HK1092988A - Quality packet radio service for a general packet radio system - Google Patents

Description

Quality packet radio service for general packet radio system

Technical Field

The present invention relates to advanced cellular communication networks and, more particularly, to a system for improving the performance of existing cellular communication networks by enhancing the slow general packet radio service medium access procedure to include fast in-session access capability.

Background

A problem with conventional cellular communication networks is that they provide data services to users, such as the internet, as well as providing access to packet switched data communications. This is due to the fact that conventional cellular communication networks have a circuit-switched architecture designed primarily for voice traffic, whose network topology is point-to-point in nature. This example represents a historical view of cellular communications as a wireless equivalent of a traditional wireless telephony communication network for interconnecting a calling party and a called party. A further problem with cellular communication networks is that the need to utilize the limited bandwidth available in a cellular communication network while serving many voice users has prevented the provision of wide bandwidth communication services, such as data, to these users.

The internet has emerged as the primary driver behind the development of new communication network technologies. The number of wireless cellular users worldwide is also rapidly increasing, which leads to an increasing need for both ubiquitous chainless (uninterrupted) communication and constant service availability. The convergence of these two powerful trends has prompted an exponential growth in the mobile access needs for internet applications. However, the internet and other data services require the use of packet switched data networks to achieve the required performance. Conventional 1G (first generation) and 2G (second generation) cellular communication networks have a circuit-switched architecture designed primarily for voice services. This has prompted the development of packet switched network coverage known as 2.5G (generation 2.5) networks implemented over existing 2G cellular communication networks. 2.5G networks form a temporary solution to provide packet switched data services to existing 2G cellular communication networks until 3G (third generation) cellular communication networks providing both circuit switched voice services and packet switched data services are developed and deployed at full scale. Moreover, a 2.5G network would thus provide a legacy platform by which cost-effective upgrades to 3G cellular communication networks can be implemented and deployed.

However, a problem with general packet radio service packet switched network coverage is that it is designed primarily for providing best effort service to bursty data traffic in only a spectrally efficient manner. The design of the gprs packet-switched network coverage is very good for providing this type of service and maintaining the required level of compatibility and interoperability with GSM (global system for mobile communications). However, it is expected that 2.5G systems such as general packet radio service will eventually migrate to full 3G network deployments in an appropriate and cost-effective manner. It is therefore highly desirable to enhance these systems to include more advanced 3G functionality. One of the main attributes of 3G is to enable new service applications. These new traffic applications are supported by defining supported 3G traffic classes with different QoS (quality of service) requirement levels, including some 3G traffic classes with delay requirements much more stringent than the best effort traffic classes. ETSI (european telecommunications standardization institute) UMTS (universal mobile telecommunications system) Phase2+ general packet radio service recommendations include the following service classes:

conversational class-maintaining a session mode with strict low latency and low error rate requirements.

Example (c): voice service

Class of streams-maintaining a temporal relation between the stream information units. Example (c): streaming audio, video

Interaction class-maintaining the request response data transmission mode and the data payload content.

Example (c): web browsing

Background class-maintaining data payload content and best effort traffic requirements. Example (c): background downloading of email messages

Conversational classes have the most stringent low latency requirements followed by streaming and interactive classes. The background class is inherently insensitive to delay. Currently, general packet radio service systems only support background classes and do not have the functionality to instead serve additional traffic classes.

Disclosure of Invention

The above problems are solved and a technical advance is achieved by the present quality packet radio service which provides an enhancement of the RLC/MAC (radio link control/medium access control) layer protocol of the GPRS (general packet radio service) packet switched network overlay implemented on existing 2G (second generation) cellular communication networks in order to support additional service classes.

