HK1117968A - Symbol timing corrections in a multi carrier system by using the channel estimation - Google Patents

Description

Symbol timing calibration using channel estimation in multi-carrier systems

Cross Reference to Related Applications

This application claims priority from U.S. provisional patent application No.60/660905 entitled "interaction between timing tracking algorithm and channel estimation in wireless communications" filed on 3/10/2005, which is incorporated herein by reference in its entirety.

Technical Field

The subject technology relates generally to communication systems and methods, and more particularly to systems and methods that perform symbol timing calibration applied to channel estimation of pilot symbols in a wireless network.

Background

Orthogonal Frequency Division Multiplexing (OFDM) is a digital modulation method that divides a signal into multiple narrowband channels of different frequencies. These channels are sometimes referred to as subbands or subcarriers. This technique first arises during research to minimize interference between mutually adjacent channels over frequency. In certain aspects, OFDM is similar to conventional Frequency Division Multiplexing (FDM). The difference is that the modulation and demodulation of the signal are different. In general, priority is given to minimizing interference between channels, or crosstalk, and symbols comprising data streams.

In one area, OFDM has been used for european digital audio broadcasting services. This technology is useful for digital television and is being considered as a method for achieving high-speed digital data transmission over conventional telephone lines. It is also used in wireless local area networks. Orthogonal frequency division multiplexing may be considered as an FDM modulation technique for transmitting large amounts of digital data over a radio wave by splitting the radio signal into multiple smaller sub-signals or sub-carriers, which are then transmitted simultaneously at different frequencies to a receiver. One of the advantages of OFDM technology is that it reduces the amount of crosstalk in signal transmissions, with various OFDM aspects being employed in current specifications such as 802.11a WLAN, 802.16 and WiMAX technologies. Another example of an OFDM-based wireless system is FLO (forward link only), which is a wireless system that has been developed to efficiently broadcast real-time audio and video signals to mobile receivers using OFDM technology.

Wireless communication systems such as FLO are designed to operate in a mobile environment where the channel characteristics vary considerably over a period of time depending on the number of channel taps with significant energy, path gain and path delay. In an OFDM system, the timing synchronization block at the receiver side responds to changes in the channel shape by selecting an OFDM symbol boundary that is suitable for maximizing the energy captured in the FFT window. When such timing calibration occurs, it is important that the channel estimation algorithm take such timing calibration into account during the calculation of the channel estimate for demodulating a given OFDM symbol. In some implementations, the channel estimate is also used to determine the timing adjustments that need to be applied to the symbol boundaries of future symbols, resulting in a subtle interplay between the timing alignment that has been introduced and the timing alignment that will be determined for the future symbols. In addition, it is common to process pilot reports from multiple OFDM symbols using a channel estimation block to produce a channel estimate with better noise averaging and accounting for longer channel delay spread. When pilot reports from multiple OFDM symbols are processed together to produce channel estimates, it is important to calibrate the underlying OFDM channel estimates according to the symbol timing. Without such calibration, false channel estimates would result, thereby not ensuring proper operation of the wireless receiver.

Disclosure of Invention

The following presents a simplified summary of various embodiments in order to provide a basic understanding of some aspects of the embodiments. This summary is not an extensive overview. It is not intended to identify key/critical elements or to delineate the scope of the embodiments disclosed herein. Its sole purpose is to present some concepts in a simplified form as a prelude to the more detailed description that is presented later.

When a wireless receiver processes multiple symbols, a timing alignment needs to be determined for a multi-carrier system in a wireless network. The timing calibration is used for channel estimates obtained from pilot reports on multiple symbols. Typically, each of these symbols may actually be using a different FFT window due to different timing alignment on the relevant symbol. These timing offsets are used to calculate potential drift in the sampling clock and dynamic channel conditions caused by the mobility of the receiver during reception of a signal from a given transmitter.

In one embodiment, the symbols within a subset of symbols (e.g., 3 symbols) are first adjusted in time according to themselves. Subsequent symbol alignment or adjustment occurs during demodulation processing of the received symbols, beginning with timing information obtained and determined from such initial adjustment. For example, timing changes and corrections may be applied to preceding or following symbols when demodulating a current symbol. Thus, in some cases where a newly determined time is applied to each symbol, a different time reference is continuously determined, while in other cases, the previous timing is used to calculate the timing difference between the symbols. In one aspect, a method for timing calibration in a multi-carrier system is provided. The method comprises the following steps: two or more symbols are aligned with respect to each other from the subset of symbols to calculate a timing difference between the symbols. One or more of the subset of synchronization symbols are then timed using the timing difference between the symbols.

To the accomplishment of the foregoing and related ends, certain illustrative embodiments are described herein in connection with the following description and the annexed drawings. These ways in which the implementations may be implemented are set forth in the various ways, all of which are intended to be covered.

Brief Description of Drawings

Fig. 1 is a schematic block diagram illustrating a wireless communication network and a receiver that perform symbol timing calibration.

Fig. 2 and 3 illustrate examples of timing aspects and calibration of a wireless communication network.

Fig. 4 shows an example of timing calibration considerations for a wireless receiver.

Fig. 5 shows an example of a data boundary pattern.

Fig. 6-9 show examples of simulation data for the timing calibration process.

Fig. 10 shows an example of timing alignment processing in a wireless system.

Fig. 11 is a diagram showing an example of a network layer of a wireless system.

Fig. 12 is a diagram showing an example of a user equipment of a wireless system.

Fig. 13 is a diagram showing an example of a base station of a wireless system.

Fig. 14 is a diagram showing an example of a transceiver of a wireless system.

Detailed Description

Systems and methods are provided for determining timing alignment in a forward link only network. In one aspect, a timing calibration method for a multi-carrier system is provided. The method comprises the following steps: two or more symbols are aligned with respect to each other from the subset of symbols to crystallize a timing difference between the symbols. One or more of the subset of synchronization symbols are then timed using the timing difference between the symbols. In one example, timing synchronization can be performed in a time filter module associated with a channel estimation module.

