principles of the invention are not limited to any particular modulation method.

If client devices 606-608 are portable data processing devices, such as PDAs, notebook computers, and/or wireless telephones, channel characteristics may change frequently as client devices may move from one location to another. In this way, in one embodiment of the present invention, the channel characteristic matrix 616 of the base station is constantly updated. In one embodiment, the base station 600 periodically (every 250 milliseconds) sends a new training signal to each client device, and each client device continuously sends its channel eigenvector back to the base station 600 to ensure that the channel characteristics remain accurate (e.g., If the environment changes or the client device moves thereby affecting the channel). In one embodiment, the training signal is interleaved in the actual data signal sent to each client device. Typically, the throughput of the training signal is much lower than the throughput of the data signal, so this will have little effect on the overall system throughput. Correspondingly, in this embodiment, the channel feature matrix 616 can be continuously updated when the base station actively communicates with each client device, so that when the client device moves from one location to the next, or the environment changes, the channel characteristics will be affected. maintain accurate channel characteristics at all times.

One embodiment of the invention shown in FIG. 7 uses MIMO technology to improve the uplink communication channel (ie, the channel from client devices 706-708 to base station 700). In this embodiment, the uplink channel characterization logic unit 741 in the base station continuously analyzes and characterizes the channel from each client device. Specifically, each client device 706-708 sends a training signal to the base station 700, where the channel characteristic logic unit 741 analyzes to generate an N x M channel characteristic matrix 741, where N is the number of client devices and M is the antenna used by the base station quantity. The embodiment shown in FIG. 7 uses three antennas 705 and three client devices 706-708 at the base station, which results in a 3×3 channel eigenmatrix 741 stored at the base station 700 . The client device may use the MIMO uplink transmission shown in FIG. 7 for sending data back to the base station 700 and transmitting the channel eigenvector back to the base station 700, as shown in FIG. 5 . However, unlike the embodiment shown in FIG. 5, in FIG. 5, the channel eigenvectors of each client device are transmitted at separate times, while the method shown in FIG. The vectors are transmitted back to the base station 700, thereby greatly reducing the impact of the channel eigenvectors on the throughput of the backhaul channel.

As noted above, the characteristics of each signal may include many factors including, for example, its phase and amplitude relative to a reference signal internal to the receiver, an absolute reference signal, a relative reference signal, characteristic noise, or other factors. For example, in a signal modulated by quadrature amplitude modulation, the feature may be the phase and amplitude offset vectors of several multipath images of the signal. As another example, in a signal modulated by Orthogonal Frequency Division Multiplexing, the feature may be the phase and amplitude offset vectors of several or all of the individual component signals in the OFDM spectrum. The training signal may be generated by the encoding and modulation subsystem 711 of each client device, the D/A converter (not shown) converts the training signal into an analog signal, and then the transmitter 709 of each client device converts it from the baseband signal converted into an RF signal. In one embodiment, to ensure synchronization of training signals, client devices transmit training signals only when requested by the base station (eg, in the case of round robin). Furthermore, the training signal may be interleaved within the actual data signal sent from each client device, or the training signal may be transmitted together with the actual data signal. Therefore, even if the client devices 706-708 are mobile, the upstream channel characteristic logic unit 741 can continuously transmit and analyze the training signal, thereby ensuring that the channel characteristic matrix 741 is kept updated.

The total channel capacity supported by the foregoing embodiments of the present invention can be defined as min(N, M), where M is the number of client devices and N is the number of base station antennas. That is, the capacity is limited by the number of antennas on the base station side or the client side. As such, one embodiment of the invention uses synchronization techniques to ensure that no more than min(N,M) antennas are transmitting/receiving at a given time.

In a typical case, the base station 700 will have fewer antennas 705 than the number of client devices 706-708. FIG. 8 shows an exemplary scenario that allows five client devices 804-808 to communicate with a base station having three antennas 802. In this embodiment, after determining the total number of client devices 804-808 and detecting the necessary channel characteristic information (eg, as described above), the base station 800 selects a first group of three clients 810 to communicate with (because min (N,M) = 3, so three customers in this example). After communicating with the first group of clients 810 for a specified time, the base station selects another group of three clients 811 with which to communicate. In order to evenly distribute the communication channels, the base station 800 selects two client devices 807, 808 not included in the first group. Additionally, base station 800 selects additional client devices 806 to be included in the first group as additional antennas are available. In one embodiment, the base station 800 cycles through the customer population in such a manner that each customer can be effectively allocated the same amount of throughput over time. For example, to distribute throughput evenly, the base station may then select any combination of three client devices other than client device 806 (ie, since client device 806 is used to communicate with the base station in the first two cycles).

In one embodiment, in addition to standard data communications, the base station may use the aforementioned techniques to transmit training signals to and receive training signals and signal characterization data from client devices.

In one embodiment, certain client devices or groups of client devices may be assigned different levels of throughput, e.g., client devices may be prioritized so that relatively higher priority client devices may be assigned lower priority client devices. Clients have more communication cycles (ie, more throughput). A customer's "priority" can be selected based on a number of variables, including, for example, the user's subscription fee for wireless bandwidth (e.g., users' willingness to pay more for extra throughput), and/or The type of data communicated to/from the client device (eg, real-time communications, such as telephone voice and video, receive priority over non-real-time communications, such as email).

In embodiments where the base station dynamically allocates throughput based on the current load required by each client device. For example, if a client device 804 is streaming live video while the other devices 805-808 are performing non-real-time functions such as email, the base station 800 may allocate relatively more throughput to the client 804. It should be noted, however, that the underlying principles of the invention are not limited to any particular throughput allocation technique.

As shown in Figure 9, two client devices 907, 908 may be in close proximity such that the channel characteristics of the clients are virtually the same. As a result, the base station will receive and store virtually equal channel eigenvectors for the two client devices 907, 908, so this will not produce a unique, spatially distributed signal for each client. Accordingly, in one embodiment, the base station will ensure that any two or more client devices that are in close proximity to each other are assigned to different groups. For example, in FIG. 9, the base station 900 first communicates with a first group 910 of client devices 904, 905, and 908, and then communicates with a second group 911 of client devices 905, 906, 907, which ensures that the client devices 907 and 908 are in different groups.

Alternatively, in one embodiment, base station 900 communicates with client devices 907 and 908 simultaneously, but multiplexes the communication channels using known channel multiplexing techniques. For example, the base station may use time division multiplexing ("TDM"), frequency division multiplexing ("FDM"), or code division multiple access ("CDMA") techniques to separate individual, spatially correlated signals between client devices 907 and 908 .

Although each client device described above is equipped with a single antenna, the underlying principles of the present invention can be implemented by using client devices with multiple antennas to increase throughput. For example, when used on the wireless system described above, a client with 2 antennas will achieve a 2X increase in throughput, a client with 3 antennas will achieve a 3X increase in throughput, etc. (i.e., assuming space and angular separation between them are sufficient). The base station can apply the same general rules when cycling through client devices with multiple antennas. For example, it may treat each antenna as a separate client and allocate throughput to that "client" as it would any other client (eg, ensuring that each client provides enough or comparable communication cycles).

As noted above, one embodiment of the present invention improves signal-to-noise ratio and throughput in near-normal-incidence sky-wave ("NVIS") using the MIDO and/or MIMO signaling techniques described above. Referring to FIG. 10 , in one embodiment of the present invention, a first NVIS base station 1001 equipped with a matrix of N antennas 1002 is used to communicate with M client devices 1004 . The NVIS antenna 1002 and various user equipment antennas 1004 transmit signals uplink at an angle within approximately 15 degrees from vertical to achieve desired NVIS and minimize ground wave interference effects. In one embodiment, the antenna 1002 and client device 1004 support at specified frequencies in the NVIS spectrum (e.g., at the carrier frequency or below 23 MHz, but typically below 10 MHz) using the various MIDO and MIMO techniques described above. Multiple independent data streams 1006, thereby significantly increasing throughput at a given frequency (ie, in direct proportion to the number of statistically independent data streams).

The NVIS antennas serving a given base station may be at a great physical distance from each other. Given the long wavelengths below 10 MHz and the long distances over which signals travel (300 miles round trip), physical separation of antennas by hundreds of yards, or even miles, can provide diversity benefits. Under such conditions, individual antenna signals can be retrieved to a central location for processing using conventional wired or wireless communication systems. Alternatively, each antenna may have local equipment to process its signal and then use conventional wired or wireless communication systems to transmit this data back to the central location. In one embodiment of the invention, NVIS base station 1001 has a broadband link 1015 to the Internet 1010 (or other wide area network), thereby providing client devices 1003 with remote, high-speed, wireless network access.

In one embodiment, the base station and/or the user may utilize polarization/pattern diversity technology to reduce array size and/or user distance while providing diversity and improving throughput. For example, in a DIMO system with HF transmission, due to polarization/pattern diversity, users can be co-located and their signals will not be correlated. In particular, by using pattern diversity, one user can communicate with the base station via ground wave, while other users can communicate with the base station via NVIS.

Additional Embodiments of the Invention

Ⅰ. Using I/Q imbalance for DIDO-OFDM precoding

One embodiment of the present invention employs a system and method for compensating for in-phase quadrature (I/Q) imbalance in a distributed-input distributed-output (DIDO) system with orthogonal frequency division multiplexing (OFDM) . In short, according to this embodiment, the user equipment estimates the channel and feeds the information back to the base station; the base station calculates the precoding matrix to eliminate the interference between carriers and users caused by I/Q imbalance ; and parallel data streams are sent via DIDO precoding to multiple user equipments; the user equipments demodulate the data via a zero forcing (ZF), minimum mean square error (MMSE) or maximum likelihood (ML) receiver, to suppress residual interference.

As detailed below, some salient features of this embodiment of the invention include, but are not limited to:

Precoding is used to eliminate inter-carrier interference (ICI) from mirror tone (mirror tone) in OFDM systems (caused by I/Q mismatch);

Precoding for eliminating inter-user interference and ICI (caused by I/Q mismatch) in DIDO-OFDM systems;

Techniques for canceling ICI (caused by I/Q mismatch) via ZF receivers in DIDO-OFDM systems using block diagonalization (BD);

Techniques for canceling inter-user interference and ICI (due to I/Q mismatch) via precoding (at transmitter) and ZF or MMSE filter (at receiver) in DIDO-OFDM systems;

Used to cancel inter-user interference and ICI (due to I/ Q mismatch caused by) technology;

Use of precoding based on channel condition information for canceling inter-carrier interference (ICI) from image modulation in OFDM systems (caused by I/Q mismatch);

Use precoding based on channel condition information for canceling inter-carrier interference (ICI) from image modulation in DIDO-OFDM systems (caused by I/Q mismatch);

Use of I/Q mismatch aware DIDO precoder at base station and I/Q aware DIDO receiver at user terminal;

Using an I/Q mismatch aware DIDO precoder at the base station, using an I/Q aware DIDO receiver at the user terminal, and using an I/Q aware channel estimator;

Using an I/Q mismatch known DIDO precoder at the base station, an I/Q known DIDO receiver at the user terminal, and an I/Q known channel estimator and I/Q known DIDO feedback generator (the generator sends channel condition information from the user terminal to the station);

Using an I/Q mismatch-aware DIDO precoder at the base station, and an I/Q-aware DIDO configurator that uses I/Q channel information to perform various functions including user selection, adaptive coding, and modulation, space-time-frequency mapping or precoder selection);

Using an I/Q-aware DIDO receiver that cancels ICI (caused by I/Q mismatch) via a ZF receiver in a DIDO-OFDM system with a Block Diagonalization (BD) precoder);

Using an I/Q-aware DIDO receiver via precoding (at the transmitter) in a DIDO-OFDM system and a non-linear detector (at the receiver) similar to a maximum likelihood (ML) detector to eliminate inter-user interference and ICI (caused by I/Q mismatch)); and

Use an I/Q-aware DIDO receiver that removes ICI (caused by I/Q mismatch) via a ZF or MMSE filter in a DIDO-OFDM system.

a. Background

The transmit and receive signals of a typical wireless communication system contain in-phase quadrature (I/Q) components. In a real system, the in-phase and quadrature components may be distorted by imperfections in mixing and baseband operation. These distortions manifest as I/Q phase, gain, and delay mismatches. Phase imbalance is caused by the sine and cosine in the modulator/demodulator not being in proper quadrature. Gain imbalance is caused by different gain between in-phase and quadrature components. Due to the difference in delay between the I and Q rails in an analog circuit, there may also be additional distortion known as delay imbalance.

In an Orthogonal Frequency Division Multiplexing (OFDM) system, I/Q imbalance causes inter-carrier imbalance (ICI) from the transmitted tone. This effect has been studied in several sources, and in the following sources, methods for compensating for I/Q mismatch in single-input single-output SISO-OFDM systems have been proposed: M.D.Benedetto and P.Mandarini, "Analysis of the effect of the I/Q baseband filter mismatch in an OFDM modem," Wireless personal communications, pp.175-186, 2000; S.Schuchert and R. Hasholzner, "Anovel I/Q imbalance compensation scheme for the reception of OFDM signals ,” IEEE Transaction on Consumer Electronics, Aug.2001; M.Valkama, M.Renfors and V.Koivunen, “Advanced methods for I/Q imbalance compensation incommunication receivers,” IEEE Trans.Sig.Proc, Oct.2001; R.Rao and B. Daneshrad, "Analysis of I/Qmismatch and a cancellation scheme for OFDM systems," IST Mobile Communication Summit, June 2004; A. Tarighat, R. Bagheri, and A.H. Sayed, "Compensation schemes and performance analysis of IQ imbalances in OFDM receivers," Signal Processing , IEEE Transactions on [see also Acoustics, Speech, and Signal Processing, IEEE Transactions on], vol.53, pp.3257-3268, Aug.2005.