It is desirable to enhance current GPRS to be able to support the additional stringent delay requirements of additional traffic classes, thereby obtaining a single integrated IP-based network that can provide all traffic classes from session to best effort data. In order to achieve spectral efficiency for all these traffic classes, it is desirable to be able to efficiently multiplex several data sessions with different QoS (quality of service) delay requirements on the same set of channels. Quality packet radio services achieve this by enhancing the GPRS medium access procedure to include fast in-session access capability. To maximize spectral efficiency, all of the quality packet radio services are allocated uplink radio channel resources only if they have valid data to transmit. A new set of common control channels is designed to provide these in-session network access capabilities. These channels support access and control functions similar to GPRS common control channels (e.g., packet random access channels, packet access grant channels) except that they are only used in quality packet radio service to implement in-session access. The structure of these common control channels meets the strict low delay requirements for in-session access and are referred to as fast packet common control channels. Because initial wireless channel access to the mobile subscriber station has been established, a smaller amount of additional information is required to implement the in-session access, thereby allowing these stringent low latency requirements to be met. Specifically, for those services that are allowed to use the in-session access, the allocated uplink channel resources are released by releasing its allocated USF (uplink State flag) and packet data traffic channel during periods of inactive data in the session. However, the mobile subscriber station is allowed to maintain its uplink TFI (temporary flow identifier). Thus, the mobile subscriber station can inform the base station subsystem of its identity and the specific TBF in reference by including the TFI in its in-session channel request message (temporary flow back). The base station subsystem can identify the mobile subscriber station and the referenced session very quickly and allocate the required uplink resources.

Thus, the quality packet radio service requires only software changes and fully preserves the existing network infrastructure and device hardware as it is implemented in the medium access control layer of GPRS.

Drawings

Fig. 1A and 1B show a block diagram of the general architecture of a 2G (second generation) cellular communication network equipped with GPRS (general packet radio service) packet-switched network coverage;

figures 2 to 4 show GPRS slot and frame structures;

figure 5 shows a GPRS protocol stack;

figure 6 shows GPRS uplink multiplexing;

figure 7 shows the message flow for a GPRS access procedure;

fig. 8 shows quality packet radio service fast uplink and downlink control channels;

figure 9 shows a quality packet radio service uplink access procedure for different service classes;

fig. 10 shows the message flow of the access procedure in a quality packet radio service fast session;

fig. 11 shows a general access and allocation period for quality packet radio service; and

figure 12 shows quality packet radio service duplicate Aloha (Aloha) random access.

Detailed Description

In this specification, "third generation cellular communication networks" are used to characterize a network that provides a complete complement of packet-based services to mobile subscriber stations. The technical description of the invention is based on the existing general packet radio service packet coverage on second generation circuit switched cellular communication networks, but it is not meant to limit the application of quality packet radio service to this environment, the architecture is only used to illustrate the concept of quality packet radio service.

Cellular communication network philosophy

As shown in the block diagrams of fig. 1A and 1B, a cellular communication network 100 provides a service for land-based customers, as well as other wireless customers, that connect wireless customers, each having a mobile subscriber station, to a public carrier PSTN (public switched telephone network) 108. In such a network all incoming and outgoing calls are routed through MSCs (mobile switching centers) 106, each MSC 106 being connected to a plurality of RNS (radio network subsystems) 131 and 151, which RNSs 131 and 151 communicate with mobile subscriber stations 101, 101' located within the area covered by the cell site. The mobile subscriber stations 101, 101' are served by RNSs 131-151 and each of the RNSs 131-151 is located in one cell coverage area of a larger service area. Each cell site within the service area is connected to the MSC 106 by a set of communication links. Each cell site contains a set of wireless transmitters and receivers, referred to herein as "base stations," each transmitter-receiver pair being connected to a communication link. Each transmitter-receiver pair operates on a pair of radio frequencies to create a communication channel: one frequency is used for transmitting wireless signals to mobile subscriber stations and the other frequency is used for receiving wireless signals from mobile subscriber stations. The MSC 106, along with an HLR (home location register) 161 and VLR (visitor location register) 162, manages subscriber registration, subscriber authentication, and provision of wireless services such as voice mail, call forwarding, roaming confirmation, etc. The MSC 106 is connected to the GMSC

A (gateway mobile services switching center) 106A and a radio network controller, GMSC 106A, for interconnecting the MSC 106 and the PSTN/IP (internet protocol) network 108. In addition, the radio network controller is connected to the internet via a serving GPRS (general packet radio service) support node 106C through a GGSN (gateway GPRS support node) 106B. The radio network controllers 132, 142, 152 in each cell site radio network subsystem 131-151 control the transmitter-receiver pairs in the radio network subsystem 131-151. The control process in the radio network subsystem also controls the tuning of the mobile subscriber station to the selected radio frequency. In the case of WCDMA (wideband code division multiple access), the system also selects PN codewords to enhance isolation of communications with mobile subscriber stations.