As used in this application, the terms "component," "network," "system," "module," and the like are intended to refer to a computer-related entity, either hardware, a combination of hardware and software, or software in execution. For example, an element may be, but is not limited to being, a process running on a processor, an object, an executable, a thread of execution, a program, and/or a computer. By way of example, both an application running on a communication device and the device can be an element. One or more elements may reside within a processor and/or a thread of execution and an element may be localized on one computer and/or distributed between two or more computers. In addition, these elements can execute from various computer readable media having various data structures stored thereon. The elements may communicate over local and/or remote processes such as in accordance with a signal having one or more data packets (e.g., data from one element interacting with another element in a local system, distributed system, and/or across a wired or wireless network such as the internet).

Fig. 1 shows a wireless network system 100 for performing timing calibration. The system 100 includes one or more transmitters 110 in communication with one or more receivers 120 over a wireless network. Receiver 120 may comprise virtually any type of communication device, such as a cellular telephone, computer, personal digital assistant, handheld or laptop device, and the like. Portions of the receiver 120 may be used to decode a subset of symbols 130 having one or more symbols, which may be sampled according to different symbol timing, such that the receiver uses a calibration element 140 to account for timing differences between symbols. The timing alignment applied to the channel estimates in the receiver 120 is obtained from pilot reports on the multiple symbols 130.

Typically, each of the symbols 130 actually uses a different Fast Fourier Transform (FFT) window due to the different timing alignment across the associated symbols. Thus, during reception of a signal from a given set of one or more transmitters 110, timing offsets may be generated due to potential drift in the sampling clock and channel dynamics due to movement of the receiver 120. As shown, the calibration element 140 may be associated with a time filter module 150 that works in conjunction with a channel estimation module 160. The subset of symbols 130 is typically transmitted in an Orthogonal Frequency Division Multiplexing (OFDM) network that employs a Forward Link Only (FLO) protocol for multimedia data transmission. Channel estimation is typically based on evenly spaced pilot tones inserted in the frequency domain and in each OFDM symbol. In a particular implementation, the pilots are spaced apart by a distance of 8 carriers, and the number of pilot carriers is set to 512 (12.5% overhead).

In one aspect, multicarrier communication system 100 is considered to be where frequency-domain multiplexed (FDM) pilots placed within transmitted symbols are used for channel estimation. In this system, by using FDM pilot interlaces, more information about the propagation channel can be extracted (obtaining a longer channel estimate) using multiple consecutive received symbols 130. In one example, this may be performed by the calibration element 140 in the time filter module 150 of the channel estimation module 160. Since timing calibration may be performed concurrently with this process, the calibration element facilitates different OFDM symbol timings that may occur on multiple adjacent symbols to be accounted for in the time filtering module 150 at 130. The time alignment process also solves the interaction problem between the channel estimation and timing synchronization modules. In one aspect, a time alignment element for a wireless receiver is provided. This may include means (e.g., 120) for receiving a subset of symbols in an OFDM broadcast, and means (e.g., 150) for filtering the subset of symbols. This may also include means for calibrating the symbols within the subset and means (e.g., 140) for calibrating one symbol based on the current demodulation condition in the subset of symbols.

In one embodiment, the symbols within the subset of symbols 130 are first aligned in time with respect to each other. For example, if three symbols are used for channel estimation and subsequently for timing offset determination, then an adjustment of the difference between the three symbols may be determined. Based on the information obtained and determined from the initial calibration, subsequent symbol calibrations or adjustments may be made during the demodulation process to demodulate the received symbols sampled from the current symbol at different symbol timings. For example, while demodulating the current symbol (which may be the fourth symbol in the subset), timing changes and calibration may be applied simultaneously to the previous or subsequent symbols, e.g., to symbol two in the subset, via calibration element 140. In this way, different timings are successively determined in the case where the newly determined timing offsets are applied to the respective symbols, and in other cases, the previous timings are applied to calculate the timing difference between the symbols 130. It should be noted that symbol timing calibration may occur in multiple combinations. For example, if three symbols are used, then in the event that the timing of one symbol is maintained or adjusted according to the other two symbol members in the subset 130, in fact 8 different combinations of adjustments may occur. For example, the second symbol may have a timing that is aligned according to the first symbol timing and the third symbol timing. In another example, symbol one may be adjusted based on symbol two and symbol three, etc. It is to be understood that different numbers of symbol subsets 130 and timing calibrations can be employed.

Timing synchronization in a multi-carrier system involves determining the correct position of the FFT sampling window for demodulating OFDM symbols. Assuming that the equivalent channel between the transmitter and the receiver is characterized by a delay spread shorter than the length of the cyclic prefix embedded at the beginning of each symbol, unwanted inter-symbol interference (ISI) can be avoided. This can happen at any time depending on the receiver's capabilities to recover the correct timing position from the incoming data stream. The optimal location for FFT window placement (also referred to as symbol sampling) begins at the first sample after the cyclic prefix. In OFDM systems, information about timing synchronization can be extracted from the channel estimate. These can be obtained with the help of pilot tones using some preliminary knowledge about the correct sampling positions. The choice of synchronization for channel estimation assistance is motivated by the reporting that an offset in the position of the FFT window for channel estimation results in an estimate of the appropriate drift. Therefore, estimating this drift is generally equivalent to estimating the sample offset. In many OFDM systems, the timing synchronization block uses channel estimates obtained from previous data symbols to calculate the drift from the ideal sampling position and applies the offset to arrive at the OFDM symbol.