An extension of this work to multiple-input multiple-output MIMO-OFDM systems is shown in: R. Rao and B. Daneshrad, "I/Q mismatch cancellation for MIMO OFDM systems," in Personal, Indoor and Mobile Radio Communications, 2004 ; PIMRC 2004.15th IEEE International Symposium on, vol.4, 2004, pp.2710-2714. For spatial multiplexing (SM), see R.M.Rao, W.Zhu, S.Lang, C.Oberli, D.Browne, J.Bhatia, J.F.Frigon, J.Wang, P; Gupta, H.Lee, D.N.Liu , S.G. Wong, M. Fitz, B. Daneshrad, and O. Takeshita, “Multitantennatestbeds for research and education in wireless communications,” IEEE Communications Magazine, vol.42, no.12, pp.72-81, Dec.2004; S. Lang, M.R.Rao and B.Daneshrad, "Design and development of a 5.25GHz software defned wireless OFDM communication platform," IEEE Communications Magazine, vol.42, no.6, pp.6-12, June2004; for orthogonal space-time block codes ( OSTBC), see A.Tarighat and A.H.Sayed, "MIMOOFDM receivers for systems with IQ imbalances," IEEE Trans.Sig.Proc, vol.53, pp.3583-3596, Sep.2005.

Unfortunately, there is currently no literature describing how to correct for I/Q gain and phase imbalance errors in Distributed-Input Distributed-Output (DIDO) communication systems. Embodiments of the present invention described below provide a solution to these problems.

The DIDO system consists of a base station with distributed antennas, which transmits parallel data streams (precoded) to multiple users while utilizing the same radio resources (i.e., the same slot duration and frequency band) as conventional SIO systems, to enhance downlink throughput. Application No. 10/902,978, filed July 30, 2004 by S.G. Perlman and T. Cotter, entitled "System and Method for Distributed Input-Distributed Output Wireless Communications" ("the earlier application") gives a detailed description of the DIDO system, This application is assigned to the assignee of the present application and is hereby incorporated by reference.

There are various ways to implement a DIDO precoder. One scheme is block diagonalization (BD) as described in: Q.H.Spencer, A.L.Swindlehurst and M.Haardt, "Zero forcing methods for downlinkspatial multiplexing in multipleuser MIMO channels," IEEETrans.Sig.Proc, vol.52, pp.461-471, Feb.2004; K.K.Wong, R.D.Murch, and K.B.Letaief, "A joint channel diagonalization for multiuser MIMO antennasystems," IEEE Trans.Wireless Comm., vol.2, pp.773-786, JuI 2003; L.U.Choi and R.D.Murch, “A transmit preprocessing technique for multiuser MIMOsystems using a decomposition approach,” IEEE Trans. Wireless Comm., vol.3, pp.20-24, Jan 2004; Z. Shen, J.G. Andrews, R.W. Heath and B.L. Evans , "Lowcomplexity user selection algorithms for multiuser MIMO system with blockdiagonalization," accepted for publication in IEEE Trans.Sig.Proc, Sep.2005; Z.Shen, R.Chen, J.G.Andrews, R.W.Heath and B.L.Evans, "Sum capacity of multiuserMIMObroadcast channels with block diagonalization," submitted to IEEE Trans.WirelessComm., Oct. 2005; R. Chen, R.W. Heath, and J.G. Andrews, "Transmitselection diversity for unitary precoded multipleuser spatial multiplexing systems with linear receivers," accepted to IEEE Trans, on Signal Process sing, 2005. The methods for I/Q compensation presented in these materials envisage a BD precoder, and this precoder can be extended to any type of DIDO precoding.

In DIDO-OFDM systems, I/Q mismatch can cause two effects: ICI and inter-user interference. Similar to that in the SISO-OFDM system, the former is due to the interference from the image tone. The latter is caused by the fact that the I/Q mismatch destroys the quadrature of the DIDO precoder, thereby creating interference between users. Both types of interference can be canceled at the transmitter and receiver by the methods described herein. Three methods for I/Q compensation in DIDO-OFDM systems are described here and their performance is compared with and without I/Q mismatch. Results are presented based on simulations and actual measurements performed with a DIDO-OFDM prototype.

This embodiment is an extension of the earlier application. In particular, these embodiments relate to the following features of the prior application:

The system described in the earlier application, where the I/Q rails are affected by gain and phase imbalance;

At the transmitter, a DIDO precoder with I/Q compensation is computed using the training signal employed for channel estimation; and

The signal characterization data takes into account the distortion due to I/Q imbalance and is used at the transmitter to calculate the DIDO precoder according to the method proposed in this material.

b. Embodiments of the present invention

First, the mathematical model and architecture of the present invention will be described.

Before presenting the solution, it is useful to explain the core mathematical concepts. We explain it by assuming I/Q gain and phase imbalance (phase delay is not included in this description, but it will be handled automatically in the algorithm of DIDO-OFDM form). To explain the basic idea, suppose we want to multiply two complex numbers s=sI+jsQ and h=hI+jho such that x=h*s. We use subscripts to represent the in-phase quadrature components. Call the following equation:

x _I ＝s _I h _I -s _Q h _Q

and

x _Q =s _I h _Q +s _Q h _I

Its matrix form can be rewritten as:

The normalization transformation is denoted by the channel matrix (H). Now assume that s is the transmitted symbol and h is the channel. The presence of I/Q gain and phase imbalance can be modeled by creating the following unnormalized transform:

What this trick does is confirm that it can be written as:

Now rewrite (A):

We make the following definitions:

and

These two matrices have a normalized structure and can therefore be represented as complex numbers:

h _e =h ₁₁ +h ₂₂ +j(h ₂₁ -h ₁₂ )

and

h _c =h ₁₁ -h ₂₂ +j(h ₂₁ +h ₁₂ )

Using all this knowledge, we can derive the efficient equation back to a scalar form with two channels, the equivalent channel he and the conjugate channel hc. Thus, the effective transformation in (5) becomes:

x=h _e s+h _c s ^*

We call the first channel the equivalent channel and the second channel the conjugate channel. If there is no I/Q gain and phase imbalance, then the equivalent channel is the channel we want to observe.

Using similar arguments, the input-output relations of discrete-time MIMON×M systems with I/Q gain and phase imbalance can be shown as (by using scalar equivalents to establish their matrix counterparts):

where t is the discrete time index, he _e , h _c ∈ C ^M×N , s=[s ₁ ,...,s _N ], x=[x ₁ ,...,x _M ] and L is the channel The number of taps (channel tap).

In DIDO-OFDM systems, the received signal is represented in the frequency domain. Recall from signal and system if the following equations are satisfied:

FFT _K {s[t]}=S[k] then FFT _K {s ^* [t]}=S ^* [(-k)]=S ^* [Kk]for k=0,1,...,K -1

Using OFDM, for subcarrier k, the equivalent input-output relationship of MIMO-OFDM system is:

Among them, k=0,1,..., _{K-1 is the OFDM subcarrier index, He and H c represent} the equivalent and conjugate channel matrix respectively, defined as follows:

and

The second base value in (1) is the interference from the image tone. It can be handled by constructing the following stacked matrix system (note the conjugate values carefully):

in with are the vectors of the transmitted and received symbols in the frequency domain, respectively.

Using this method, an effective matrix can be built for DIDO operations. For example, with a DIDO 2×2 input-output relationship (assuming each user has a single receive antenna), the first user equipment may consider the following equation (in the absence of noise):

And the second user notices the following equation:

in, represent the m-th row of matrices He and H _c respectively, and _W∈C ^4x4 is the DIDO precoding matrix. According to (2) and (3), it can be noticed that the symbol received by user m Two sources of interference caused by I/Q imbalance (i.e., intercarrier interference from image tone (i.e., ) and inter-user interference (ie, as well as p≠m)). The DIDO precoding matrix W in (3) is designed to cancel these two interference terms.

There are several different implementations of DIDO precoders that can be used here, depending on the joint detection applied at the receiver. In one embodiment, it is possible to use the synthetic channel based on (instead of ) computed block diagonalization (BD) (see, e.g., QHSpencer, ALSwindlehurst, and M.Haardt, "Zeroforcing methods for downlink spatialmultiplexing in multiuser MIMO channels," IEEETrans.Sig.Proc, vol.52, pp.461- 471, Feb.2004.KK; Wong, RDMurch, and KBLetaief, "Ajointchannel diagonalization for multiuser MIMO antenna systems," IEEETrans.Wireless Comm., vol.2, pp.773-786, JuI 2003; LUChoi and RDMurch, "Atransmitpreprocessing technique for multiuser MIMO systems using adecomposition approach,” IEEE Trans.Wireless Comm., vol.3, pp.20-24, Jan2004; Z. Shen, JG Andrews, RW Heath, and BL Evans, “Low complexity user selection algorithms for multiuser MIMO systems with block diagonalization ,” accepted for publication in IEEE Trans.Sig.Proc, Sep.2005; Z.Shen, R.Chen, JG Andrews, RW Heath, and BL Evans, “Sumcapacity of multiuserMIMO broadcast channels with block diagonalization,” submitted to IEEE Trans.Wireless Comm., Oct. 2005). Therefore, currently the DIDO system selects the precoder such that:

Among them, α _{i, j} are constants, and This approach is very beneficial because by using this precoder the other aspects of the DIDO precoder can be kept intact as the effects of I/Q gain and phase imbalance are completely removed at the transmitter.

The DIDO precoder can also be designed to pre-cancel inter-user interference without pre-cancelling ICI due to IQ imbalance. Using this approach, the receiver (rather than the transmitter) can compensate for IQ imbalance by employing one of the receive filters described below. Therefore, the precoding design criteria in (4) can be modified as:

and

where for the mth transmitted signal, and is the symbol vector received by user m.

On the receiving side, in order to send symbol vectors to For estimation, user m applies a ZF filter, and the estimated symbol vector is given as:

While the ZF filter is the easiest to understand, the receiver can apply any number of other filters known to those skilled in the art. A popular choice is the MMSE filter, where:

And ρ is the signal-to-noise ratio. Optionally, the user can perform maximum likelihood symbol detection (or sphere decoder, iterative variation). For example, a first user may use an ML receiver and solve the following optimization:

Among them, S is the set of all possible vectors s, and depends on the size of the constellation diagram. The ML receiver gives better performance but requires higher complexity at the receiver. A similar set of equations can be applied to the second user.

Note that in (6) and (7) with is assumed to have zero entries. This assumption is only valid if the transmit precoder is able to completely cancel the inter-user interference for the criterion in (4). akin, with Diagonal matrix only if the transmit precoder is able to completely cancel the intercarrier interference (i.e. from image tone).

Figure 13 shows an embodiment of the architecture of a DIDO-OFDM system with I/Q compensation, the DIDO-OFDM system includes an IQ-DIDO precoder 1302 located in the base station (BS), a transmission channel 1304, located in the user Channel estimation logic 1306 and ZF, MMSE or ML receiver 1308 within the device. The channel estimation logic 1306 estimates the channel via the training signal with Estimates are made and these estimates are fed back to the precoder within the AP. The BS calculates DIDO precoder weights (matrix W) to pre-cancel interference caused by I/Q gain and phase imbalance and user interference, and transmits data to users through the wireless channel 1304 . User equipment m employs a ZF, MMSE or ML receiver 1308 to cancel the remaining interference by utilizing the channel estimate provided by unit 1304 and demodulates the data.

The following three implementations can be used to implement the I/Q compensation algorithm.

Method 1 - TX Compensation: In this embodiment, the transmitter calculates the precoding matrix according to the criteria in (4). At the receiver, the UE employs a "simplified" ZF receiver, where with is assumed to be a diagonal matrix. Therefore, formula (8) simplifies to:

Method 2 - RX Compensation: In this embodiment, the transmitter is based on R.Chen, R.W.Heath, and J.G.Andrews, "Transmit selection diversity for unitary precodedmultiuserspatial multiplexing systems with linear receivers," accepted to IEEETrans, onSignal Processing, 2005 The traditional BD method described in , calculates the precoding matrix, and does not address the criteria in (4) to eliminate inter-carrier and inter-user interference. Using this approach, the precoding matrices in (2) and (3) simplify to:

At the receiver, the user equipment employs a ZF filter as in (8). Note that this method does not pre-cancel interference at the transmitter as in Method 1 above. Therefore, it cancels inter-carrier interference at the receiver, but not inter-user interference. Also, compared to method 1 requiring feedback with In method 2, the user only needs to feed back the vector for the transmitter to compute the DIDO precoder. Therefore, Method 2 is especially suitable for DIDO systems with low-rate feedback channels. On the other hand, Method 2 requires a slightly higher computational complexity at the user equipment to compute the ZF receiver in (8) instead of (11).

Method 3-TX-RX Compensation: In one embodiment, the above two methods are combined. The transmitter calculates the precoding matrix as in (4), and the receiver estimates the transmitted symbols according to (8).

I/Q imbalances (whether phase imbalances, gain imbalances, or delay imbalances) can detrimentally degrade signal quality in wireless communication systems. For this reason, conventional circuits have been designed to have low unbalance. However, as mentioned above, this problem can be corrected by using digital signal processing in the form of transmit precoding and/or specific receivers. One embodiment of the present invention includes a system with several new functional units, each of which is important for realizing I/Q correction in an OFDM communication system or a DIDO-OFDM communication system.

One embodiment of the present invention uses precoding based on channel condition information to eliminate inter-carrier interference (ICI) (caused by I/Q mismatch) from image modulation in OFDM systems. As shown in Figure 11, the DIDO transmitter according to this embodiment includes a user selector unit 1102, a plurality of code modulation units 1104, a plurality of corresponding mapping units 1106, a DIDO IQ known precoding unit 1108, and a plurality of RF transmitters unit 1114 , user feedback unit 1112 , and DIDO configurator unit 1110 .