In fig. 1B, the mobile subscriber station 101 communicates with two base stations 133 and 143 at the same time, thereby constituting a soft handover. However, soft handoff is not limited to a maximum of two base stations. During soft handoff, the base stations serving a given call must act in concert so that the commands issued on the RF (radio frequency) channels 111 and 112 are consistent with each other. To achieve this consistency, one of the serving base stations may operate as a primary base station relative to the other serving base stations. Of course, if it is sufficient for the cellular communication network to determine that mobile subscriber station 101 communicates with only one base station, mobile subscriber station 101 may communicate with only one base station.

The control channels available in the system are used to set up the communication connection between the subscriber station 101 and the base station 133. When a call is initiated, the control channel is used to communicate between the mobile subscriber station 101 and the local serving base station 133 involved in the call. The control message locates and identifies the mobile subscriber station 101, determines the dialed number, and identifies the available voice/data communication channel consisting of a pair of radio frequencies (and orthogonal codes of a CDMA (code division multiple access) system) selected by the base station 133 for the communication connection. The radio unit in mobile subscriber station 101 retunes the transmitter-receiver equipment contained therein to use these assigned radio frequencies and orthogonal codes. Once a communication connection is established, control messages are typically sent when a user moves from the current cell to one of the neighboring cells in order to adjust the transmitter power and/or change the transmission channel when a handover of the mobile user station 101 to a neighboring cell is required. Because the amplitude of the signal received by base station 133 varies with the subscriber station transmitter power and the distance to base station 133, the transmitter power of mobile subscriber station 101 is adjusted. Thus, by scaling the transmitter power to correspond to the distance to the base station 133, the received signal amplitude can be maintained within a predetermined range of values to ensure accurate signal reception without interfering with other transmissions in the cell.

Voice communications between mobile subscriber station 101 and other subscriber stations, such as land-line based subscriber station 109, are accomplished by routing communications received from mobile subscriber station 101 through telephone switching center 106 and trunks to PSTN 108, and routing communications in PSTN 108 to local switching operator 125 serving land-line based subscriber station 109. There are a number of MSCs 106 connected to the PSTN 108, thereby enabling both landline based subscriber stations and users of mobile subscriber stations to communicate between selected subscriber stations. Data communication between mobile subscriber station 101 and other data communication systems, such as server 120, is accomplished by routing data communications received from mobile subscriber station 101 through IP (internet protocol) network 107. This architecture represents the current architecture of wireless and wired communication networks.

General packet radio service

As shown in fig. 1A, GPRS (general packet radio service) is a 2G circuit switched cellular communication network that can be used to provide TDMA-based (time division multiple access): GSM (global system for mobile communications) and the 2.5G packet-switched upgraded packet network coverage of IS-136 in north america. In view of the current dominance of serving 70% of cellular subscribers worldwide, the implementation of quality packet radio service in general packet radio service packet network coverage over GSM cellular communication networks is described herein. The extension of general packet radio service to the north american IS-136 cellular communication network IS similar to the implementations disclosed herein, and a description of the implementations IS omitted for the sake of brevity.

The general packet radio service overlay based on circuit switched GSM cellular communication networks provides a separate IP (internet protocol) based packet switched core network. The current evolution of general packet radio service is primarily designed to provide best effort packet service and allow IP based applications such as efficient internet access. However, the problem with the overlap of general packet radio service packet-switched networks is that it is primarily designed to provide best-effort service to bursty data traffic in a particularly efficient manner only.

The focus of the essential packet radio service is to develop enhancements to the RLC/MAC (radio link control/medium access control) layer protocols of the general packet radio service to enable the provision of all other service classes-class 1 to class 4-to be supported by future 3G systems as specified by the ETSI (european telecommunications standardization institute) UMTS (universal mobile telecommunications system) Phase2+ (2+ Phase) general packet radio service recommendation. These classes are:

session-voice, video technology (ultra low latency)

Streaming-multimedia (maintaining internal time relationships)

Interaction-web browsing, gaming (maintaining data integrity)

Background-email (time insensitive, maintain data integrity)

These are called QoS (quality of service), and the frame error rate can range from 10% to 10% frame error rate^-6The error rate varies within a range. For CDMA (code division multiple access) systems, the lower the error rate, the higher the spreading sequences, meaning greater transmit power and bandwidth, used with a correspondingly higher quality of service. From the mobile subscriber station's point of view, the quality of service and the occupied baseband data rate affect the final encoded data rate on a frame-by-frame basis, and in its functionality, the call appears to be circuit-switched and continuous in nature. This is especially true during soft or softer handover.