The channel estimation system 100 may be designed in a manner that allows it to process channels up to twice the length of the cyclic prefix or longer. This may be achieved by pilot interleaving. On the one hand the receiver 120 comprises a so-called uncorrelated time filter which combines channel reports from at least three consecutive OFDM symbols to calculate a longer channel estimate, which is then used for demodulation. If the synchronization unit indicates that a non-zero offset should be used when sampling the next OFDM symbol, then the corresponding channel is not calibrated using the first two channel reports (since this would result in using a non-zero calibrated channel estimate). Thus, the combination of these three reports produces a distorted result. A remedy is to apply the appropriate transformation on the first two channel estimates when the timing offset is non-zero to keep them calibrated with the current one.

Some introductory mathematical discussions are provided below for the purpose of the more detailed reports given below. In the frequency domain k^thThe received OFDM symbol can be written as:

Y(k)＝H(k)+w(k)＝W_P，Dh (k) + w (k) equation 1

Wherein

● P is the number of pilot carriers and D is the number of channel taps assumed by the receiver.

● vector Y, H, w is of length P and noise w is of variable N₀White complex gaussian.

● matrices WP, D are non-standard DFT matricesWhere N is the total number of subcarriers.

● vector h (k) has a length D and is normalized so that E [ h (k)^Hh(k)]＝E_PIn which E_PIs the energy of the received pilot symbols. From the above definition, it is easy to see that the channel value of each carrier in the frequency domain satisfies:

E|H_P(k)|²＝E_P

from equation 1, it is apparent that the number of channel taps D ≦ P. However, it is generally desirable for longer channel estimates a) to be precisely timing synchronized when providing longer channel estimates and positioning the FFT window to maximize the collected energy; b) the case where the channel has a larger delay spread than the cyclic prefix is handled. To produce longer channel estimates, the pilots are interleaved on consecutive OFDM symbols in frequency on the one hand, i.e. the pilot carrier indications vary over consecutive OFDM symbols. For simplicity, two symbol interleaving patterns may be assumed: for example, for a FLO system with 96 guard carriers, the pilot carrier indication is {50, 58.., 4042} in even symbols and {54, 62.., 4046} in odd symbols. More generally, if the uniformly spaced pilot carriers are in the form of even symbolsThen in the odd symbols they will be

By this interleaving, it is possible to connectThe received estimate may be up to 2P in length by using pilot reports from at least two adjacent OFDM symbols. Specifically, assume that the channel has 2P time domain taps (and sets n₀2). Then it is determined that,

wherein N is 8P

While

Wherein N is 8P

Thus, the pilot reports in even and odd symbols can be written as:

Y(2k)＝W_P，PΛ₁[h^aclual(2k)-jh^excess(2k)]+w(2k)

Y(2k+1)＝W_P，PΛ₂[h^aclual(2k+1)+jh^excess(2k+1)]+w(2k+1)

equation 2

WhereinAnd "actual" and "processes" refer to taps corresponding to 1 ═ 0., P-1 and 1 ═ P., 2P-1.

To determine the channel estimate from the report in equation 2, one step is to use the least squares criterion:

equation 3

The above estimation includes actual and redundant elements. One possible way to obtain a full 2P tap channel estimate is to:

however, this is a special case of more general operation, where the time domain estimate in equation 3 (obtained per OFDM symbol) is averaged over multiple OFDM symbols. This is the timing filtering step of the channel estimation. Timing filtering may be performed separately for each time domain tap, and the estimated tap l generated in any OFDM symbol m (even or odd) may be written as:

wherein N is_fAnd N_bRespectively the number of uncorrelated and correlated taps. It should be noted that due to the staggering, the pilots in the frequency domain may not be filtered, and thus timing filtering is performed in the time domain. In other words, the order of least squares estimation and temporal filtering cannot be swapped. Filter coefficient { alpha_nProvides a trade-off between gain due to collecting conventional pilot energy from the symbol instead of the current symbol (reduced pilot noise) and loss due to channel variations over the symbol (reduced time-varying noise). In addition, the first and second substrates are,as shown above, since the estimate in equation 3 includes a contribution from the excess delay component, timing filter coefficients may be used to suppress such a contribution. It is important to ensure that the channel estimates are adjusted in time before timing filtering the channel estimates from multiple symbols. The following discussion provides one example of an apparatus for determining and performing such timing calibration for channel estimates collected from multiple symbols prior to timing filtering.

Fig. 2-3 and the additional discussion provide different examples of how timing calibration may be performed in an OFDM system. It can be assumed that the actual composite channel is limited to M taps in the time domain, i.e.

In addition, in the following, bold letters represent matrices and vectors. Symbols Wk are stored as K KDFT matrices and Ik is stored as an identification matrix of size K. If H (z) is the channel transfer function defined in (1), then H [ k ] is used to represent its own kth DFT coefficient (0 ≦ k-1), defined as:

if K is 4096, then the coefficient (2) corresponds to the frequency domain channel gain on the carrier tone. In general, the dimension (dimension) K of the fourier transform is always clear from the context.

In the following description, the effect of misplacing symbol samples in the channel report is considered. To investigate these effects, when the timing alignment is good, a channel report α is generated based on the interleaved pilots, (0 ≦ α ≦ 7), but the length of the channel M ≦ N ≦ 8P ≦ 4096 (where N is the number of subcarriers and P is the number of pilots). This results in

The channel value a (0 ≦ l ≦ p-1) centered on the pilot interlace, resulting in

Note that the summation in parenthesis represents the P-point DFT of the channel response within the r-th alias bin (alias bin). Referring to fig. 2, a diagram 200 identifies eight pseudonyms for a channel of length N4096. (3) The results of these two phase calibration periods in (1): (a) consider non-zero interleaving, and (b) have fewer reports than channel taps, which leads to the nomenclature of alias (alias). It is clear from (3) that considering only a single interlace, it is not possible to identify channels of length M > P due to this alias. This fact motivates the pilot interleaving technique. For different 1-value convergence equations (3), matrix equations

Has the advantages ofOne conclusion is that each channel report of form (4) consists not only of channel samples contained in the zeroth pseudonym band with proper phase alignment, but also of an overlap of content from all pseudonyms. Under the above analogous symbols, by Y_α(n) on the left-hand side of (4)Where the index n represents the point in time at which the report is collected and the subscripts represent the corresponding interleaving.