The user selector unit 1102 selects data associated with a plurality of users U ₁ -U _M based on the feedback information obtained by the feedback unit 1112, and provides the information to each coded modulation unit 1104 in the plurality of coded modulation units 1104 Unit 1104. Each coding and modulating unit 1104 codes and demodulates the information bits of each user, and sends them to the mapping unit 1106 . The mapping unit 1106 maps the input bits to complex symbols and sends the result to the DIDO IQ-aware precoding unit 1108 . The DIDO IQ known precoding unit 1108 calculates DIDO IQ known precoding weights by using the channel state information obtained from the user by the feedback unit 1112 , and precodes the input symbols obtained from the mapping unit 1106 . Each precoded data stream is sent by the DIDO IQ known precoding unit 1108 to the OFDM unit 1115, and the OFDM unit 1115 calculates IFFT and adds a cyclic prefix. This information is sent to the D/A unit 1116 which performs digital to analog conversion and sends it to the RF unit 1114 . The RF unit 1114 upconverts the baseband signal to IF/RF and sends it to the transmit antenna.

The precoder operates on both normal and mirrored modulation to compensate for I/Q imbalance. Any number of precoder design criteria can be used, including ZF, MMSE or weighted MMSE designs. In a preferred embodiment, the precoder can completely remove the ICI caused by the I/Q mismatch, so that the receiver does not need to perform any additional compensation.

In one embodiment, the precoder uses a block diagonalization criterion to completely cancel inter-user interference without completely canceling the I/Q impact of each user, which requires additional receiver processing. In another embodiment, the precoder uses a zero forcing criterion to completely eliminate inter-user interference and ICI interference caused by I/Q imbalance. This embodiment can use a conventional DIDO-OFDM processor at the receiver.

One embodiment of the present invention uses precoding based on channel condition information to eliminate inter-carrier interference (ICI) (caused by I/Q mismatch) from image modulation in DIDO-OFDM systems, and each user adopts IQ DIDO receivers are known. As shown in FIG. 12, in one embodiment of the present invention, the system (including the receiver 1202) includes a plurality of RF units 1208, a corresponding plurality of A/D units 1210, an IQ known channel estimator unit 1204, and a DIDO Feedback generator unit 1206 .

The RF unit 1208 receives the signal transmitted from the DIDO transmitter unit 1114 , down-converts the signal to baseband, and provides the down-converted signal to the A/D unit 1210 . Afterwards, the A/D unit 1210 performs analog-to-digital conversion on the signal and sends it to the OFDM unit 1213 . The OFDM unit 1213 removes the cyclic prefix and performs FFT to report the signal to the frequency domain. During the training period, the OFDM unit 1213 sends output to the IQ-aware channel estimation unit 1204, which computes the channel estimate in the frequency domain. Optionally, the channel estimate may be computed in the time domain. During a data period, the OFDM unit 1213 sends an output to the IQ-aware receiver unit 1202 . The IQ known receiver unit calculates an IQ receiver and demodulates/decodes the signal to obtain data 1214. The IQ-aware channel estimation unit 1204 sends the channel estimate to the DIDO feedback generator unit 1206, which may quantize the channel estimate and send it back to the transmitter via the feedback control channel 1112 .

The receiver 1202 shown in FIG. 12 may operate under any number of standards known to those skilled in the art, including ZF, MMSE, maximum likelihood or MAP receivers. In a preferred embodiment, the receiver uses an MMSE filter to remove ICI caused by IQ imbalance on image modulation. In another preferred embodiment, the receiver uses a non-linear detector similar to a maximum likelihood search to jointly detect the image-on-tone symbols. This method has good performance but has higher complexity.

In one embodiment, an IQ-aware channel estimator 1204 is used to determine receiver coefficients to remove ICI. Therefore, we require a DIDO-OFDM system (using precoding based on channel condition information to cancel inter-carrier interference (ICI) from image tones (caused by I/Q mismatch)), an IQ-aware DIDO receiver, and an IQ The benefit of the known channel estimator. The channel estimator may use conventional training signals, or may use specially constructed training signals sent on in-phase quadrature signals. Any number of estimation algorithms can be implemented, including least squares, MMSE or maximum likelihood. The IQ-aware channel estimator provides input to an IQ-aware receiver.

Channel condition information may be provided to the stations through channel reciprocity or through a feedback channel. An embodiment of the present invention includes a DIDO-OFDM system with an I/Q-aware precoder and an I/Q-aware feedback channel for transmitting channel condition information from a user terminal to a station. The feedback channel can be a physical or logical control channel. It can be dedicated or shared in the random access channel. Feedback information can be generated by using a DIDO Feedback Generator at the user terminal (which we also claim the rights to). The DIDO feedback generator takes as input the output of the I/Q known channel estimator. It can quantize the channel coefficients, or it can use any number of finite feedback algorithms known in the art.

Allocation of users, modulation and coding rates, mapping to space-time-frequency coded slots can vary according to the results of the DIDO feedback generator. Therefore, an embodiment includes an IQ-aware DIDO configurator that uses IQ-aware channel estimates from one or more users to configure a DIDO IQ-aware precoder, selects modulation rate, coding rate, users allowed to transmit and their mappings to space-time-frequency coded slots.

To evaluate the performance of the proposed compensation method, three DIDO 2×2 systems will be compared:

1. With I/Q mismatch: send through all modulations (except DC modulation and edge modulation), and do not compensate for I/Q mismatch;

2. With I/Q compensation: send through all tunes, and compensate for I/Q mismatch by using "method 1" above;

3. Ideal: transmit over odd tones only to avoid inter-user interference as well as inter-carrier (ie, from imaged tones) interference due to I/Q mismatch.

Following this, the results obtained from measurements with the DIDO-OFDM prototype in real propagation scenarios are shown. Fig. 14 shows the 64-QAM constellation diagrams obtained from the above three systems. These constellations were acquired at the same user location and with a fixed average SNR (~45dB). The first constellation 1401 is very noisy (interference from image tones due to I/Q imbalance). The second constellation diagram 1402 shows some improvement (due to I/Q compensation). Note that the second constellation diagram 1402 is not as clean as the ideal case shown in constellation diagram 1403 (due to the presence of phase noise that may generate inter-carrier interference (ICI)).

Figure 15 shows the average SER (Symbol Error Rate) 1501 and the actual throughput per user ( Goodput) 1502. OFDM bandwidth is 250KHZ, has 64 tones and cyclic prefix length L _cp =4. Since in an ideal case, we only send data through a subset of tones, the SER and actual throughput performance are evaluated based on the average transmit power per tone (rather than the total transmit power) to ensure fairness between different cases Compare. Also, in the following results, we use the normalized value of transmit power (expressed in decibels), since our goal here is to compare the relative (not absolute) performance of different schemes. Figure 15 shows that in the presence of I/Q imbalance, the SER is saturated and the target SER (~10 ^-2 ) is not reached, which is consistent with A.Tarighat and A.H.Sayed, "MIMO OFDM receivers for systems with IQ imbalances, "The results reported in IEEE Trans.Sig.Proc, vol.53, pp.3583-3596, Sep.2005 are consistent. This saturation effect is due to the fact that the signal power and the interference power (from the image tone) increase with increasing TX power. However, with the proposed I/Q compensation method, interference can be eliminated and better SER performance can be obtained. Note that since 64-QAM modulation requires large transmit power, there may be a slight increase in SER at high SNR due to amplitude saturation effects in the DAC.

Furthermore, it can be observed that in the presence of I/Q compensation, the SER performance is very close to ideal. The 2dB gap in TX power between these two cases is due to phase noise which may create additional interference between adjacent OFDM tones. Finally, the actual throughput curve 1502 shows that when the I/Q method is applied, it can send twice as much data compared to the ideal case because we use all data tones instead of only odd tones (for the ideal case).

Figure 16 illustrates the SER performance of different QAM constellations with or without I/Q compensation. We can observe that, in this embodiment, the proposed method is particularly advantageous for 64-QAM constellations. For 4-QAM and 16-QAM, the I/Q compensation method will produce worse performance than the case with I/Q mismatch, which may be because the proposed method requires more power for data transmission and Interference cancellation from mirrored tones. Furthermore, 4-QAM and 16-QAM are not as affected by I/Q mismatch as 64-QAM due to the larger minimum distance between constellation points. See A.Tarighat, R. Bagheri, and A.H. Sayed, "Compensation schemes and performance analysis of IQ imbalances in OFDM receivers," Signal Processing, IEEE Transactions on [see also Acoustics, Speech, and Signal Processing, IEEE Transactions on], vol.53, pp.3257-3268, Aug.2005. This conclusion can also be drawn by observing Figure 16 and comparing the I/Q mismatch to the ideal case for 4-QAM and 16-QAM. Therefore, the additional power required for a DIDO precoder with interference cancellation (from image modulation) does not warrant a small benefit of I/Q compensation for the 4-QAM and 16-QAM cases. Note that this problem can be solved by employing I/Q compensation methods 2 and 3 above.

Finally, the relative SER performance of the above three methods is measured under different propagation scenarios. The SER performance in the presence of I/Q mismatch is also described for reference. FIG. 17 shows the measured SER at two different user locations for a 64-QAM DIDO 2×2 system with a carrier frequency of 450.5 MHz and a bandwidth of 250 KHz. At location 1, the user is ~6λ away from the BS in a different room and in NLOS (no line-of-sight) state. At location 2, the user is at a distance of ~λ from the BS with LOS (Line of Sight).

Figure 17 shows that all three compensation methods outperform no compensation. However, it should be noted that Method 3 outperforms the other two compensation methods in any channel situation. The relative performance of methods 1 and 2 depends on the propagation situation. From actual measurement campaigns, it can be concluded that method 1 generally outperforms method 2 because it preempts (at the transmitter) inter-user interference caused by I/O imbalance. When the inter-user interference is small, as shown in the graph 1702 of FIG. 17 , method 2 can outperform method 1 because it does not suffer from the power loss caused by the I/Q compensation precoder.

So far, the different methods have been compared by considering only the finite group propagation case (as shown in Figure 17). Afterwards, the relative performance of these methods is measured in ideal i.i.d. (independent and identically distributed) channels. Simulate DIDO-OFDM systems with I/Q phase and gain imbalances on the transmit and receive sides. Fig. 18 shows the performance of the proposed method with gain balancing on the transmitter side only (ie, with a gain of 0.8 on the I rail of the first transmit chain and a gain of 1 on the other rails). It can be seen that method 3 outperforms all other methods. Furthermore, method 1 may perform better than method 2 in the i.i.d. channel compared to the results obtained at position 2 in the graph 1702 of FIG. 17 .

Therefore, three novel methods are given to compensate the I/Q imbalance in the above DIDO-OFDM system, and method 3 outperforms other proposed compensation methods. In systems with low-rate feedback channels, method 2 can be used to reduce the amount of feedback required for DIDO precoding, but results in poorer SER performance.

Ⅱ. Adaptive DIDO transmission scheme

Another embodiment of a system and method for enhancing the performance of a Distributed-Input Distributed-Output (DIDO) system will be described. The method dynamically allocates radio resources to different user equipments by tracking the changing channel state to increase throughput while meeting certain target bit error rates. The user equipment estimates the channel quality and feeds it back to the base station (BS); the base station processes the channel quality obtained from the user equipment to select the best user equipment set, DIDO scheme, Modulation/coding scheme (MCS) and array configuration; the base station transmits parallel data to multiple user equipments via precoding, and the signal is demodulated at the receiver.

A system for efficiently allocating resources for DIDO wireless links is also described. The system includes a DIDO base station with a DIDO configurator that processes feedback received from users to select the best set of users, DIDO scheme, modulation/coding scheme (MCS), and array configuration for the next transmission; DIDO A receiver in the system that measures the channel and other related parameters to generate a DIDO feedback signal; and a DIDO feedback control channel for transmitting feedback information from users to the base station.

As detailed below, some salient features of this embodiment of the invention may include, but are not limited to:

Used to adaptively select the number of users, DIDO transmission scheme (i.e., antenna selection or multiplexing), modulation/coding scheme (MCS), and array configuration to minimize SER, or maximize spectrum per user, based on channel quality information efficiency or downlink spectral efficiency techniques;

Techniques for defining multiple sets of DIDO transmission patterns as a combination of DIDO schemes and MCS;

Techniques for assigning different DIDO patterns to different slots, OFDM tones and DIDO sub-streams depending on channel state;

Techniques for dynamically assigning different DIDO patterns to different users based on their channel quality;

Criteria for activation of adaptive DIDO handover based on link quality metrics computed in time, frequency and space domains;

Standard for lookup table based activation of adaptive DIDO switching.

19 shows a DIDO system with a DIDO configurator at the base station, which can adaptively select the number of users, DIDO transmission scheme (i.e., antenna selection or multiplexing), modulation/coding scheme ( MCS) and array configurations to minimize SER, or maximize spectral efficiency per user or downlink spectral efficiency;

A DIDO system with a DIDO configurator at the base station and a DIDO feedback generator at each user equipment as shown in Figure 20 uses the estimated channel conditions and/or other parameters at the receiver (similar to the estimated SNR) to generate a feedback message that is input to the DIDO configurator.

A DIDO system with a DIDO configurator (at the base station), a DIDO feedback generator, and a DIDO feedback control channel (the DIDO feedback channel is used to transmit DIDO specific configuration information from the user to the base station).

a. Background

In multiple-input multiple-output (MIMO) systems, diversity schemes (e.g., Orthogonal Space-Time Block Codes (OSTBC)) can be conceived (see V.Tarokh, H. Jafarkhani, and A.R. Calderbank, "Spacetime block codes from orthogonal designs," IEEE Trans.Info.Th., vol.45, pp.1456-467, JuI.1999) or antenna selection (see R.W.Heath Jr., S.Sandhu, and A.J.Paulraj, "Antenna selection for spatial multiplexing systems with linear receivers, "IEEE Trans.Comm., vol.5, pp.142-144, Apr.2001) to prevent channel fading and improve link reliability (which translates into better coverage). On the other hand, Spatial multiplexing (SM) can use multiple parallel data transmissions as a means to enhance system throughput. See G.J. Foschini, G.D. Golden, R.A. Valenzuela, and P.W. Wolniansky, "Simplifed processing for high spectral efficiency wireless communication employing multielement arrays," IEEE Jour.Select. Areas in Comm., vol.17, no.11, pp.1841-1852, Nov. 1999. According to L. Zheng and D.N.C.Tse, "Diversity and multiplexing: a fundamental tradeoffin multiple antenna channels," IEEE Trans.Info. Th., vol.49, no.5, pp.1073-1096, May 2003 Theoretical diversity/multiplexing trade-offs, these benefits can be realized simultaneously in MIMO systems.A practical implementation form is by tracking the changing channel state, Adaptive switching between diversity and multiplexing transmission schemes.