These quality packet radio service enhancements are only implemented in the MAC/RLC layer of the general packet radio service protocol stack. Therefore, only software upgrades of the mobile station (101) and BSS (base station subsystem) are necessary without requiring hardware improvements. The GSM physical layer, TDMA time slots, and framing structure are maintained, thereby allowing a conventional GSM handset to continue to function in GSM quality packet radio service coverage. Finally, the quality packet radio service core network is compatible with the proposed EDGE (enhanced data rates for GSM service) upgraded to 8-PSK (phase Shift keying) modulation and UMTS W-CDMA for a clear transition to the final 3G system.

Quality packet radio service requires only software modifications and fully maintains the existing network infrastructure and device hardware as it is primarily related to the medium access control layer. Thus, the following description focuses on the corresponding general packet radio service RLC/MAC protocol layer.

General packet radio service network architecture

To understand the operation of the quality packet radio service, the basic architecture of the general packet radio service is described. The following description relates to the parts of the general packet radio service implementing the RLC/MAC protocol layers.

Because the gprs network architecture was originally designed to overlay GSM networks, the gprs network architecture is based primarily on the GSM system concept to achieve maximum interoperability and requires a minimum amount of additional infrastructure and modifications to implement the overlay. A general packet radio service enabled mobile subscriber station (MS) communicates directly with a GSM base station transceiver located in a cell site across the same GSM physical radio link. In a GSM PLMN (public land mobile network), a BSC (base station controller) controls the radio link between a mobile subscriber station and a base station transceiver, which together with its associated BSC is called a BSS (base station subsystem). The BSC is connected to the GSM PLMN backbone through an MSC, which provides switching, routing, and transport of voice, message, and control signals within the GSM PLMN. GMSC (gateway mobile switching centre) is the GSM PLMN to PSTN interface. The communication between the network elements in the GSM PLMN employs signaling No. 7 (SSN 7).

General packet radio service overlay adds two new network router components: SGSN (serving general packet radio service support node) and GGSN (gateway general packet radio service support node). Furthermore, hardware upgrades of the base station subsystem are required. The CCU (channel codec unit) is incorporated into an existing base station transceiver to enable a general packet radio service specific coding scheme. The base station subsystems are connected to their serving GPRS support nodes. A PCUSN (packet control unit support node) unit is added to each base station subsystem to support the frame relay packet data interface between the base station subsystem and the GPRS support node. The GPRS support node serves as an access router for the general packet radio service core network. The GGSN is a gateway router that connects the general packet radio service core network to external IP or x.25/x.75pdn (packet data network). Circuit-switched traffic destined for the PSTN continues to be routed from the base station subsystem to the MSC and then to the PSTN through the GMSC. Packet switched traffic, on the other hand, is routed independently from the base station subsystem, through the serving GPRS support node on the general packet radio service core network, to the GGSN, and to the common packet data network.

In general packet radio service where each cell site provides 200KHz bandwidth channels on different operator frequencies, the GSM physical layer is used. Each 200KHz channel is further formed as a channel by being time division multiplexed into 8 time slots per TDMA (time division multiple access) frame. Each time slot is 0.577 ms in duration and each TDMA frame is 4.615 ms in duration. General packet radio service TDMA time slot and frame structures are shown in fig. 2 to 4. The transmission between different mobile subscriber stations and the base transceiver station can take place in 8 time slots. The mobile subscriber station transmits only during certain time slots and when it is idle its transmitter is powered down to conserve battery power. The transmitter periodic switching of the mobile subscriber station is called burst (burst), and the transmission during a single slot is called burst (burst). The structure of a 156.25 bit normal burst carrying user traffic or network control signaling is as follows:

bit content

1-3 tail position (T)

4-60 coded data (data)