In one case, timing synchronization is assumed to be good, and there is no drift in the channel estimate. In other words, the channel impulse response estimated at 200 in fig. 2 begins at h (0) of the 0 position, as shown at 210 in fig. 2. Canceling the assumption that the maximum delay spread corresponds to 1024 samples yields a channel report Y_α(n) consists of only pseudonym bands 0 and 1. Thus, in the ideal case, produce

By passing through_a(n) a vector representing the first P taps of the channel at time n, and h_a(n) represents the second P tap, and the right hand side of (5) becomesThus, when pilots occupy interlaces 2 and 6, a channel of length 2P can be estimated from two consecutive reports as follows:

and

this operation is implemented in a so-called uncorrelated time filter operating in the channel estimation block, which is only spread over 3 consecutive reports to get a better noise averaging. In the following, a description will be given of how these reports and channel estimates are affected by timing synchronization errors.

Channel estimation at time nIs according to the report y_α1(n-1)y_α2(n)，y_α2(n) and y_α1(n + 1). Based on h, a data pattern timing tracking (DMTT) unit calculates the correct sampling position for the next OFDM symbol. The result of taking into account the timing synchronization indicates that the sampling position needs to be changed. This implies that inaccurate samples may be used to obtain the previous channel report. Next, the generated defects are described.

At 220 in fig. 2, two possible timing errors are shown, which result in erroneous FFT window positions. Window position 1 of 230 is referred to as early sampling, which results in a delayed channel estimate-240 shown. Note that in this case, a total of three consecutive pseudonyms contain channel taps, which is reflected in the changed channel report.

Similarly, in the case of late sampling (window position 2 of 250), advanced channel estimates are reported. Assume that it starts at sample x and ends at 2P-x. However, since the insertion of a cyclic prefix in an OFDM system transforms the linear into a cyclic convolution, an equivalent channel estimate is shown at 260. Third, three circularly continuous pseudonymous frequency bands, frequency bands 7, 0 and 1, are occupied. The corresponding channel report is:

the timing tracking unit assumes the sampling instant used to calibrate future symbols, but to ensure uninterrupted performance of the channel estimation time filter, the distortion in the previous channel report will not work.

Subsequent operations performed by the channel estimation and DMTT block are shown at 300 of fig. 3. During reception of symbol n, channel reports from symbols n-3, n-2 and n-1 are prepared and a channel estimate h (n-2) is calculated. At this point, the demodulation block begins operating on symbol n-2. While the DMTT unit observes h (n-2) and estimates the exact sample position of the next symbol (n +1) based on the algorithm used for timing synchronization.

Assume that a non-zero sample offset is detected at this time, i.e., sample calibration should be applied to symbol n +1, see 300 in the figure. This signal triggers two other operations: cyclic rotation of current channel estimates and calibration of previous channel reports. Without loss of generality, it is assumed that the position offset x is detected by the DMTT, i.e., the symbol samples that have passed at this time have been outdated. This corresponds to 240 of fig. 2. Note that the timing calibration applied does not affect the calculation of the channel estimate for symbol n-1, which uses pilot reports from symbols n-2, n-1, and n. Thus, no modification of the demodulated data from symbol n-1 is required. During the next OFDM symbol, h (n) is calculated using channel reports y (n-1), y (n), and y (n + 1). Note that y (n +1) is obtained using the latest timing, and y (n-1) and y (n) may be corrupted. However, the channel estimate h (n) computed for symbol n should match the timing window used to sample the nth symbol. If this is not done, there is a mismatch between the channel gain experienced by the nth symbol and the channel estimate h (n) generated to decode the data in the nth OFDM symbol. Thus, a suitable calibration would be applied to y (n +1) to match the symbol timing of y (n-1) and y (n) to produce h (n).

During the next OFDM symbol, channel report y (n), y (n +1) and y (n +2) will be used to generate channel estimate h (n +1) to demodulate the data from y (n + 1). Note that new timing has been used from y (n +1), so when y (n) arrives at different timing, y (n +1) and y (n +2) also arrive at the same timing. Thus, to generate a channel estimate h (n +1) to demodulate y (n +1), it should be ensured that h (n +1) carries the timing for y (n + 1). Thus, timing calibration will be applied to y (n) to match the timing of y (n +1) and y (n +2) to produce h (n + 1). In this approach, it is assumed that the channel estimates from h (n +2) are calibrated with zeros until the next channel drift causes the DMTT cell to reflect. The characteristics of the timing alignment applied in conjunction with the operations involved in the relatively early and late sampling of the OFDM symbol are given below.

The transformation back to the desired form is given by the following matrix (a below stands for pilot interleaving)

Early sampling calibration:

and (3) late sampling calibration:

note by first looking at y_α ^early(n) performing a cyclic right shift of X samples and then multiplying the first X samples (those that overflow) byTo be easily realized at y_α ^early(n) performing an early sample calibration of x samples. Similarly, at y_α ^lateLate sample calibration by x samples on (n) may be performed by sampling at y_α ^late(n) performing a circular left shift of x samples and then going to the endx samples (those that overflow) timesTo be implemented.

These transformation amounts to a simple cyclic shift allowed by constant complex multiplication are applied to a portion of the samples. The sequence of operations is summarized below:

1. if the offset provided by the DMTT unit at OFDM code element (n +1) is x > 0

● begins the sampling of the next OFDM symbol n +2 by delaying x samples.

● to obtain the channel estimate h (h +1) for symbol n +1, the future channel report is cyclically shifted x samples to the right from y (n +2) and the sample calibration previously given in equation 8 above is applied.

● to obtain the channel estimate h (n +2) for symbol n +2, the y (n +1) x samples are reported to the left by circularly shifting the previous channel to the left and applying the late sample calibration given in equation 9 above.