A large number of adaptive MIMO transmission techniques have been proposed. The diversity/multiplexing switching method in R.W.Heath and A.J.Paulraj, "Switching between diversity and multiplexing in MIMO systems," IEEETrans.Comm., vol.53, no.6, pp.962-968, Jun.2005 is designed based on Instantaneous channel quality information, improving BER (Bit Error Rate) for fixed rate transmissions. Alternatively, as in S.Catreux, V.Erceg, D.Gesbert, and R.W.Heath.Jr., "Adaptive modulation and MIMO coding for broadband wirelessdatanetworks," IEEE Comm.Mag., vol.2, pp.108- 115, as in June 2002 ("Catreux"), uses statistical channel information to activate the adaptation, thereby reducing the feedback overhead and the number of control messages. Adaptive transmission algorithms in Catreux are designed to enhance spectral efficiency for predetermined target bit error rates in Orthogonal Frequency Division Multiplexing (OFDM) systems based on channel time/frequency selection indicators. Also for narrowband systems, a similar low-feedback adaptive approach is proposed that exploits channel spatial selectivity to switch between diversity schemes and spatial multiplexing. See, eg, A. Forenza, M.R. McKay, A. Pandharipande, R.W. Heath. Jr., and I.B. Collings, "Adaptive MIMO transmission for exploiting the capacity of spatially correlated channels," accepted to the IEEE Trans, on Veh. Tech., M ar. 2007 ; M.R.McKay, I.B.Collings, A.Forenza, and R.W.Heath.Jr., “Multiplexing/beamforming switching for coded MIMO in spatially correlated Rayleigh channels,” Accepted to IEEE Trans, on Veh.Tech., Dec. 2007; A.Forenza , M.R.McKay, R.W.Heath.Jr., and I.B.Collings, "Switching between OSTBC and spatial multiplexing with linear receivers spatially correlated MIMO channels," Proc.IEEE Veh.Technol.Conf.,vol.3,pp.1387-1391,May 2006; M.R.McKay, I.B.Collings, A.Forenza, and R.W.Heath Jr., “Athroughput-based adaptive MIMO BICMproach for spatially correlated channels,” in Proc. IEEE ICC, June 2006.

In this work, we extend the scope of work presented in various previous publications to DIDO-OFDM systems. See eg R.W.Heath and A.J.Paulraj, "Switching betWeendiversity and multiplexing in MIMO systems," IEEE Trans.Comm., vol.53, no.6, pp.962-968, Jun. 2005; S.Catreux, V.Erceg, D .Gesbert, and R.W.Heath Jr., "Adaptive modulation and MIMO coding for broadband wireless datanetworks," IEEE Comm.Mag., vol.2, pp.108-115, June 2002; A.Forenza, M.R.McKay, A.Pandharipande, R.W. Heath Jr., and I.B. Collings, "Adaptive MIMO transmission for exploiting the capacity of spatially correlated channels," IEEE Trans, on Veh. Tech., vol.56, n.2, pp.619-630, Mar. 2007; M.R. McKay , I.B.Collings, A.Forenza, and R.W.Heath Jr., “Multiplexing/beamforming switching for coded MIMO in spatially correlated Rayleighchannels,” accepted to IEEE Trans, on Veh.Tech., Dec. 2007; A.Forenza, M.R.McKay, R.W. Heath, Jr., and I.B. Collings, “Switching between OSTBC and spatial multiplexing with linear receivers in spatially correlated MIMO channels,” Proc.IEEEVeh.Technol.Conf., vol.3, pp.1387-1391, May 2006; M.R.McKays, I.B.Coll , A. Forenza, and R.W. Heath Jr., “A throughput-based adaptive MIMO BICMproach forspatially co related channels," appeared in Proc. IEEE ICC, June 2006.

A novel adaptive DIDO transmission strategy is described here that switches between different numbers of users, different numbers of transmit antennas, and transmission schemes based on channel quality information as a means to improve system performance. Note, M. Sharif and B. Hassibi, "On the capacity of MIMO broadcast channel with partialside information," IEEE Trans.Info.Th., vol.51, p.506522, Feb.2005 and W.Choi, A.Forenza, J.G. Andrews, and R.W. Heath Jr., "Opportunistic space division multiple access with beam selection," appeared in IEEE Trans, on Communications. A scheme for adaptive selection of users in multi-user MIMO systems has been proposed. However, the opportunistic space division multiple access (OSDMA) schemes in these publications are designed to maximize the total capacity by exploiting multi-user diversity, and they are only able to realize the Partial theoretical capacity, since the interference is not fully pre-cancelled at the transmitter. In the DIDO transmission algorithm described herein, block diagonalization is employed to pre-cancel inter-user interference. However, the proposed adaptive transmission strategy can be applied to any DIDO system regardless of the type of precoding technique.

This patent application describes the invention described above and extensions of the embodiments of the earlier applications, including but not limited to the following additional features:

1. The link quality metrics in the adaptive DIDO scheme can be evaluated by wireless client devices using the training symbols used for channel estimation in the prior application.

2. A base station receives signal characteristic data from a client device as described in the prior application. In the current implementation, signal characteristic data is defined as a link quality metric for activating adaptation.

3. The prior application describes a mechanism for selecting antennas and number of users and defines throughput allocation. Furthermore, different levels of throughput can be dynamically assigned to different clients as in the prior application. The current implementation of the invention defines a new type of criteria related to this selection and throughput allocation.

b. Embodiments of the present invention

The proposed adaptive DIDO technique aims to enhance per-user spectral efficiency or downlink spectral efficiency by dynamically allocating radio resources in time, frequency, and space to different users in the system. This overall adaptive criterion is used to increase throughput while meeting the target bit error rate. Depending on the propagation state, this adaptive algorithm can also be used to improve the link quality (or coverage) of the user via a diversity scheme. Figure 21 shows a flowchart describing the steps of the adaptive DIDO scheme.

At 2102, a base station (BS) collects channel condition information from all users. At 2104, based on the received CSI, the base station calculates a link quality metric in time domain/frequency domain/space domain. At 2106, the link quality metrics are used to select the users to be served in the next transmission, and the transmission mode for each user. Note that the transmit modes include different combinations of modulation/coding and DIDO schemes. Finally, at 2108, the BS transmits the data to the user via DIDO precoding.

At 2102, the base station selects channel condition information (CSI) from all user equipments. At 2104, the base station uses the CSI to determine instantaneous or statistical channel quality for all user equipments. In DIDO-OFDM systems, channel quality (or link quality metrics) can be estimated in time, frequency and space domains. Thereafter, at 2106, the base station uses the link quality metric to determine the best subset of users and transmission mode for the current propagation state. A set of DIDO transmission patterns is combined as a combination of DIDO scheme (ie, antenna selection or multiplexing), modulation/coding scheme (MCS), and array configuration. At 2108, the data is transmitted to the user equipment using the selected number of users and the transmission mode.

Mode selection may be performed by a look-up table (LUT) that is pre-computed based on the bit error rate performance of the DIDO system in different propagation environments. These LUTs map channel quality information to bit error rate performance. In order to construct the LUT, the BER performance of the DIDO system in different propagation scenarios can be evaluated according to the SNR. From the bit error rate curve, it can be seen that the minimum SNR required to achieve a predetermined target bit error rate can be calculated. We define this SNR requirement as the SNR threshold. Afterwards, the SNR threshold is evaluated in different propagation scenarios and for different DIDO transmission modes and stored in the LUT. For example, the SER results in Figure 24 and Figure 26 can be used to construct the LUT. Then, based on this LUT, the base station can select a transmission mode for active users that improves throughput while meeting a predetermined target bit error rate. Finally, the base station transmits the data to the selected users via DIDO precoding. Note that different DIDO patterns can be assigned to different time slots, OFDM tones and DIDO sub-streams so that adaptation can be done in time, frequency and space domains.

Figures 19-20 show an embodiment of a system using DIDO adaptation. Several new functional units are introduced to implement the proposed DIDO adaptive algorithm. Specifically, in one embodiment, the DIDO configurator 1910 can perform various functions based on the channel quality information 1912 provided by the user equipment, including selecting the number of users, DIDO transmission scheme (ie, antenna selection and multiplexing), Modulation/coding scheme (MCS) and array configuration.

User selector unit 1902 selects data associated with multiple users U _{1 -UM} based on the feedback information obtained by DIDO configurator 1910 and provides this information to each coded modulation unit 1904 in each plurality of coded modulation units 1904. unit. Each coding and modulation unit 1904 codes and modulates the information bits of each user, and sends them to the mapping unit 1906 . The mapping unit 1906 maps input bits to complex symbols and sends them to the precoding unit 1908 . Both the coded modulation unit 1904 and the mapping unit 1906 use the information obtained from the DIDO configurator unit 1910 to select the type of modulation/coding scheme to be used for each user. The information may be calculated by the configurator unit 1910 by utilizing the channel quality information of each user provided by the feedback unit 1912 . The DIDO precoding unit 1908 uses the information obtained by the DIDO configurator unit 1910 to calculate DIDO precoding weights and precodes the input symbols obtained from the mapping unit 1906 . Each precoded data stream is sent by DIDO precoding unit 1906 to OFDM unit 1915, which calculates IFFT and adds cyclic prefix. This information is sent to D/A unit 1916 which performs digital to analog conversion and sends the final analog signal to RF unit 1914 . The RF unit 1914 upconverts the baseband signal to IF/RF and sends it to the transmit antenna.

The RF unit 2008 of each client device receives the signal transmitted from the DIDO transmitter unit 1914 , down-converts the signal to baseband, and supplies the down-converted signal to the A/D unit 2010 . Afterwards, the A/D unit 2010 converts the signal from analog to digital and sends it to the OFDM unit 2013 . The OFDM unit 2013 removes the cyclic prefix and performs FFT to report the signal to the frequency domain. During the training period, the OFDM unit 2013 sends output to the channel estimation unit 2004, which computes the channel estimate in the frequency domain. Alternatively, the channel estimate can be computed in the time domain. During data periods, OFDM unit 2013 sends output to receiver unit 2002 which demodulates/decodes the signal to obtain data 2014 . The channel estimation unit 2004 sends the channel estimate to the DIDO feedback generator unit 2006 which may quantize the channel estimate and send it back to the transmitter via the feedback control channel 1912 .

The DIDO configurator 1910 may use information obtained at the base station or, in a preferred embodiment, additionally use the output of the DIDO feedback generator 2006 (see Fig. 20 ) operating at each user equipment. The DIDO feedback generator 2006 uses the estimated channel conditions 2004 and/or other parameters at the receiver like the estimated SNR to generate feedback messages to be input to the DIDO configurator 1910 . The DIDO feedback generator 2006 can compress, quantize and/or use some limited feedback strategy known in the art at the receiver to compress the information.

The DIDO configurator 1910 may use information recovered from the DIDO feedback control channel 1912 . The DIDO Feedback Control Channel is a logical or physical control channel that can be used to transmit the output of the DIDO Feedback Generator 2006 from the user to the base station. Control channel 1912 may be implemented in any number of ways known in the art and may be a logical or physical control channel. As a physical channel, it may comprise dedicated time/frequency slots assigned to users. It can also be a random access channel shared by all users. The control channel may be pre-assigned, or may be created by stealing bits in an existing control channel in a predetermined manner.

In the following discussion, the results obtained by making measurements with the DIDO-OFDM prototype will be described in a real propagation environment. These results demonstrate the achievability of potential gains in adaptive DIDO systems. The performance of different classes of DIDO systems is first presented, showing that the number of antennas/users can be increased to achieve greater downlink throughput. Afterwards, the DIDO performance in relation to the location of the user equipment is described, showing the need to track changing channel states. Finally, the performance of the DIDO system using diversity technology is described.

i. Performance of different levels of DIDO systems

The performance of different DIDO systems is evaluated with an increasing number of transmit antennas (N=M, where M is the number of users). Compare the performance of the following systems: SISO, DIDO 2×2, DIDO 4×4, DIDO 6×6, and DIDO 8×8. DIDON x M refers to DIDO with N transmit antennas and M users at the BS.

Figure 22 shows the transmit/receive antenna layout. The transmit antennas 2201 are arranged in a square array configuration with users positioned around the transmit array. In FIG. 22 , T refers to a "transmit" antenna and U refers to a "user equipment" 2202 .

Different subsets of antennas in the 8-element transmit array are active, depending on the value of N chosen for different measurements. For each DIDO level (N), the subset of antennas that can cover the largest real area targeted by the fixed size constraint of the 8-element array is selected. This criterion is expected to enhance spatial diversity for a given value of N.

Figure 23 shows array configurations for different DIDO levels fitting the available real area (ie, dashed lines). The square dotted box has dimensions of 24"x24", corresponding to ~λxλ at the 450MHz carrier frequency.

Based on the comments related to Figure 23 and with reference to Figure 22, the performance of each of the following systems will now be defined and compared:

SISO with T1 and U1 (2301)

DIDO 2×2 with T1,2 and U1,2 (2302)

DIDO 4×4 with T1,2,3,4 and U1,2,3,4 (2303)

DIDO 6×6 with T1,2,3,4,5,6 and U1,2,3,4,5,6 (2304)

DIDO 8×8 with T1,2,3,4,5,6,7,8 and U1,2,3,4,5,6,7,8 (2305)

Figure 24 shows the SER, BER, SE (Spectral Efficiency) and actual throughput performance as a function of transmit (TX) power for the DIDO system described above for 4-QAM and 1/2 FEC (Forward Error Correction) rate . It has been observed that the SER and BER performance degrades as the value of N increases. This effect is caused by two phenomena: for a fixed TX power, the input power of the DIDO array is split among more and more users (or data streams); the spatial diversity increases with the actual DIDO channel decrease as the number of users increases.