61 stealing flag bit (F)

62-87 training sequence (training)

88 index bit for use (F)

89-145 encode data (data)

146- & ltwbr/& gt148 tail position (T)

148-156.25 GUARD period (No Send) (GUARD)

Fig. 3 shows a detailed structure of a general burst transmission. The tail bit (T3) and GUARD period (GUARD) are GUARD times used to compensate for timing instability and synchronization errors. The stealing flag (F1) indicates whether the burst contains user traffic or network control signaling data and the training sequence allows the receiver to equalize the effect of radio propagation multipath. In a normal burst transmission of 156.25 bit duration, there are only 114 bits of coded data bits in total. Only these coded data bits carry user traffic or network control signalling data. The amount of coding depends on the level of channel error correction coding selected for application.

General packet radio service protocol architecture

While the general packet radio service protocol architecture supports IP and x.25 based (and potentially other packet data protocols) applications, general packet radio service specific protocols are employed within the general packet radio service core network. Figure 5 gives a diagram of the current version of the general packet radio service protocol stack, the devices in which the general packet radio service protocol stack is located, and the interfaces between the devices. The structure includes the following well-known components:

MS mobile subscriber station

BSS base station subsystem

SGSN service general packet radio service support node

GGSN gateway general packet radio service support node

GTP GPRS tunneling protocol

SNDCP subnet dependent convergence protocol

BSSGP base station GPRS protocol

LLC logical link channel

RLC radio link control

MAC medium access channel

Interface between Gb SGSN and base station subsystem

Interface between Um mobile subscriber station and GPRS fixed network part for providing packet network services

Interface between Gn SGSN and GGSN

Interface between Gi GPRS network and other IP or X.25 networks

GTP (GPRS tunneling protocol) is used to transfer packets between two GPRS support nodes (e.g., SGSN, GGSN). Tunneling protocols encapsulate IP or x.25 packets into GTP PDUs (packet data units). The GTP PDUs are routed over the IP-based GPRS backbone network using TCP (transmission control protocol) based on x.25 applications or UDP (user data protocol) based on IP applications. This process is called GPRS tunneling. In the transfer of IP or x.25 packets between a mobile subscriber station and its serving GPRS support node, the general packet radio service uses a set of different network protocols, namely SNDCP (subnetwork dependent convergence protocol) and LLC (logical link control) layers. SNDCP is used to map network protocol layer features to specific features of the underlying network. The LLC provides a secure logical conduit between the GPRS support node and each mobile subscriber station and performs tasks such as decoding, flow control, and error control. LLC is used by SNDCP to transfer network layer PDUs between a mobile subscriber station and its serving GPRS support node. LLC PDUs are transmitted over a radio link using services provided by the RLC/MAC protocol layer. The RLC/MAC protocol layers reside both within the mobile subscriber station and within the base station subsystem. LLC PDU transfers between a plurality of mobile subscriber stations and a core general packet radio service network use a shared radio medium. The LLC layer is responsible for:

1. the LLC PDU is fragmented and reassembled.

2. Providing a link level including means for recovering uncorrectable data block transmission errors

An option for ARQ (automatic repeat request) procedure.

The MAC layer operates between the mobile subscriber station and the base station subsystem and is responsible for:

1. signaling of procedures related to wireless medium access control

2. Contention resolution between access attempts, arbitration between multiple service requests from different mobile subscriber stations, and media allocation in response to service requests are performed.

The RLC/MAC layer performance largely determines the multiplexing efficiency and access delay of the radio interface based general packet radio service application.