2. Otherwise, if the offset provided by the DMTT cell is x < 0

● begins sampling the next OFDM symbol by x samples in advance.

● to obtain the channel estimate h (n +1) for symbol n +1, the future channel report is cyclically shifted x samples to the left from y (n +2) and the late sample calibration given in equation 9 above is applied.

● to obtain the channel estimate h (n +2) for symbol n +2, the previous channel report y (n +1) x samples are cyclically shifted to the right and the early sample calibration given in equation 8 above is applied.

Note that while the above discussion was performed using the example of an uncorrelated time filter with one uncorrelated tap, the techniques discussed herein are fairly generic in scope and can be readily extended to time filters of any length. In the above example, it is also assumed that only one symbol is out of synchronization and therefore needs to be aligned with the other symbols. In a more general case, all symbols processed by the channel estimation algorithm will see different symbol timing. The concepts of early sampling and late sampling alignment described above will apply to each symbol with an argument (argument) given by the corresponding timing offset. In particular, it should be ensured that the timing of all channel reports for processing must match the timing of the samples used to generate the OFDM symbol to be decoded.

At the end of the above-described operating set, the channel estimates are all time aligned to enable timing filtering of the time domain channel estimates. Assuming that there is no excess delay spread (see discussion below) and the channel is truncated to P taps, the loss with channel estimation can be analyzed.

Fig. 4 shows an example of timing calibration considerations 400. At 410, timing alignment is considered based on the lesser or greater delay spread of the transmitted symbols. For a given set of coefficients { alpha }_nThe SNR loss from a good channel estimate can be given by the following equation:

equation 4

One can assume sampling (D ═ P), and E_dRepresenting the data symbol energy. Parameters r and σ_h ²In relation to the time filter coefficients and the channel variation over the symbol:

and

where R (n) is a correlation function for each channel tap, with a normalization to the OFDM symbol interval T_sIs used as the argument of (1). For a frequency f with Doppler_dThe model of the Jakes of (1),

R(n)＝J₀(2πf_dT_sn)

SNR_effincluding ICI contribution due to doppler, as a function of actual SNR as follows:

where σ is the Doppler spectrum given_ICI ²Can be obtained exactly. For the Jakes spectrum, the upper stringency limit is given as follows:wherein T is_FFTIs the duration of the FFT (excluding the cyclic prefix).

Continuing to 420 of FIG. 4, consider the selection of a non-causal filter. Associated FIR time filter (N) for channel estimation_f0) has been studied in detail in the past. The filter taps are optimized using the Robust MMSE method and the general linear decay technique. However, analytical compromises and simulation results indicate that they do not achieve reasonable gains over non-timed filtering, with the overall range of speeds (up to 120 km/h) and spectral efficiency (< 2bps/Hz) being current goals. These results indicate the limitation of using the correlation filter.

An improved compromise can be made if a non-causal filter is provided. The use of more than one uncorrelated tap may be prohibitive in terms of buffering requirements, and therefore, one uncorrelated tap is preferred-however, more than one may be used. For simplicity, one past symbol is used, giving a total of 3 taps for the time filter. To obtain unbiased estimates in statistical channels, one constraint is:

∑α_n＝1

in addition, by symmetric channel correlation in time, in effect, equal weighting is applied to the past and future OFDM symbols, α_-1＝α₁. Under this constraint, the selection of uncorrelated filter coefficients is reduced to the selection of one parameter, the center tap α₀. Using equation 4 above, α can be transformed₀To provide a compromise between statistical losses and high speed losses. Referring briefly to fig. 5, the trade-offs of a 4-tap MMSE filter and a 3-tap non-causal filter at an operating SNR of 20dB and using equation (4) are compared by a graph 500. For a non-causal filter, the region of interest is the lower line from (3, 3) corresponding to no time filter to (1.25, 1.4) corresponding to equal weighting for the three symbols. From FIG. 3 can be seenIt is clear that the non-causal filter is robust to time variations at high speed and provides a better compromise than the associated MMSE filter. In addition, it is best to make all three taps of the non-causal filter equal (to 1/3) as it minimizes the loss in the statistical channel, and this loss is almost the same in the high speed channel as well. However, it should also be considered that the tap weights α are selected₀Influence of time excess delay spread: robustness to excess delay spread of the channel.

Continuing back to 430 of FIG. 4, consideration of excess delay is described. Since the channel is sampled in the frequency domain strictly at 512 pilots per OFDM symbol, over 512 time domain channel samples will be known as (alias into) the first 512 taps. Thus, in the presence of excess delay spread, the reported l^thThe time domain channel taps can be written as (for even k and pilot in interlace 2)

In addition, but with four carrier interleaved (starger) pilots on consecutive OFDM symbols, the reporting channels in future and past OFDM symbols can be written as:

thus, by using a non-causal filter, a good channel estimate becomes:

for the case of statistical channels, the actual and excess channels are independent of k, and the filter output is reduced to h_l ^actual-(2α₀-1)h_l ^excess。

The ideal case is to eliminate the excess delay contribution to the reporting channel and estimate only the actual channel. This can be achieved by mixing alpha₀Is provided as 1/2 instead of 1/3. Another problem is how the non-causal filter handles time variations in the excess channel. However, the excess channel variation is nearly linear over three symbols, and it is clear that any symmetric choice of taps will also eliminate the time variation in the excess channel. As discussed above, the selection of 0.25, 0.5, 0.25 for the uncorrelated taps removes the time variation in the actual channel, any wrap-around of the excess delay channel taps and any time variation of these excess taps. One problem with selecting these taps instead of equal taps is that the statistical loss increases from 1.25dB to 1.38dB, but this is relatively small. Therefore, using a three-tap filter with coefficients 0.25, 0.5, 0.25 produces a code packet error result in the next section.