As shown in Fig. 24, in order to compare the relative performance of different levels of DIDO systems, the target BER is fixed at 10 ^-4 (this value can vary according to the system), which roughly corresponds to SER=10 ^-2 . We refer to the TX power value corresponding to this target as TX Power Threshold (TPT). For any N, if the TX power is lower than TPT, we assume that it is not possible to transmit at DIDO level N, and we need to switch to a lower level of DIDO. Furthermore, in Fig. 24, it can be observed that when the TX power exceeds the TPT for any value of N, the SE and actual throughput performance saturates. Based on these results, an adaptive transmission strategy can be designed to switch between different levels of DIDO to enhance SE or actual throughput for a fixed predetermined target error rate.

ii. Performance in case of variable user location

The goal of this experiment is to evaluate the DIDO performance at different user locations via simulations in spatially correlated channels. A DIDO 2x2 system is considered to have 4QAM and 1/2FEC rate. As shown in FIG. 25 , User 1 is located in the broadside direction of the transmit array, while User 2's position changes from the broadside direction to the endfire direction. The transmitting antennas are separated by -λ/2 and separated by -2.5λ from the user.

Fig. 26 shows the SER and SE results per user for different locations of user equipment 2. The user equipment has an Angle of Arrival (AOA) of 0° to 90° as measured from the broadside of the transmitting array. It is observed that as the angular separation of the UEs increases, the DIDO performance will improve due to the greater diversity within the DIDO channel. Furthermore, at the target SER=10 ⁻² , there is a gap of 10 dB between the two cases of AOA2=0° and AOA2=90°. This result is consistent with the simulation results obtained for an angular extension of 10° in Figure 35. Also, note that for the case of AOA1=AOA2=0°, there may be coupling effects between the two users (caused by the proximity of their antennas), which may make their performance different from the simulation results in Figure 35 different.

ⅲ. Optimal situation for DIDO 8×8

Figure 24 shows that DIDO 8×8 produces larger SEs than lower-level DIDOs, but has higher TX power requirements. The goal of this analysis is to show that there is a case where DIDO 8×8 outperforms DIDO 2×2 not only in terms of peak spectral efficiency (SE), but also in terms of TX power requirements (or TPT) to achieve said peak SE .

Note that in the i.i.d. (ideal) channel, there is a ~6dB gap in TX power between the SE of DIDO 8×8 and DIDO 2×2. The gap is due to the fact that DIDO 8x8 splits the TX power between 8 data streams, while DIDO 2x2 splits it between only two streams. The results are shown via simulation in FIG. 32 .

However, in spatially correlated channels, TPT is a function of propagation environment characteristics (eg, array orientation, user location, angular spread). For example, Figure 35 shows a ~15dB gap for low angular spread between two different user equipment locations. Similar results are shown in Figure 26 of the present application.

Similar to MIMO systems, the performance of DIDO systems degrades (due to lack of diversity) when the user is located in the end-fire direction of the TX array. This effect can be observed by taking measurements with the current DIDO prototype. So, one way to show that DIDO 8x8 is better than DIDO 2x2 is to put the user in an end-fire orientation relative to the DIDO 2x2 array. In this case, DIDO 8x8 outperforms DIDO 2x2 because the 8-antenna array provides higher diversity.

In this analysis, the following systems were considered:

System 1: 4-QAM DIDO 8×8 (send 8 parallel data streams per slot);

System 2: 64-QAM DIDO 2×2 (every 4 time slots, send users X and Y once). For this system, we consider four combinations of TX and RX antenna positions: a) T1,T2U1,2 (endfire direction); b) T3,T4U3,4 (endfire direction); c) T5,T6U5,6 ( ~30° from endfire direction); d) T7,T8U7,8 (NLOS (no line-of-sight));

System 3: DIDO 8×8 with 64-QAM; and

System 4: MISO 8×1 of 64-QAM (every 8 time slots, the sending user X is sent once).

For all these cases, an FEC rate of 3/4 is used.

Figure 27 depicts the user's location.

In Fig. 28, the SER results show a ~ 15dB gap between systems 2a and 2c due to different array orientations and user positions (similar to the simulation results in Fig. 35). The first panel in the second row shows the value of the TX power at which the SE curve is saturated (ie, corresponding to BER 1e-4). We observe that System 1 yields a larger SE per user than System 2 for lower TX power requirements (less than ~5dB). Moreover, due to the multiplexing gain of DIDO 8×8 over DIDO 2×2, the benefits of DIDO 8×8 over DIDO 2×2 are more obvious for DL (downlink) SE and DL actual throughput. System 4 has lower TX power requirements (less than 8dB) than System 1 due to beamforming array gain (ie, MRC with MISO 8×1). But System 4 only generates 1/3 of the SE per user compared to System 1. System 2 performed worse than System 1 (ie, yielded lower SE for larger TX power demands). Finally, System 3 yields a much larger SE (due to the larger order modulation) than System 1 for larger TX power requirements (~15dB).

From these results, the following conclusions can be inferred:

One channel scenario was identified as DIDO 8×8 outperforms DIDO 2×2 (i.e. yields larger SE for lower TX power requirements);

In this channel scenario, DIDO 8×8 produces larger SE and DLSE per user than DIDO 2×2 and MISO 8×1; and

The performance of DIDO 8×8 can be further increased by using higher order modulation (i.e. 64-QAM instead of 4-QAM) at the expense of larger TX power requirements (greater than ~15dB).

iv. DIDO with antenna selection

Below, we evaluate the benefits of the antenna selection algorithm described in "Transmit selection diversity for unitary precoded multiuser spatial multiplexing systems with linear receivers" published in Signal Processing by R. Chen, R.W. Heath and J.G. Andrews in 2005, accepted by IEEE Transactions on . We present results for a specific DIDO system with two users, 4-QAM, and a FEC rate of 1/2. The following systems are compared in Figure 27:

DIDO 2×2 with T1,2 and U1,2; and

DIDO 3×2 using antenna selection with T1,2,3 and U1,2.

The transmit antenna locations and user equipment locations are the same as in FIG. 27 .

Figure 29 shows that DIDO 3x2 with antenna selection can provide ~5dB gain compared to a DIDO 2x2 system (without selection). Note that the channel is nearly static (i.e. no Doppler effect), so the selection algorithm works with path loss and channel spatial correlation, rather than fast fading. We should see different gains in cases with high Doppler effect. Also, in this particular experiment, it was observed that the antenna selection algorithm selected antennas 2 and 3 for transmission.

iv. SNR threshold for LUT

In option [0171], we state that mode selection is achieved via LUTs. LUTs can be pre-computed by evaluating SNR thresholds to achieve certain predefined target bit error rate performances for DIDO transmission modes in different propagation environments. Below, we provide the performance of DIDO systems with and without antenna selection and variable number of users, which can be used as a guideline for constructing LUTs. Although Figure 24, Figure 26, Figure 28, and Figure 29 were obtained by actual measurement with the DIDO prototype, the following figures are obtained by simulation. The following BER results assume no FEC.

Fig. 30 shows the average BER performance of different DIDO precoding schemes in independent identically distributed channels. Curves marked "no choice" refer to the case where BD was used. In the same figure, the performance of antenna selection (ASel) is shown for different numbers of extra antennas (for different numbers of users). It can be seen that ASel provides better diversity gain (characterized by the slope of the BER curve in the high SNR region) as the number of additional antennas grows, resulting in better coverage. For example, if we fix the target BER to 10-2 (a practical value for an uncoded system), the SNR gain provided by ASel grows with the number of antennas.

Figure 31 shows the SNR gain of ASel as a function of the number of additional transmit antennas in the IID channel for different target BERs. It can be seen that by adding only 1 or 2 antennas, ASel produces a huge SNR gain compared to BD. In the following sections we will evaluate the performance of ASel by fixing the target BER to 10 ⁻² (for an uncoded system) only for the case of 1 or 2 extra antennas.

Figure 32 shows the SNR threshold as a function of the number of users (M) for BD and ASel with 1 and 2 extra antennas in i.d. channel. We observe that the SNR threshold increases with M due to the larger receive SNR requirement for a larger number of users. Note that we assume a fixed total transmit power (with varying numbers of transmit antennas) for any number of users. Furthermore, Figure 32 shows that the gain due to antenna selection is constant for any number of users in an i.d. channel.

Below, we show the performance of the DIDO system in spatially correlated channels. We simulated by the COST-259 spatial channel model described in "Channel models for link and system level simulations" published by X.Zhuang, FWVook, KLBaum, TAThomas and M.Cudak in September 2004 on IEEE 802.16Broadband Wireless Access Working Group channel for each user. We generate a single group for each user. As a case study, we assume an NLOS channel with a uniform linear array (ULA) at the transmitter with an element spacing of 0.5λ. For the case of a 2-user system, we simulate the swarm with the mean angles of arrival AOA1 and AOA2 for the first and second user, respectively. AOA is measured relative to the lateral orientation of the ULA. When there are more than two users in the system, we generate groups of users with uniformly spaced average AOA in the range [-φ _m , φ _m ], where we define

K is the number of users, and △φ is the angular distance between users' average AOA. Note that the angular range [ _-φm , _φm ] is centered at 0°, corresponding to the ULA's side-fire direction. In the following, we investigate the BER performance of the DIDO system as a function of the channel angular distribution (AS) and the angular separation between users using BD and ASel transmission schemes and different numbers of users.

Figure 33 shows the BER relative to the average SNR per user for two users with different AS values located in the same angular direction (ie AOA1 = AOA2 = 0° with respect to the side-fire direction of the ULA). It can be seen that as the AS increases, the BER performance improves and approaches the IID situation. In fact, a higher AS statistically yields less coverage between the two users' eigenpatterns and better performance of the BD precoder.

Figure 34 shows similar results to Figure 33, but with a higher angular distance between users. We consider AOA1=0°, AOA2=90° (that is, 90° angular distance). The best performance is achieved with low AS. In fact, for the case of high angular separation, there is less overlap between the user's characteristic patterns when the angular separation is low. Interestingly, we observe better BER performance in low AS than IID channels for the same reasons just mentioned.

Next, we calculate the SNR threshold for a target BER of 10 ⁻² in different correlation scenarios. Figure 35 plots the SNR threshold as a function of AS for different values of the average AOA of a user. For low user angular separations, reliable transmission with reasonable SNR requirements (ie 18dB) is only possible for channels characterized by high AS. On the other hand, when users are spatially separated, a smaller SNR is required to meet the same target BER.

Fig. 36 shows the SNR thresholds for the case of 5 users. Generate user average AOA with different values of angular distance Δφ according to the definition in (13). We observe that for △φ = 0° and AS < 15°, due to the small angular separation between users, BD performance is poor and the target BER is not met. For increasing AS, the SNR requirement to meet a fixed target BER decreases. On the other hand, for Δφ = 30°, the minimum SNR requirement is obtained at low AS, consistent with the results in Fig. 35. As AS increases, the SNR threshold saturates to one of the i.d. channels. Note that Δφ = 30° with 5 users corresponds to an AOA range of [−60°,60°], which is typical for a base station in a cellular system with 120° sectored cells.

Next, we investigate the performance of the ASel transmission scheme in spatially correlated channels. Figure 37 compares the SNR thresholds of BD and ASel with 1 and 2 extra antennas for two user cases. We consider two different cases of angular distance between users: {AOA1=0°, AOA2=0°} and {AOA1=0°, AOA2=90°}. The curves for the BD scheme (ie no antenna selection) are the same as in FIG. 35 . We observe that ASel yields SNR gains of 8dB and 10dB with 1 and 2 extra antennas for high AS, respectively. As AS decreases, the gain due to ASel on BD becomes smaller due to the reduced number of degrees of freedom in the MIMO broadcast channel. Interestingly, for AS=0° (ie close to the LOS channel) and the case {AOA1=0°, AOA2=90°}, ASel does not provide any gain due to differences in the spatial domain. Figure 38 shows similar results to Figure 37, but for the case of 5 users.

We calculated the SNR threshold as a function of the number of users (M) in the system (assuming a general target BER of 10 ⁻² ) for the BD and ASel transmission schemes. The SNR threshold corresponds to the average SNR such that the total transmit power is constant for any M. We assume the maximum separation between the mean AOAs of each user group within the azimuth range [-φ _m ,φ _m ] = [-60°,60°]. Then, the angular distance between users is Δφ=120°/(M-1).

Figure 39 shows the SNR thresholds for BD schemes with different AS values. We observe that the lowest SNR requirement is obtained for AS=0.1° (ie low angular spread) with a relatively small number of users (ie K≤20) due to the large angular separation between users. However, for M>50, the SNR requirement is much greater than 40dB because Δφ is very small and BD cannot be implemented. Furthermore, for AS > 10°, the SNR threshold remains almost constant for any M, and the DIDO system in spatially correlated channels approaches the performance of independent and identically distributed channels.

To reduce the value of the SNR threshold and improve the performance of the DIDO system, we apply the ASel transmission scheme. Fig. 40 shows the SNR thresholds in a spatially correlated channel with AS = 0.1° for BD and ASel with 1 and 2 additional antennas. For reference, we also report the curves for the IID case shown in Figure 32. It can be seen that for fewer users (i.e. M≤10), antenna selection does not help reduce SNR requirements due to the lack of diversity in the DIDO broadcast channel. As the number of users increases, ASel benefits from multi-user diversity, resulting in an SNR gain (ie 4dB for M=20). Furthermore, for M≤20, the performance of ASel with 1 or 2 extra antennas in highly spatially correlated channels is the same.

We then calculate the SNR thresholds for two additional channel situations: AS = 5° in Fig. 41 and AS = 10° in Fig. 42 . Fig. 41, compared with Fig. 40, shows that due to the larger angular spread, ASel yields SNR gains also for a relatively small number of users (ie M≤10). As reported in Fig. 42, for AS = 10°, the SNR threshold decreases further, as the gain of ASel becomes higher.