General packet radio service frame and data structure

General packet radio service uses the same physical time slot and TDMA frame structure as GSM. The basic PDU between the mobile subscriber station and the base station subsystem is called an RLC block (also called RLC/MAC block). The LLC PDU is segmented into an appropriate number of RLC blocks. The structure of each RLC block is such that it can be channel coded and transmitted in interleaved form within 4 time slots in 4 consecutive TDMA frames. Therefore, the logical channel resource allocation unit in the RLC/MAC layer is one RLC block, and the transmission unit in its physical layer uses a normal burst within 4 slots. As described above, in each normal burst, there are 114 bits of channel encoded data that are transmitted. One RLC block transmits 4 × 114-456 bit coded data bits. The time for transmitting 4 TDMA frames is 18.46 ms, 4 × 4.615 ms. However, in GSM, a multi-frame structure consisting of 52 TDMA frames is used, with every 13 th frame being used for a purpose other than data transmission (e.g. channel measurement). Therefore, only 48 of the 52 frames are used for data transmission, and the average time for transmitting 4 TDMA data frames is (4 × 4.615) × (52/48) ms, which is 20 ms. This results in a maximum radio interface data throughput of 456 coded bits per 20 ms (22.8 Kbps per channel) (note here that the time guard, training sequence, and control bit overhead in a normal burst are not considered here). If each RLC block is consideredThe RLC header, MAC protocol overhead bits, other control bits, and channel coding error protection bits, the actual information throughput rate is much lower. In the general packet radio service standard, four different channel coding schemes CS-1 to CS-4 are defined with the following information throughput rates:

code	Encoding rate	Throughput (Kbps)
			CS-1	1/2	8

CS-2CS-3CS-4

2/33/41

1214.420

Since there is no channel coding in CS-4 (since the coding rate is equal to 1), this represents the maximum possible information throughput rate of 20Kbps per channel. However, some procedures are included in general packet radio service to allow a mobile subscriber station to utilize several channels simultaneously, thereby improving its information throughput rate. While up to 8 channels can be allocated to one mobile subscriber station.

Each set of 4 TDMA frames containing 8 RLC blocks is called a logical frame. Fig. 2 shows the structure of a logical frame in units of a basic TDMA frame. Although there are 8 time slots in each TDMA frame, there are 32 time slots in each logical frame. Since the channel resource allocation unit in the RLC/MAC layer is an RLC block of a normal burst within 4 slots, a channel supporting consecutive logical frames of the 4 slots is called a PDCH (packet data channel). The PDCH may be in the uplink direction (mobile subscriber station to base station transceiver transmissions) or the downlink direction (base station transceiver to mobile subscriber station transmissions). In general packet radio service, PDCH assignment is a simplex channel so that an uplink (uplink) PDCH can be used by one mobile subscriber station, while a downlink (downlink) PDCH occupying the same time slot can be used by a different mobile subscriber station. PDCHs are mapped to various different logical channels for providing specific data transfer functions. PDCHs, which are used only for conveying user data traffic, are referred to as packet data traffic channels, which may be uplink channels or downlink channels, to name a few examples useful for the following discussion. A PAGCH (packet access grant channel) is a downlink channel used by a base transceiver station to transmit a resource allocation message to a mobile subscriber station. The PACCH (packet associated control channel), which may be uplink or downlink, conveys network control signaling information and can also be used to convey resource allocation messages to mobile subscriber stations.

User mobility can cause mobile subscriber stations to be located in cells at different locations and distances from the base transceiver station, resulting in different transmission propagation delays. In GSM and GPRS networks, therefore, precise timing synchronization must be obtained and maintained at the mobile subscriber station so that the normal bursts in different time slots do not overlap. However, in some cases, such as before a wireless link is established, timing synchronization does not exist and reliable normal burst transmission may not be possible. To avoid this problem, the mobile subscriber station uses a shorter burst, called a random access burst, which allows the base station transceiver to measure the propagation delay to the mobile subscriber station and allows the base station transceiver to then control the mobile subscriber station timing by sending timing advance information to the mobile subscriber station. The random access burst is short enough that no overlap with other bursts occurs within the maximum possible propagation delay difference in a cell of 35 km radius (which is the maximum allowed cell size in GSM). Thus, when timing information is not available, the mobile subscriber station initiates a network channel access request using a random access burst having the following structure:

bit content

1-8 tail position (tail)

9-49 Sync sequence (Sync)

50-85 encode data (data)

86-88 tail position (T)

89-156.25 GUARD period (no transmission) (GUARD)

Fig. 4 shows a detailed structure of random burst transmission. The guard period is long enough to accommodate the propagation delay caused by distances up to 75 kilometers, allowing for a cell radius of 35 kilometers. A long synchronization sequence is set to allow for more accurate timing measurements. Considerable error protection is included so that 36 coded bits carry at most 8 or 11 bits of information. A logical uplink channel, called PRACH (packet random access channel), is allocated a TDMA time slot for use by mobile subscriber stations to initiate network channel access and resource allocation requests. Normally, one time slot is allocated for PRACH every TDMA frame. There are 4 network channel access start opportunities in each logical frame. This is illustrated in fig. 2, where slot #1 in each TDMA frame is allocated as a PRACH.