Continuing to 440 of fig. 4, energy considerations are discussed to optimize data-to-pilot energy ratios. The previous discussion assumed that the data symbol energy was about the same as the pilot symbol energy. Increasing the pilot symbol energy results in better channel estimation (or lower pilot noise) under the constraint that the total pilot + data energy is fixed, at the expense of lower data symbol SNR (higher data noise). This ratio can be chosen for compromise optimization. For statistical channels, the trade-off can be analytically optimized and improved when the energy ratio is not optimized as follows:

the equation in brackets is the statistical loss after data-pilot energy optimization. For a non-causal filter, the improvement is equal to about 0.16 dB.

Fig. 6-9 show simulation examples of the timing calibration process. The simulation results show QPSK/16 QAM with rate 1/2 coding (and thus spectral efficiency of 1bps/Hz and 2bps/Hz) and low/high speed channels. For low speeds, the duplicate ATSC channel model is considered to have a delay of 5dB in the second cluster below the main cluster and at 40 mus. The ATSC channel model has a robust spectral content that is truly statistical, and the Rayleigh component of the channel is assumed to decay at a rate of 20 km/h. For high speed, the repetitive PEDB shape was used with the same cluster delay of 40 μ s and power difference of 5 dB. All paths in the "PEDB" channel are Rayleigh fading at 120 km/h speed. Thus, it can be assumed that there are 96 guard carriers and that the frequency domain interpolation assumes that the channel value of the guard pilot is the same as the channel value of the most recently transmitted pilot.

The results of fig. 6-9 include ICI effects due to channel variations within an OFDM symbol. ICI should be included in the noise variance estimate used for LLR calculation. An actual noise variation estimation algorithm is used. In addition, threshold techniques are used to mitigate pilot noise, with a threshold of 0.1. The thresholding operation is performed after the timing filtering operation is performed in the time domain.

Fig. 6 and 7 show the performance of QPSK and 16QAM modulation in a slow fading channel with an ATSC shape. It can be seen that the three tap non-causal filter produces a gain of about 1.6dB compared to the non-timed filtering case in both cases with slow fading channels. The performance results of fig. 8 and 9 at high speed confirm that the non-causal filter truly cancels the time variations in the channel, resulting in a channel estimate that is robust to time variation errors. The robustness of the non-causal filter is more pronounced when compared to the associated filter (e.g., robust MMSE) in high speed situations. Figure 9 shows the performance of a QPSK rate 1/2 code corresponding to doppler of approximately 195Hz over a repeated channel shape per dB at a speed of 300 km/h.

Fig. 10 shows a timing calibration process 1000 for a wireless system. While, for purposes of simplicity of explanation, the methodologies are shown and described as a series or number of acts, it is to be understood and appreciated that the processes described herein are not limited by the order of acts, as some acts may, in fact, occur in different orders and/or concurrently with other acts from that shown and described herein. For example, those skilled in the art will understand and appreciate that a methodology could alternatively be represented as a series of interrelated states or events, such as in a state diagram. Moreover, not all illustrated acts may be required to implement a methodology in accordance with the subject matter disclosed herein.

Continuing to 1010, a subset of symbols is received by a wireless receiver. At 1020, the symbols in the received subset of symbols are aligned in time with respect to each symbol in the subset for which timing alignment is employed. After initial alignment between symbols at 1020, subsequent symbol alignment and adjustment is determined at 1030, where new and old timing patterns are determined. At 1040, during demodulation processing of the received symbols in the wireless receiver, a symbol is aligned with a new or old timing pattern when the current symbol is demodulated based on additional time considerations. As described above, a subset of the filter taps may be selected to perform time synchronization for symbols within the contents of a time filter module in the channel estimation module. Thus, as previously described, when demodulating a current symbol that may be the nth symbol in the subset (n being an integer), timing changes and calibration to previous or subsequent symbols in the subset may be employed simultaneously. In some cases where a newly determined time is applied to each symbol, a new or previous timing pattern may be determined, while in other cases, previous timing is applied to account for timing differences between one symbol and the remaining members of the subset.

Fig. 11 shows an example of a network layer 1100 of a wireless system. The Forward Link Only (FLO) air interface protocol reference model is shown in fig. 11. In general, the FLO air interface specification covers protocols and services corresponding to OSI6 with layer 1 (physical layer) and layer 2 (data link layer). The data link layer is further partitioned into two sub-layers, namely a Medium Access (MAC) sub-layer, and a stream sub-layer. The upper layers may include compression of multimedia content, access control of multimedia, and content and formatting of control information.

The FLO air interface specification typically does not specify upper layers to allow design flexibility in supporting various specifications and services. These layers are shown to provide a relationship between above and below. The stream layer includes multiplexing three upper layer streams into one logical channel, bundling upper layer packets into a stream for each logical channel, and providing packetization and residual error processing functions. Characteristics of the Medium Access Control (MAC) layer include controlling access to the physical layer, performing mapping between logical channels and physical channels, multiplexing logical channels for transmission onto physical channels, demultiplexing logical channels at the mobile device, and/or enforcing quality of service (QOS) requirements. The characteristics of the physical layer include the channel structure that provides the forward link and defines the frequency, modulation, and coding requirements.

In general, FLO technology utilizes Orthogonal Frequency Division Multiplexing (OFDM), which is also used by Digital Audio Broadcasting (DAB)7, digital video broadcasting-terrestrial (DVB-T)8, and integrated services digital broadcasting-terrestrial (ISDB-T). In general, OFDM technology can achieve high spectral efficiency while effectively meeting the mobility requirements of large cellular SFNs. In addition, OFDM may use cyclic prefixes of suitable lengths to handle long delays from multiple transmitters; a guard interval added to the front of the symbol, which is a replica of the last part of the data symbol, is used to help orthogonalize and mitigate inter-carrier interference. As long as the length of the interval is greater than the maximum channel delay, reflection of the previous symbol is removed, thereby preserving orthogonality.