Finally, we summarize the results presented so far for correlated channels. Figures 43 and 44 show the SNR threshold as a function of the number of users (M) and angular spread (AS) for the BD and ASel schemes with 1 and 2 additional antennas, respectively. Note that the case of AS = 30° actually corresponds to i.i.d. channels, and we use this value of AS in the figure for graphical representation. We observe that while BD is affected by channel spatial correlation, ASel yields almost the same performance for any AS. Furthermore, for AS = 0.1°, ASel performs similarly to BD for low M and outperforms BD for large M (ie, M ≥ 20) due to multiuser diversity.

Figure 49 compares the performance of different DIDO schemes in terms of SNR threshold. The DIDO schemes considered are: BD, ASel, BD with Eigenmode Selection (BD-ESel), and Maximum Ratio Combining (MRC). Note that MRC does not pre-cancel interference at the transmitter (unlike other methods), but provides greater gain in cases where users are spatially separated. In Fig. 49 we plot the SNR threshold for a DIDO N×2 system for a target BER = 10 ⁻² when two users are located at -30° and 30° from the side-fire direction of the transmit array, respectively. We observe that for low AS, the MRC scheme provides a 3dB gain compared to other schemes because the spatial channels of users are well separated and the influence of inter-user interference is small. Note that the gain of MRC over DIDO N×2 is due to array gain. For AS larger than 20°, the QR-ASel scheme outperforms the others and yields about 10 dB gain compared to BD 2x2 without the option. QR-ASel and BD-ESel provide about the same performance for arbitrary values of AS.

Described above is a new adaptive transmission technique for DIDO systems. The method dynamically switches between DIDO transmission modes to different users to enhance throughput for a fixed target bit error rate. The performance of different levels of DIDO systems is measured under different propagation conditions, and it is observed that a huge gain in throughput can be achieved by dynamically selecting the DIDO mode and number of users as a function of the propagation conditions.

III. Precompensation of frequency and phase difference

a. Background

As mentioned earlier, wireless communication systems use carrier waves to transfer information. These carriers are typically sinusoidal waves whose amplitude and/or phase are modulated in response to the information being transmitted. The nominal frequency of the sine wave is known as the carrier frequency. To create this waveform, the transmitter synthesizes one or two sine waves and uses up-conversion to create a modulated signal superimposed on the sine wave with the specified carrier frequency. This can be achieved by direct conversion, where the signal is modulated directly on a carrier or through multiple up-conversion stages. To process this waveform, the receiver must demodulate the received RF signal and effectively remove the modulating carrier. This requires the receiver to synthesize one or more sinusoidal signals to invert the modulation process at the transmitter, known as down conversion. Unfortunately, the sine wave signals generated at the transmitter and receiver are obtained from different reference oscillators. No reference oscillator creates a perfect frequency reference; in fact, there is usually some deviation from the actual frequency.

In wireless communication systems, differences in the output of reference oscillators at the transmitter and receiver create a phenomenon at the receiver known as carrier frequency offset, or simply frequency offset. Essentially, after down conversion, there is some residual modulation (corresponding to the difference in the transmit and receive carrier) in the received signal. This creates distortion in the received signal, resulting in a higher bit error rate and lower throughput.

Different techniques exist for dealing with carrier frequency offset. Most methods estimate the carrier frequency offset at the receiver and then apply a carrier frequency offset correction algorithm. The carrier frequency offset estimation algorithm is blind using the following method: Offset QAM (T. Fusco and M. Tanda, "Blind Frequency-offset Estimation for OFDM/OQAM Systems," Signal Processing, IEEE Transactions on [see also Acoustics , Speech, and Signal Processing, IEEETransactions on], vol.55, pp.1828-1838, 2007); periodic characteristics (E.Serpedin, A.Chevreuil, G.B.Giannakis and P.Loubaton, "Blind channel and carrier frequency offset estimation using periodic modulation precoders, "SignalProcessing, IEEETransactions on [see also Acoustics, Speech, and Signal Processing, IEEETransactions on], vol.48, no.8, pp.2389-2405, Aug.2000); or Orthogonal Frequency Division Multiplexing ( Cyclic prefixes in OFDM) structural approaches (J.J. van de Beek, M. Sandell and P.O. Borjesson, "MLestimation of time and frequency offset in OFDM systems," Signal Processing, IEEE Transactions on [see also Acoustics, Speech, and Signal Processing, IEEETransactions on], vol.45, no.7, pp.1800-1805, July 1997; U.Tureli, H.Liu and M.D.Zoltowski, "OFDM blind carrier offset estimation: ESPRIT," IEEETrans.Commun., vol.48, no.9, pp.1459-1461, Sept.2000; M.Luise, M.Marselli and R.Reggiannini, "Low-complexity blind carrier frequency recovery for OFDM signals o ver frequency-selective radio channels," IEEE Trans. Commun., vol. 50, no. 7, pp. 1182-1188, July 2002).

Alternatively, dedicated training signals can be utilized, including repeated data symbols (P.H. Moose, "A technique for orthogonal frequency division multiplexing frequency offset correction," IEEE Trans. Commun., vol.42, no.10, pp.2908-2914, Oct.1994); two different symbols (T.M.Schmidl and D.C.Cox, "Robust frequency and timing synchronization for OFDM," IEEE Trans.Commun., vol.45, no.12, pp.1613-1621, Dec.1997); Or a known symbol sequence inserted periodically (M.Luise and R.Reggiannini, "Carrier frequency acquisition and tracking for OFDM systems," IEEE Trans. Commun., vol.44, no.11, pp.1590-1598, Nov. .1996). Correction can occur analog or digitally. The receiver can also use the carrier frequency offset estimate to pre-correct the transmitted signal to remove the offset. Due to the sensitivity of multi-carrier and OFDM systems to frequency offset, carrier frequency offset correction has been extensively studied for multi-carrier and OFDM systems (J.J.van deBeek, M.Sandell and P.O.Borjesson, "ML estimation of time and frequency offset in OFDM systems , "Signal Processing, IEEE Transactions on [see also Acoustics, Speech, and Signal Processing, IEEE Transactions on], vol.45, no.7, pp.1800-1805, July 1997; U.Tureli, H.Liu and M.D. Zoltowski, "OFDM blindcarrier offset estimation: ESPRIT," IEEE Trans.Commun., vol.48, no.9, pp.1459-1461, Sept.2000; T.M.Schmidl and D.C.Cox, "Robust frequency and timing synchronization for OFDM," IEEETrans.Commun.,vol.45,no.12,pp.1613-1621,Dec.1997;M.Luise,M.Marselli and R.Reggiannini,"Low-complexity blind carrier frequency recovery for OFDM signals overfrequency-selective radio channels ,"IEEE Trans.Commun.,vol.50,no.7,pp.1182-1188,July 2002).

Frequency offset estimation and correction is an important problem for multi-antenna communication systems or more generally MIMO (Multiple Input Multiple Output) systems. In a MIMO system, the transmit antenna is locked to one frequency reference and the receiver is locked to another frequency reference, with a single offset between the transmitter and receiver. Several algorithms were proposed to deal with this problem using training signals (K.Lee and J.Chun, "Frequency-offset estimation for MIMO and OFDM systems using orthogonal training sequences," IEEE Trans.Veh.Technol.,vol.56,no.1 , pp.146-156, Jan.2007; M.Ghogho and A.Swami, "Training design for multipath channel and frequency offset estimation in MIMO systems," Signal Processing, IEEETransactions on [see also Acoustics, Speech, and Signal Processing, IEEETransactions on ], vol.54, no.10, pp.3957-3965, Oct.2006; and adaptive tracking C. Oberli and B. Daneshrad, "Maximum likelihood tracking algorithms for MIMOOFDM," in Communications, 2004 IEEE International Conference on, vol.4, June 20 -24, 2004, pp. 2468-2472). A more significant problem is encountered in MIMO systems, where the transmit antennas are not locked to the same frequency reference, but the receive antennas are locked together. This actually happens in the uplink of Space Division Multiple Access (SDMA) systems, which are considered as MIMO systems, where different users correspond to different transmit antennas. In this case, the compensation of the frequency offset is more complicated. In particular, the frequency offset creates interference in the different transmitted MIMO streams. Correction can be performed using complex joint estimation and equalization algorithms (A.Kannan, T.P.Krauss and M.D.Zoltowski, "Separation of cochannel signals under imperfecttiming and carriersynchronization," IEEE Trans.Veh.Technol., vol.50, no.1, pp.79-96, Jan.2001), and equalization after frequency offset estimation (T.Tang and R.W.Heath,"Joint frequency offset estimation and interference cancellation for MIMO-OFDM systems[mobile radio],"2004.VTC2004-Fall .2004IEEE 60th Vehicular Technology Conference, vol.3, pp.1553-1557, Sept.26-29, 2004; X.Dai, "Carrier frequency offset estimation for OFDM/SDMA systems using consecutive pilots," IEEE Proceedings-Communications, vol.152, pp.624-632, Oct.7, 2005). Some works have dealt with the related problem of residual phase offset and tracking error, where the residual phase offset is estimated and compensated after frequency offset estimation, but this work only considers the uplink of SDMAOFDMA system (L Haring, S.Bieder and A. Czylwik, "Residual carrier and sampling frequency synchronization in multiuser OFDM systems," 2006. VTC 2006-Spring. IEEE 63rd Vehicular Technology Conference, vol. 4, pp. 1937-1941, 2006). The worst case occurs in MIMO systems when all transmit and receive antennas have different frequency references. The only available work on this topic deals only with the asymptotic analysis of estimation errors in flat-fading channels (O. Besson and P. Stoica, "Onparameter estimation of MIMO flat-fading channels with frequency offsets," Signal Processing, IEEE Transactions on [See also Acoustics, Speech, and Signal Processing, IEEE Transactions on], vol.51, no.3, pp.602-613, Mar.2003).

An already well-studied situation occurs when the different transmit antennas of a MIMO system do not have the same frequency reference, and the receive antennas process signals independently. This occurs in what is known as a Distributed Input Output (DIDO) communication system (also known in the literature as a MIMO broadcast channel). The DIDO system consists of an access point with distributed antennas that transmit parallel data streams (via precoding) to multiple users to enhance downlink throughput, while using the same radio resources (i.e., the same time slot duration and frequency band) as a conventional SISO system. A detailed description of the DIDO system is presented in US Patent Application 20060023803, filed July 2004, by S.G. Perlman and T. Cotter, entitled "System and method for distributed input-distributed output wireless communications". There are many ways of implementing the DIDO precoder. One solution is block diagonalization (BD), described for example in: Q.H.Spencer, A.L.Swindlehurst and M.Haardt, "Zero-forcing methods for downlink spatial multiplexing inmultiuser MIMO channels," IEEE Trans.Sig.Proc, vol .52, pp.461-471, Feb.2004; K.K.Wong, R.D.Murch and K.B.Letaief, "A joint-channel diagonalization for multiuser MIMO antenna systems," IEEETrans.WirelessComm., vol.2, pp.773-786, JuI 2003; L.U.Choi and R.D.Murch, "Atransmitpreprocessing technique for multiuser MIMO systems using adecomposition approach," IEEE Trans.Wireless Comm., vol.3, pp.20-24, Jan 2004; Z.Shen, J.G.Andrews, R.W.Heath and B.L Evans, "Low complexity userselection algorithms for multiuser MIMO systems with block diagonalization," accepted for publication in IEEE Trans.Sig.Proc, Sep.2005; Z.Shen, R.Chen, J.G.Andrews, R.W.Heath and B.L Evans, "Sum capacity of multiuser MIMO broadcast channels with block diagonalization," submitted to IEEE Trans.Wireless Comm., Oct. 2005; R. Chen, R.W. Heath and J.G. Andrews," Transmit selection diversity for unitary precoded multiuserspatial multiplexing systems with linear receivers," accepted to IEEE Trans, on Signal Pr ocessing, 2005.

In DIDO systems, transmit precoding is used to separate data streams for different users. Carrier frequency offset causes several problems related to system implementation when transmit antenna RF chains do not share the same frequency reference. When this happens, each antenna effectively transmits at a slightly different carrier frequency. This breaks the integrity of the DIDO precoder, causing each user to experience additional interference. Proposed below are several solutions to this problem. In one embodiment of the solution, DIDO transmit antennas share a frequency reference through a wired, optical or wireless network. In another implementation of the solution, one or more users estimate the frequency offset difference (relative difference in offset between antenna pairs) and send this information back to the transmitter. The transmitter then pre-corrects the frequency offset and proceeds with training for DIDO and precoder estimated phase. This implementation has problems when there is a delay in the feedback channel. The reason is that there may be residual phase errors created by the correction process, which does not take into account the subsequent channel estimation. To solve this problem, an additional embodiment solves this problem by estimating the delay using a new frequency offset and phase estimator. Results are presented based on simulations and actual measurements performed by a DIDO-OFDM prototype.

The frequency and phase offset compensation method proposed in this document may be sensitive to estimation errors due to noise at the receiver. Therefore, another embodiment proposes a method for time and frequency offset estimation that is also robust under low SNR conditions.

There are different methods for performing time and frequency offset estimation. Many of these methods have been proposed specifically for OFDM waveforms due to their sensitivity to synchronization errors.