RLC/MAC multiplexing

The RLC/MAC layer is designed to support burst traffic best effort transport services in a particularly efficient manner. Multiple data streams may be supported on the same packet data traffic channel and a given data stream may be supported using multiple packet data traffic channels. Data transfer in general packet radio service is implemented using an entity called TBF (temporary back flow). A TBF is a virtual connection that supports unidirectional transfer of LLC PDUs on a packet data physical channel between a mobile subscriber station and a base station subsystem. The maintenance time of the virtual connection is the duration of the data transfer and the virtual connection consists of a number of RLC blocks. The TBF may be open or closed. A closed TBF limits the data to be transferred to the amount negotiated between the mobile subscriber station and the base station subsystem during initial network channel access. In an open TBF, any amount of data may be transferred. Each TBF is identified by a TFI (temporary flow identifier). For the uplink, the TFI is 7 bits long, and for the downlink, the TFI is 5 bits long. In each direction, the TFI assigned by the base station subsystem is unique, so RLC blocks destined for different mobile subscriber stations are distinguished by their additional TFI embedded in the RLC block header. After the data transfer in the session is completed, the TBF is terminated and its TFI is released.

Downlink multiplexing of multiple data streams on the same packet data traffic channel is achieved by assigning a unique TFI for each data transmission. Each mobile subscriber station listens to its set of allocated downlink packet data traffic channels and receives only RLC blocks with its TFI. The base station subsystem is thus able to communicate with mobile subscriber stations on any of its assigned packet data traffic channels and to multiplex several TBFs intended for different mobile subscriber stations on the same packet data traffic channel.

Uplink multiplexing is achieved by allocating a set of channels for each data transmission and a unique USF (uplink state flag) for each of these channels. The USF is 3 bits long, allowing multiplexing up to 7 different data transfers on one channel (USF 111 reserved by the network). The base station subsystem uses a centralized in-band polling scheme to poll the desired mobile subscriber stations. This is achieved by setting the USF in the MAC header of the RLC block transmitted on the corresponding downlink channel to an appropriate value for identifying the particular data transfer. Thus, the mobile subscriber station listens to all downlink channels paired with the uplink channel allocated thereto. If its USF appears on a certain downlink channel, the mobile subscriber station transmits its data using the corresponding uplink channel in the next logical frame. Fig. 6 shows the operation of the procedure by way of example. In this example, channel 6 of each uplink logical frame is allocated to both mobile subscriber station 1 and mobile subscriber station 2. Accordingly, after detecting its USF in channel 6 of downlink frame 1, mobile subscriber station 1 can use the corresponding uplink channel (channel 6 of uplink logical frame) in frame 2. Meanwhile, the USF of the mobile subscriber station 2 appears in the channel 6 of the downlink frame 2. Mobile subscriber station 2 is therefore now allowed to transmit on the corresponding uplink channel in frame 3. The USF of mobile subscriber station 1 occurs immediately in channel 6 in both downlink frames 3 and 4, thereby allowing mobile subscriber station 1 to transmit on the corresponding uplink channel in frames 4 and 5, respectively. The data carried by the downlink channel 6 in each frame may be destined for any mobile subscriber station and its recipient is identified by the TFI header in the RLC data block. This procedure enables multiplexing of different users on the same uplink physical channel. Thus, even though the downlink RLC database may be destined for a certain mobile subscriber station, the USF carried in the MAC header of that data block may be targeted to a different mobile subscriber station.

Medium access program

General packet radio service allows two types of data transfer access procedures: one step or two steps. Fig. 7 shows these two procedures.

One-step procedure

The mobile subscriber station transmits a packet channel request on the PRACH. As previously mentioned, the random access burst occupies only 1 TDMA time slot. General packet radio service utilizes slotted ALOHA-based random access procedures to achieve contention resolution for PRACH. The 8 or 11 bit encryption information field in the 36 bit coded data bits in the random access burst carries only a limited amount of information, i.e. the access cause: whether it is a one-step access, a two-step access, or a paging response; mobile subscriber station class and radio priority; and the number of blocks to be transmitted (relative to the page response only). The identity of the mobile subscriber station or connection and the amount of data to be transmitted (other than the page response) are not included in the channel request and are not known to the network at this time.