Fig. 12 illustrates a user equipment 1200 employed in a wireless communication environment in accordance with one or more aspects set forth herein. User device 1200 includes a receiver 1202 that receives a signal from, for instance, a receive antenna (not shown), performs typical operations thereon (e.g., filters, amplifies, downconverts, etc.) the received signal, and digitizes the state signal to obtain samples. Receiver 1202 may be a non-linear receiver. A processor 1206 may be used for timing synchronization and channel estimation. FLO channel element 1210 is provided to process FLO signals as previously described. Processor 1206 can be a processor dedicated to analyzing information received by receiver 1202. User device 1200 additionally can include a memory 1208 operatively coupled to processor 1206 and storing information and instructions related to embodiments described herein.

It will be appreciated that the data store (e.g., memories) components described herein can be either volatile memory or nonvolatile memory, or can include both volatile and nonvolatile memory. By way of example, and not limitation, nonvolatile memory can include Read Only Memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), electrically erasable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM), which acts as external cache memory. By way of example, but not limitation, RAM may exist in a variety of forms, such as Synchronous RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), and direct random access bus RAM (DRRAM). The memory 1208 of the subject systems and methods includes, but is not limited to, these and other suitable types of memory. User device 1200 further includes a background monitor 1214 for processing FLO data.

Fig. 13 shows an example of a system 1300 that includes a base station 1302 having a receiver 1310 for receiving signals from one or more user devices 1304 via multiple receive antennas 1306 and a transmitter 1324 for transmitting to the one or more user devices 1304 via transmit antennas 1308. Receiver 1310 can receive information from receive antennas 1306 and is operatively associated with a demodulator 1312 that demodulates received information. Demodulated symbols are analyzed by a processor 1314, which is similar to the processor described above, coupled to a memory 1316 that stores user rankings, lookup tables related thereto, and/or any other suitable information related to performing the various operations and functions described herein. Processor 1314 is further connected to FLO channel 1318 elements that facilitate sending FLO information to one or more user devices 1304, respectively.

A modulator 1322 can multiplex the signal for transmission by a transmitter 1324 through transmit antennas 1308 to user devices 1304. FLO channel element 1318 can add information to a signal associated with an update data stream given a transport stream used to communicate with user device 1304, which can be transmitted to user device 1304 to indicate that a new best channel has been identified and acknowledged. In this manner, the base station 1302 can interoperate with a user device 1304 that provides FLO information and employs a decoding protocol in conjunction with a non-linear receiver.

Fig. 14 shows an exemplary wireless communication system 1400. The wireless communication system 1400 depicts one base station and one terminal for sake of brevity. However, it is to be appreciated that the system can include more than one base station and/or more than one terminal, and that the other base stations and/or terminals can be substantially the same or different for the exemplary base station and terminal described below.

Referring now to fig. 14, on a downlink, at access point 1405, a Transmit (TX) data processor 141O receives, formats, codes, interleaves, and modulates (or symbol maps) traffic data and provides modulation symbols (data symbols). A symbol modulator 1415 receives and processes the data symbols and pilot symbols and provides a stream of symbols. A symbol modulator 1420 multiplexes data and pilot symbols and provides them to a transmitter unit (TMTR) 1420. Each transmit symbol may be a data symbol, a pilot symbol, a signal value of zero. The pilot symbols may be continuous in each symbol period. The pilot symbols may be Frequency Division Multiplexed (FDM), Orthogonal Frequency Division Multiplexed (OFDM), Time Division Multiplexed (TDM), Frequency Division Multiplexed (FDM), or Code Division Multiplexed (CDM).

TMTR1420 receives and converts the stream of symbols into one or more analog signals and further conditions (e.g., amplifies, filters, frequency upconverts) the analog signals to generate a downlink signal suitable for transmission over the wireless channel. The downlink signal is then transmitted through an antenna 1425 to the terminals. At terminal 1430, an antenna 1435 receives the downlink signal and provides a received signal to a receiver unit (RCVR) 1440. Receiver unit 1440 conditions (e.g., filters, amplifies, and frequency downconverts) the received signal and digitizes the conditioned signal to obtain samples. A symbol demodulator 1445 demodulates and provides received pilot symbols to a processor 1450 for channel estimation. Symbol demodulator 1445 further receives a frequency response estimate for the downlink from processor 1450, performs data demodulation on the received data symbols to obtain data symbol estimates (which are estimates of the transmitted data symbols), and provides the data symbol estimates to an RX data processor 1455, which demodulates (e.g., symbol demaps), deinterleaves, and decodes the data symbol estimates to recover the transmitted traffic data. At access point 1405, the processing by symbol demodulator 1445 and RX data processor 1455 is complementary to the processing by symbol modulator 1415 and TX data processor 1410, respectively.

On the uplink, a TX data processor 1460 processes traffic data and provides data symbols. A symbol modulator 1465 receives and multiplexes the data symbols with pilot symbols, performs modulation, and provides a stream of symbols. A transmitter unit 1470 then receives and processes the stream of symbols to generate an uplink signal, which is transmitted via the antenna 1435 to the access point 1405.

At access point 1405, the uplink signal from terminal 1430 is received by the antenna 1425 and processed by a receiver unit 1475 to obtain samples. A symbol modulator 1480 then processes the samples and provides received pilot symbols and data symbol estimates for the uplink. An RX data processor 1485 processes the data symbol estimates to recover the traffic data transmitted by terminal 1430. A processor 1490 performs channel estimation for each active terminal transmitting on the uplink. Multiple terminals may transmit pilot concurrently on the uplink on their respective assigned sets of pilot subbands, which may be interlaced.

Processors 1490 and 1450 direct (e.g., control, cooperate, manage, etc.) operation at access point 1405 and terminal 1430, respectively. Respective processors 1490 and 1450 can be associated with memory units (not shown) that store program codes and data. Processors 1490 and 1450 can also perform computations to generate frequency and impulse response estimates for the uplink and downlink, respectively.