These algorithms do not typically use the structure of OFDM waveforms and thus are generally adequate for single-carrier and multi-carrier waveforms. The algorithms described below are among a class of techniques that use known reference symbols (eg, training data) to assist in synchronization. Many methods are extensions of Moose's frequency offset estimator (see P.H.Moose, "A technique for orthogonal frequency division multiplexing frequency offset correction," IEEE Trans. Commun., vol.42, no.10, pp.2908-2914, Oct.1994 ). Moose proposes to use two repeated training signals and use the phase difference between the received signals to obtain the frequency offset. Moose's method is only able to correct for fractional frequency offsets. An extension of Moose's method was proposed by Schmidl and Cox (T.M.Schmidl and D.C.Cox, "Robust frequency and timing synchronization for OFDM," IEEE Trans. Commun., vol.45, no.12, pp.1613-1621, Dec.1997 ). Their main innovation lies in the use of a periodic OFDM symbol and additional differentially encoded training symbols. Integer offset correction implemented by differential encoding in the second symbol. Coulson considered a similar setup described in T.M.Schmidl and D.C.Cox, "Robust frequency and timing synchronization for OFDM," IEEE Trans.Commun., vol.45, no.12, pp.1613-1621, Dec.1997, and in A.J.Coulson,"Maximum likelihoodsynchronization for OFDM using a pilot symbol:analysis,"IEEE J.Select.AreasCommun.,vol.19,no.12,pp.2495-2503,Dec.2001 and A.J.Coulson,"Maximum likelihoodsynchronization for OFDM using a A detailed discussion of the algorithms and analysis is provided in pilot symbol: algorithms," IEEE J. Select. Areas Commun., vol. 19, no. 12, pp. 2486-2494, Dec. 2001. A major difference is that Coulson uses repeated maximal length sequences to provide good correlation properties. He also recommends using chirp signals because of their constant envelope properties in both the time and frequency domains. Coulson considered practical details but did not include integer estimates. Multiple repeated training signals were used by Minn et.al. in H.Minn, V.K.Bhargava and K.B.Letaief, "A robust timing and frequency synchronization for OFDM systems," IEEETrans.Wireless Commun.,vol.2,no.4,pp. 822-839, considered in July 2003, but the training structure was not optimized. Shi and Serpedin proposed some optimality of the training structure with the idea of forming frame synchronization (K. Shi and E. Serpedin, "Coarse frame and carrier synchronization of OFDM systems: a new metric and comparison," IEEE Trans.Wireless Commun., vol .3, no.4, pp.1271-1284, July 2004). One embodiment of the present invention uses the method of Shi and Serpedin to perform frame synchronization and fractional frequency offset estimation.

Many methods in the literature focus on frame synchronization and fractional frequency offset correction. Integer offset correction using additional training symbols in T.M.Schmidl and D.C.Cox, "Robust frequency and timing synchronization for OFDM," IEEE Trans.Commun., vol.45, no.12, pp.1613-1621, Dec.1997 to be resolved. For example, Morrelli et al. obtained in M.Morelli, A.N.D'Andrea and U.Mengali, "Frequency ambiguityresolution in OFDM systems," IEEE Commun. Lett., vol.4, no.4, pp.134-136, Apr.2000 An improved version of T.M.Schmidl and D.C.Cox, "Robust frequency and timing synchronization for OFDM," IEEE Trans.Commun., vol.45, no.12, pp.1613-1621, Dec.1997. An alternative method using a different preamble structure was proposed by Morelli and Mengali (M. Morelli and U. Mengali, "An improved frequency offset estimator for OFDM applications," IEEE Commun. Lett., vol. 3, no. 3, pp. 75-77, Mar. 1999). This method uses the correlation between M repetitions of the same training symbols to increase the range of the fractional frequency offset estimator by a factor of M. This is the best linear unbiased estimator and accepts the largest skew (with a suitable design), but does not provide good timing synchronization.

System specification

One embodiment of the present invention uses precoding based on channel condition information to eliminate frequency and phase offsets in DIDO systems. See Figure 11 and the related description above for the description of this embodiment.

In one embodiment of the invention, each user uses a receiver equipped with a frequency offset estimator/compensator. As shown in FIG. 45, in one embodiment of the invention, a system including a receiver includes a plurality of RF units 4508, a corresponding plurality of A/D units 4510, equipped with a frequency offset estimator/compensator 4512 receiver and DIDO feedback generator unit 4506.

The RF unit 4508 receives a signal transmitted from the DIDO transmitter unit, down-converts the signal to baseband, and supplies the down-converted signal to the A/D unit 4510 . The A/D unit 4510 then converts the signal from analog to digital and sends it to the frequency offset estimator/compensator unit 4512 . The frequency offset estimator/compensator unit 4512 estimates the frequency offset and compensates for the frequency offset as described herein, and then sends the compensated signal to the OFDM unit 4513. OFDM unit 4513 removes the cyclic prefix and runs a Fast Fourier Transform (FFT) to report the signal to the frequency domain. During training, OFDM unit 4513 sends output to channel estimation unit 4504 to compute channel estimates in the frequency domain. Alternatively, the channel estimate can be computed in the time domain. During the data period, the OFDM unit 4513 sends the output to the DIDO receiver unit 4502 which demodulates/decodes the signal to obtain the data. Channel estimation unit 4504 sends the channel estimates to DIDO feedback generator unit 4506, which may quantize the channel estimates and send them back to the transmitter via a feedback control channel, as shown.

Description of one embodiment of the algorithm for the DIDO 2x2 scenario

Described below is an embodiment of an algorithm for frequency/phase offset compensation in a DIDO system. DIDO system models are initially described with and without frequency/phase offsets. For simplicity, a specific implementation of the DIDO 2×2 system is provided. However, the basic principles of the invention can also be implemented in higher order DIDO systems.

DIDO system model with/without frequency and phase offset

The received signal of DIDO 2x2 can be written for the first user as:

r ₁ [t]=h ₁₁ (w ₁₁ x ₁ [t]+w ₂₁ x ₂ [t])+h ₁₂ (w ₁₂ x ₁ [t]+w ₂₂ x ₂ [t]) (1)

and for the second user writes:

r ₂ [t]h ₂₁ (w ₁₁ x ₁ [t]+w ₂₁ x ₂ [t])+h ₂₂ (w ₁₂ x ₁ [t]+w ₂₂ x ₂ [t]) (2)

where t is the discrete time index, h _mn and w _mn are the channel and DIDO precoding weights between the mth user and the nth transmit antenna, respectively, and x _m is the transmitted signal for user m. Note that h _mn and w _mn are not functions of t because we assume the channel is constant over the period between training and data sending.

In the presence of frequency and phase offsets, the received signal is expressed as

as well as

where T _s is the symbol period; for the n-th transmit antenna, ωT _n =2∏f _Tn ; for the m-th user, W _Um =2∏F _Um ; and f _Tn and f _Um are the Antenna and actual carrier frequency of the mth user (affected by offset). The value t _mn represents a random delay causing a phase shift on channel h _mn . Figure 46 depicts the DIDO 2x2 system model.

For time, we use the following definitions:

Δω _mn =ω _Um -ω _Tn (5)

is used to denote the frequency offset between the mth user and the nth transmit antenna.

Description of an embodiment of the invention

A method according to one embodiment of the invention is illustrated in FIG. 47 . The method comprises the following general steps (including sub-steps, as shown): training period 4701 for frequency offset estimation; training period 4702 for channel estimation; data transmission 4703 via DIDO precoding with compensation. These steps are described in detail below.

(a) Training period for frequency offset estimation (4701)

During a first training period, the base station sends one or more training signals from each transmit antenna to one of the users (4701a). As described herein, a "user" is a wireless client device. For the DIDO 2×2 case, the signal received by the mth user is given by:

where p1 and _p2 are the training sequences sent from the _first and second antennas, respectively.

The mth user can use any type of frequency offset estimator (ie by convolution of the training sequence) and estimate the offsets Δω _ml and Δω _m2 . Then, based on these values, the user calculates the frequency offset between the two transmit antennas:

Δω _T ＝ Δω _m2 - Δω _m1 ＝ ω _T1 - ω _T2 (7)

Finally, the value in (7) is fed back to the base station (4701b).

Note that p ₁ and p ₂ in (6) are designed to be orthogonal so that the user can estimate Δω _ml and Δω _m2 . Alternatively, in one embodiment, the same training sequence is used in two consecutive time slots from which the user estimates the offset. Furthermore, to improve the estimation of the offset in (7), the same calculation described above can be done for all users of the DIDO system (not only for the m-th user), and the final estimate can be obtained from all users The (weighted) average of the values. However, this solution requires more computation time and amount of feedback. Finally, an update of the frequency offset estimate is only required if the frequency offset changes over time. Thus, depending on the stability of the clock at the transmitter, step 4701 of the algorithm may be performed on a long-term basis (ie for each data transmission) such that the above-mentioned feedback is reduced.

(b) Training period for channel estimation (4702)

During the second training period, the base station first obtains a frequency offset feedback with the value in (7) from the mth user or from multiple users. The value in (7) is used to precompensate the frequency offset at the transmitter. The base station then sends training data to all users for channel estimation (4702a).

For a DIDO 2×2 system, the signal received at the first user is given by:

and at the second user:

in, and Δt is the random or known delay between the base station's first transmission and the second transmission. Furthermore, p1 and _p2 are the training sequences transmitted from the _first and second antennas for the user frequency offset and channel estimation, respectively.

Note that precompensation is only applied to the second antenna in this embodiment.

Expanding (8), we get

Similar for the second user:

in,

At the receiving end, the user compensates the remaining frequency offset by using the training sequences _p1 and _p2 . Then, the user estimates by training the vector channel (4702b):

These channels or Channel Situation Information (CSI) in (12) are fed back to the base station (4702b), which calculates the DIDO precoder as described in the following sections.

(c) DIDO precoding with precompensation (4703)

The base station receives the channel state information (CSI) in (12) from the user and calculates the precoding weight (4703a) through block diagonalization (BD), so that

where the vector h ₁ is defined in (12), and w _m =[w _m1 ,w _m2 ]. Note that the invention proposed in this disclosure can be used in any other DIDO precoding method than BD. The base station also precompensates for frequency offset by using the estimate in (7), and precompensates for phase offset by estimating the delay (Δt ₀ ) between the second training transmission and the current transmission (4703a). Finally, the base station sends the data to the user via the DIDO precoder (4703b).

After the sending process, the signal received at user 1 is given by:

in, Using property (13), we obtain

Similarly for user 2 we get:

expand(16):

in,

Finally, the user calculates the residual frequency offset and channel estimate to demodulate the data streams x ₁ [t] and x ₂ [t] (4703c).

Generalization to DIDO N×M

In this section, the previously described techniques are generalized to a DIDO system with N transmit antennas and M users.

i. Training Period for User Frequency Offset Estimation

During the first training period, the signal received by the mth user due to the training sequence sent from the N antennas is given by:

where p _n is the training sequence sent from the nth antenna.

After estimating the offset Δω _mn , The mth user calculates the frequency offset between the first and nth transmit antennas:

Δω _T,1n = Δω _mn - Δω _m1 = ω _T1 - ω _Tn (19)

Finally, the value in (19) is fed back to the base station.

ii. Training period for channel estimation

During the second training period, the base station first obtains frequency offset feedback with the value in (19) from the mth user or from multiple users. The value in (19) is used to precompensate the frequency offset at the transmitter. Then, the base station sends training data to all users for channel estimation.

For a DIDO N×M system, the received signal at the mth user is given by:

in, And Δt is the random or known delay between the base station's first and second transmissions. Also, P _n is a training sequence transmitted from the nth antenna for frequency offset and channel estimation.

On the receiving side, the user compensates for the remaining frequency offset by using the training sequence _Pn . Then, each user m is estimated by training the vector channel:

and fed back to the base station, which computes the DIDO precoder as described in the following sections.

iii. DIDO precoding with precompensation

The base station receives the channel condition information (CSI) in (12) from the user and calculates the precoding weights by block diagonalization (BD), so that

where the vector h _m is defined in (21), and w _m =[w _m1 ,w _m2 ,...,w _mN ]. The base station also precompensates for the frequency offset by using the estimate in (19), and for the phase offset by estimating the delay (Δt ₀ ) between the second training transmission and the current transmission. Finally, the base station sends the data to the users via the DIDO precoder.

After the sending process, the signal received at user i is given by:

in, Using property (22), we get:

Finally, the user computes the residual frequency offset and channel estimate to demodulate the data stream _xi [t].

result

Figure 48 shows the SER results for the DIDO 2x2 system with and without frequency offset. It can be seen that the proposed method completely removes the frequency/phase offset, yielding the same SER as the system without offset.

Next, we evaluate the sensitivity of the proposed compensation method to fluctuations in frequency offset errors and/or real-time offsets. Therefore, we rewrite (14) as:

where ε represents the estimation error and/or change in frequency offset between training and data transmission. Note that the effect of ε is to destroy the orthogonality property in (13), such that the interference terms in (14) and (16) are not completely pre-cancelled at the transmitter. Because of this, the SER performance degrades with increasing ε value.

Fig. 48 shows the SER performance of the frequency offset compensation method for different values of ε. These results assume T _s =0.3ms (ie a signal with 3KHz bandwidth). We observe that for ε = 0.001 Hz (or less), the SER performance is similar to the case without offset.

f. Description of one embodiment of an algorithm for time and frequency offset estimation

Below, we describe an additional embodiment (4701b in Figure 47) that performs time and frequency offset estimation. The transmit signal structure considered is proposed in H.Minn, VKBhargava and KBLetaief, "A robust timing and frequency synchronization for OFDM systems," IEEETrans.Wireless Commun., vol.2, no.4, pp.822-839, July 2003, It is detailed in K.Shi and E.Serpedin, "Coarse frame and carriersynchronization of OFDM systems: a new metric and comparison," IEEE Trans. Wireless Commun., vol.3, no.4, pp.1271-1284, July 2004 Research. Usually sequences with good correlation properties are used for training. For example, for our system, the Chu sequence is used, as described in D. Chu, "Polyphasecodes with good periodic correlation properties (corresp.)," IEEE Trans. Inform. Theory, vol.18, no.4, pp. 531-532, described in July 1972. These sequences have the interesting property that they are perfectly circularly correlated. Let _Lcp denote the length of the cyclic prefix and _Nt denote the length of the component training sequence. Such that N _t =M _t , where M _t is the length of the training sequence. Under these assumptions, the sequence of symbols sent for head start can be written as:

s[n]=t[nN _t ] for n=-1,...,-L _cp

s[n]=t[n] for n=0,...,N _t -1

s[n]=t[nN _t ] for n=N _t ,…,2N _t -1

s[n]=-t[n-2N _t ] for n=2N _t ,…,3N _t -1

s[n]=t[n-3N _t ] for n=3N _t ,…,4N _t -1

Note that the structure of the training signal can be extended to other lengths, but the block structure is repeated. For example, to use 16 training signals, we consider a structure such as:

[CP,B,B,-B,B,B,B,-B,B,-B,-B,B,-B,B,B,-B,B,].

By using this structure, and making N _t =4M _t , all algorithms to be described can be used without modification. Effectively, we repeat the training sequence. This is especially useful in situations where suitable training signals may not be available.

After filtering and downsampling to match the symbol rate, consider the following received signal:

where ε is the unknown discrete-time frequency offset, Δ is the unknown frame offset, h[l] is the unknown discrete-time channel coefficient, and v[n] is the additive noise. To explain the key ideas in the following sections, the presence of additional noise is ignored.

i. Coarse frame synchronization

The purpose of coarse frame synchronization is to account for the unknown frame offset Δ. Let's make the following definitions:

The proposed coarse frame synchronization algorithm is from K.Shi and E.Serpedin, "Coarse frame and carrier synchronization of OFDM systems: a new metric and comparison," IEEETrans.Wireless Commun.,vol.3,no.4,pp.1271- 1284, inspired by the algorithm in July 2004, obtained according to the maximum likelihood criterion.

Approach 1 - Improved Coarse Frame Synchronization: The coarse frame synchronization estimator addresses the following optimizations:

in,

Such that the corrected signal is defined as:

Additional correction terms are used to compensate for small initial pulses in the channel and can be adjusted based on the application. This extra delay will then be included in the channel.

ii. Fractional Frequency Offset Correction

Fractional frequency offset correction follows the coarse frame sync block.

Method 2 - Improved Fractional Frequency Offset Correction: Fractional Frequency Offset is the solution below:

This is known as a fractional frequency offset because the algorithm can only correct for the offset

This issue will be addressed in the next section. Let the fine frequency offset correction signal be defined as:

Note that methods 1 and 2 are for K.Shi, E.Serpedin, who work better in frequency selective channels, "Coarseframe and carrier synchronization of OFDM systems: a newmetric and comparison," IEEE Trans.Wireless Commun., vol.3, no.4, pp.1271-1284, improvement of July 2004. A particular innovation here is the use of the r and The use of σ improves previous estimators because it ignores samples that are affected due to inter-symbol interference.

iii. Integer Frequency Offset Correction

In order to correct integer frequency offsets, it is necessary to write an equivalent system model for the received signal after fine frequency offset correction. Absorbing the retained timing error into the channel, the received signal without noise has the following structure:

where n=0,1,...,4N _t –1. The integer frequency offset is k and the unknown equivalent channel is g[l].

Method 3 - Improved Integer Frequency Offset Correction: The integer frequency offset is the solution of:

in:

r=D[k]Sg

This gives an estimate of the total frequency offset:

In fact, method 3 has high complexity. To reduce complexity, the following observations can be made. First, the product S(S*S ) ^-1 S can be precomputed. Unfortunately, this still leaves a rather large matrix multiplication. An alternative is to use observations with the proposed training sequence, S*S≈I. This results in the following reduced loading method.

Method 4 - Improved Integer Frequency Offset Correction with Low Complexity:

A low-complexity integer frequency offset estimator solves the

iv. Results

In this section, we compare the performance of different proposed estimators.

First, in Figure 50, we compare the amount of overhead required by each approach. Note that two new methods reduce the overhead by a factor of 10 to 20. To compare the performance of different estimators, MonteCarlo experiments were performed. The considered setup is our usual NVIS transmit waveform constructed from linear modulation with a symbol rate of 3K symbols per second, corresponding to a passband bandwidth of 3KHz, and raised cosine pulse shaping. For each Monte Carlo implementation, frequency offsets are generated from a uniform distribution on [-f _max , f _max ].

A simulation with a small frequency offset of f _max =2 Hz and no integer offset correction is shown in FIG. 51 . It can be seen from this performance comparison that the performance with N _t /M _t =1 is slightly degraded from the original estimator, although the overhead is substantially reduced. The performance with N _t /M _t =4 is better, almost 10 dB. All curves experience meandering at low SNR points due to errors in integer offset estimation. Small errors in integer offsets can create large frequency errors and large splice squared errors. Integer offset correction can be turned off in small offsets to improve performance.

In the presence of multipath channels, the performance of the frequency offset estimator generally degrades. However, in Figure 52, turning off the integer offset estimator exhibits very good performance. Therefore, in multipath channels, an improved fine correction algorithm after performing a robust coarse correction is more important. Note that the offset performance with N _t /M _t =4 is much better in the multipath case.

Embodiments of the present invention may include the various steps set forth above. The steps may be implemented in the form of machine-executable instructions that cause a general or special purpose processor to perform certain steps. For example, the various components within the base station/AP and client devices described above may be implemented as software executing on a general-purpose or special-purpose processor. In order to avoid obscuring relevant aspects of the present invention, various well-known personal computer components, such as computer memory, hard disk, input devices, etc. have been omitted from the figures.

Alternatively, in one embodiment, the various functional blocks and associated steps shown here may be implemented by dedicated hardware components (such as application-specific integrated circuits (ASICs)) that contain hard-wired logic to perform the steps, or by programmed computer components and Any combination of custom hardware components can be implemented.

In one embodiment, specific blocks such as the encoding, modulation, and signal processing logic 903 described above may be implemented on a programmable digital signal processor (DSP) (or set of DSPs) using, for example, a Texas Instruments ) DSP of the TMS320x architecture (for example, TMS320C6000, TMS320C5000, etc.). The DSP in this embodiment may be embedded in a personal computer add-in card, eg, a PCI card. Of course, a variety of different DSP architectures could be used while consistent with the underlying principles of the invention.

Various components of the present invention may also be provided as machine-readable media for storing machine-executable instructions. A machine-readable medium may include, but is not limited to, flash memory, optical disks, CD-ROMs, DVD ROMs, RAM, EPROMs, EEPROMs, magnetic or optical cards, transmission media, or other types of machine-readable media suitable for storing electronic instructions. For example, the present invention can be downloaded as a computer program that can be transmitted from a remote computer (such as a server) to the requesting computer via a communication link (such as a modem or network connection) by means of a data signal embodied in a carrier wave or other propagation medium. Computers (such as clients).

Throughout the foregoing description, for purposes of explanation, numerous specific details were set forth in order to provide a thorough understanding of the present systems and methods. It will be apparent, however, to one skilled in the art that the systems and methods may be practiced without some of these specific details. Accordingly, the scope and spirit of the present invention should be judged from the appended claims.

In addition, in the foregoing description, numerous documents were cited to provide a more comprehensive understanding of the invention. All such cited references are incorporated into this application by reference.

Claims (33)

1. A system for compensating for frequency and phase offsets for multiuser multiple antenna system (MU-MAS) communications, the system comprising:

one or more code modulation units for coding and modulating information bits for each of a plurality of wireless client devices to generate coded and modulated information bits;

one or more mapping units for mapping the coded and modulated information bits into complex symbols; and

an MU-MAS frequency/phase offset aware precoding unit to compute MU-MAS frequency/phase offset aware precoding weights using channel condition information, the MU-MAS frequency/phase offset aware precoding unit to precode complex symbols obtained from the mapping unit with the weights to pre-cancel frequency/phase offset and/or inter-user interference.

2. The system as in claim 1 wherein the channel condition information is obtained through feedback from the wireless client device to the MU-MAS frequency/phase offset aware precoding unit.

3. The system as in claim 1 wherein the channel condition information in the downlink is obtained from the uplink at the MU-MAS frequency/phase offset aware precoding unit by exploiting channel reciprocity.

4. The system of claim 1, further comprising: one or more Orthogonal Frequency Division Multiplexing (OFDM) units to receive the precoded signals from the MU-MAS frequency/phase offset aware precoding units and to modulate the precoded signals according to an OFDM standard.

5. The system of claim 4, wherein the OFDM standard includes computing an Inverse Fast Fourier Transform (IFFT) and adding a cyclic prefix.

6. The system of claim 4, further comprising: one or more D/A units to perform digital-to-analog (D/A) conversion on outputs of the OFDM units to generate analog baseband signals; and one or more Radio Frequency (RF) units to up-convert the analog baseband signals to radio frequencies and transmit the signals using corresponding one or more transmit antennas.

7. The system as in claim 1 wherein the MU-MAS frequency/phase offset aware precoding unit is implemented as a Minimum Mean Square Error (MMSE), a weighted MMSE, a Zero Forcing (ZF), or a Block Diagonalization (BD) precoder.

8. The system of claim 1, wherein the channel condition information is estimated by transmitting training sequences between a plurality of transmit antennas or the wireless client device, and the channel condition information is used to estimate a frequency or phase offset between the transmit antennas.

9. A system for compensating in-phase-quadrature (I/Q) imbalance of multiuser multiple antenna system (MU-MAS) communications, the system comprising:

one or more code modulation units for coding and modulating information bits for each of a plurality of wireless client devices to generate coded and modulated information bits;

one or more mapping units for mapping the coded and modulated information bits into complex symbols; and

an MU-MAS IQ aware precoding unit to compute MU-MASIQ aware precoding weights using channel condition information, the MU-MAS IQ aware precoding unit to precode complex symbols obtained from the mapping unit using the weights to pre-cancel interference due to I/Q gain and phase imbalance and/or inter-user interference.

10. The system as in claim 9 wherein the channel condition information is obtained through feedback from the wireless client devices to the MU-MAS IQ-aware precoding unit.

11. The system according to claim 9, wherein the channel condition information in downlink is obtained from uplink at the MU-MASIQ-aware precoding unit by exploiting channel reciprocity.

12. The system of claim 9, further comprising:

one or more Orthogonal Frequency Division Multiplexing (OFDM) units to receive the precoded signals from the MU-MASIQ aware precoding unit and to modulate the precoded signals according to an OFDM standard.

13. The system of claim 12, wherein the OFDM standard comprises computing an Inverse Fast Fourier Transform (IFFT) and adding a cyclic prefix.

14. The system of claim 12, further comprising:

one or more D/A units to perform digital-to-analog (D/A) conversion on outputs of the OFDM units to generate analog baseband signals; and

one or more Radio Frequency (RF) units to up-convert the analog baseband signals to radio frequencies and transmit the signals using corresponding one or more transmit antennas.

15. The system as in claim 9 wherein the MU-MAS IQ-aware precoding unit is implemented as beamforming or Maximum Ratio Combining (MRC).

16. The system as in claim 9 wherein the MU-MAS IQ-aware precoding unit is implemented as a Minimum Mean Square Error (MMSE), a weighted MMSE, a zero-forcing (ZF), or a block-diagonalization (BD) precoder.

17. The system of claim 9, wherein the channel condition information is estimated by transmitting training sequences between a plurality of transmit antennas or the wireless client device, and the channel condition information is used to estimate I/Q gain and phase imbalance for the transmit antennas.

18. The system of claim 9, comprising one or more RF units to convert signals between baseband and RF.

19. A system for dynamically adapting communication characteristics of a multi-user multi-antenna system (MU-MAS) communication system, comprising:

one or more code modulation units for coding and modulating information bits for each of a plurality of wireless client devices to generate coded and modulated information bits;

one or more mapping units for mapping the coded and modulated information bits into complex symbols; and

an MU-MAS configurator unit for determining subsets of users and MU-MAS transmission modes based on channel characteristic data and responsively controlling the code modulation units and the mapping unit.

20. The system in claim 19 wherein the channel characteristic data is obtained through feedback from the wireless client device to the MU-MAS configurator unit.

21. The system as in claim 19 wherein the channel characterization data in the downlink is obtained from the uplink at the MU-MAS configurator unit by exploiting channel reciprocity.

22. The system of claim 19, further comprising:

an MU-MAS precoding unit operating under control of the MU-MAS configurator unit to calculate precoding weights for precoding data signals before being transmitted to the wireless client devices.

23. The system of claim 22, further comprising:

one or more Orthogonal Frequency Division Multiplexing (OFDM) units to receive the precoded signals from the MU-MAS precoding unit and to modulate the precoded signals according to an OFDM standard.

24. The system of claim 23, wherein the OFDM standard comprises computing an Inverse Fast Fourier Transform (IFFT) and adding a cyclic prefix.

25. The system of claim 23, further comprising:

one or more D/A units to perform digital-to-analog (D/A) conversion on outputs of the OFDM units to generate analog baseband signals; and

one or more Radio Frequency (RF) units to up-convert the analog baseband signals to radio frequencies and transmit the signals using corresponding one or more transmit antennas.

26. The system as in claim 22 wherein the MU-MAS precoding unit is implemented as a Minimum Mean Square Error (MMSE), a weighted MMSE, a zero-forcing (ZF), or a block-diagonalization (BD) precoder.

27. The system as in claim 19 wherein the channel characterization data is estimated by sending training sequences between multiple transmit antennas and the wireless client devices and is used to estimate a subset of wireless client devices, transmit antennas, or transmission modes for the MU-MAS communication system.

28. The system as in claim 19 wherein the MU-MAS configurator unit uses polarization and/or pattern diversity techniques as a method to reduce array size while achieving diversity over the wireless link.

29. The system of claim 19, wherein communication occurs via near-normal incidence sky-wave (NVIS) and/or ground-wave as a method to increase diversity and downlink throughput.

30. The system of claim 19, wherein pattern diversity is used to communicate with a particular user via ground waves and communicate with other users via NVIS.

31. A system as in claim 29, where each client employs spatial separation of ground waves and NVIS links as a method of increasing spatial diversity of the links.

32. The system of claim 19, further comprising a base station for adaptively switching between different array geometries and different antenna diversity techniques based on channel quality feedback from a client as a way to increase diversity and downlink throughput of a link.

33. The system of claim 19, further comprising a base station that defines groups of users and schedules different groups of users to transmit based on their priorities and/or channel conditions.