For multiple access systems (e.g., FDMA, OFDMA, CDMA, TDMA, etc.), multiple terminals may transmit simultaneously on the uplink. For such systems, the pilot subbands may be shared among different terminals. This channel estimation technique may be used in cases where the pilot subbands for each terminal span the entire operating band (possibly except for the bandwidth edges). This pilot subband structure is ideal to obtain frequency diversity for each terminal. The techniques described herein may be implemented in various ways. For example, these techniques may be implemented in hardware, software, or a combination thereof. For a hardware implementation, the processing units used for channel estimation may be implemented with one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), processors, controllers, micro-controllers, microprocessors, other electronic units designed to perform the functions described herein, or a combination thereof. In the case of software, it may be implemented with modules (e.g., procedures, functions, and so on) that perform the functions described herein. These software codes may be stored in memory units and executed by processors 1490 and 1450.

For a software implementation, the techniques described herein may be implemented with modules (e.g., procedures, functions, and so on) that perform the functions described herein. The software codes may be stored in memory units and executed by processors. The memory unit may be implemented within the processor or external to the processor, in which case it can be communicatively coupled to the processor via various means as is known in the art.

What has been described above includes exemplary embodiments. It is, of course, not possible to describe every conceivable combination of components or methodologies for purposes of describing the embodiments, but one of ordinary skill in the art may recognize that many further combinations and permutations are possible. Accordingly, these embodiments are intended to embrace all such alterations, modifications and variations that fall within the spirit and scope of the appended claims. Also, to the extent that the term "includes" is used in either the detailed description or the claims, such term is intended to be inclusive. The term comprising is to be interpreted in a manner similar to the term comprising as it is interpreted when employed in the claims in a transitional manner.

Claims (30)

1. A timing calibration method for a communication system, comprising:

calibrating symbol timing of two or more symbols from the subset of symbols with respect to each other to account for timing differences between the symbols; and

channel estimates are obtained based on symbols from a subset of symbols, where one or more symbols have been aligned in time.

2. A method according to claim 1, characterized in that the channel estimation is used to generate timing alignment information for symbols in the subset or symbols outside the subset.

3. A method according to claim 1, characterized in that said channel estimate is used to demodulate the data contained in the symbols from the subsets or symbols outside the subsets.

4. The method of claim 1, further comprising demodulating the first symbol during calibration of timing of a subsequent symbol or a previous symbol to the first symbol.

5. The method of claim 1, further comprising performing channel estimation of length 2P, where P is an integer number of pilot carriers.

6. The method of claim 5, further comprising employing pilot reports from at least two adjacent symbols to determine timing alignment.

7. The method of claim 1, further comprising determining one or more time filter taps to produce a channel estimate for data demodulation and determining a timing alignment.

8. The method of claim 1, further comprising performing a least squares criterion to determine the channel estimate.

9. The method of claim 8, further comprising determining an actual and excess portion for channel estimation.

10. The method of claim 8, further comprising averaging the time domain channel estimates over a plurality of symbols.

11. The method of claim 1, further comprising determining one or more time filter coefficients for channel estimation.

12. A channel estimation module for a wireless receiver, comprising:

a time filter element for processing a subset of symbols received in a forward link only network; and

a calibration element for adjusting timing between symbols in the subset of symbols and adjusting timing of at least one symbol relative to other symbol members in the subset during demodulation of a current symbol.

13. The module according to claim 12, characterized in that it further comprises means for determining a signal-to-noise ratio for channel estimation.

14. The module of claim 12, further comprising means for determining the symbol energy.

15. The module of claim 12, further comprising means for determining parameters related to time filter coefficients and channel variations across symbols.

16. The module of claim 12, further comprising means for determining a doppler frequency.

17. The module of claim 12, further comprising at least one non-causal filter for generating channel estimates for data demodulation and timing alignment.

18. The module of claim 12, further comprising means for determining the excess delay spread.

19. The module of claim 12 further comprising at least three time domain filter taps for generating channel estimates for data demodulation and timing alignment.

20. The module of claim 19 further comprising setting the tap values to {0.25, 0.5, and 0.25}, respectively.

21. The module of claim 12, further comprising means for optimizing data to pilot energy ratio.

22. The module of claim 12, further comprising a machine-readable medium having stored thereon machine-readable instructions to perform the time filter element or the calibration element.

23. A timing calibration element for a wireless receiver, comprising:

means for receiving a subset of symbols in an OFDM broadcast;

means for filtering the subset of symbols;

means for calibrating symbols within the subset; and

means for calibrating one or more symbols according to a current demodulation of the subset of symbols.

24. A machine-readable medium having stored thereon machine-readable instructions, comprising:

receiving a subset of symbols in a forward link only broadcast;

decoding the subset of symbols;

determining a timing alignment for symbols within the subset; and

the timing of one symbol is adjusted during the current demodulation according to the timing alignment.

25. A machine-readable medium having stored thereon a data structure, comprising:

receiving a subset of symbols in a wireless network; and

at least three non-causal filter tap structures are assigned to adjust the timing differences within the subset of symbols.

26. A wireless communications apparatus, comprising:

a memory comprising elements for determining a time alignment for a received subset of symbols; and

at least one processor associated with a receiver that decodes at least one current symbol during adjustment of timing of another symbol in the subset of symbols.

27. A method of performing time synchronization in a communication environment, comprising:

determining a timing alignment to apply with the operation based on the relatively early or late samples of the OFDM symbol; and

the sample alignment is performed based in part on early or late samples of the OFDM symbol.

28. The method of claim 27, further comprising performing said sample calibration using at least one of the following equations:

early sampling calibration:

and (3) late sampling calibration:

29. the method of claim 28 wherein said early sample calibration further comprises a value of y_a ^(early)(n) performing a circular right shift on the x samples, and then multiplying the first x sample obtained after the circular shift by

30. The method of claim 28 wherein said late sample calibration further comprises a value of y_a ^(late)(n) performing a circular left shift on the x samples, and then multiplying the last x sample obtained after the circular shift by