After receiving the packet channel request, the base station subsystem replies with a packet uplink assignment message on the PAGCH paired with the PRACH used. The message contains the resource allocation, including carrier frequency, TFI, USF and other parameters, with respect to the mobile subscriber station so that the mobile subscriber station can transmit on the allocated uplink packet data traffic channel. At this point, however, the network is unaware of the mobile subscriber station identity and the requested service.

The base station subsystem transmits the USF on the downlink packet data traffic channel paired with the assigned uplink packet data traffic channel in the next logical frame.

The mobile subscriber station listens to its USF and starts data transmission on the uplink packet data traffic channel allocated in the next logical frame. The transmitted RLC block includes an extension header with the requested service type and the mobile identity identified by its TLLI (temporary logical link identifier).

When the network successfully decodes the TLLI, the network sends an acknowledgement to the mobile subscriber station with an uplink ACK/NAK message on the PACCH. Contention resolution is done on the network side; and after the mobile subscriber station successfully receives the acknowledgement, contention resolution is also completed at the mobile subscriber station side.

Data transfer from the mobile subscriber station to the base station subsystem may continue while the mobile subscriber station listens to its USF to begin data transfer on the assigned packet data traffic channel.

Two-step procedure

The mobile subscriber station transmits a packet channel request in the same manner as the one-step procedure.

After receiving the packet channel request, the base station subsystem replies with an uplink assignment message on the PAGCH. The allocation is a single block on the uplink PACCH. The message contains the resource allocation, including carrier frequency, TFI, timeslot, and other parameters, relative to the mobile subscriber station so that the mobile subscriber station can transmit on the allocated PACCH.

The mobile subscriber station transmits a detailed packet resource request message on the allocated uplink PACCH. The resource request includes the mobile TLLI and the requested service details.

The base station subsystem then allocates the required resources to the mobile subscriber station in response to the request using an uplink packet allocation message sent on the PACCH. The message includes the carrier frequency, TFI and USF parameters so that the mobile subscriber station can transmit on the assigned uplink packet data traffic channel.

The base station subsystem transmits the USF on the downlink packet data traffic channel paired with the assigned uplink packet data traffic channel in the next logical frame.

The mobile subscriber station listens to its USF and starts data transmission on the uplink packet data traffic channel allocated in the next logical frame.

The choice of which of these two procedures to use is left to the GPRS system operator. The essential difference is that in a one-step procedure, uplink data transfer starts simultaneously with service negotiation and mobile authentication; whereas in the two-step procedure, uplink data transfer is started only after mobile authentication and service negotiation is completed. Thus, if the requested service negotiation is acceptable to the network and mobile subscriber station applications, the one-step procedure may be slightly faster than the two-step procedure (a minimum of 3 to 4 logical frame times for the one-step procedure compared to 4 to 5 logical frame times relative to the two-step procedure without contention). However, because mobile authentication cannot be achieved before data transfer in a one-step procedure, the system operator considers it insecure and favors a two-step procedure in network deployment. Also, because a two-step procedure is currently used in GSM systems, compatibility reasons are also desirable.

Quality packet radio service-general packet radio service RLC/MAC protocol layer enhancement

Defects of general packet radio service medium access program

In GPRS systems, time slots can be shared between GSM circuit switched voice services and GPRS packet switched data services to implement a total capacity on demand system (on demand system). GPRS must therefore have a high level of compatibility and interoperability with GSM. GPRS must operate within the physical constraints imposed by GSM cellular networks and with the same physical layer transport channels. Furthermore, for GSM operators who prefer GPRS procedures over similar symmetric GSM procedures, ease of maintenance and operation is very important. This level of compatibility and interoperability extends to the use of GSM timeslots, framing, random burst, and the normal burst structure in GPRS. These constraints limit the multiple access efficiency of GPRS medium access procedures to a large extent. It also imposes corresponding constraints on the types of enhancements that can be implemented to improve performance or provide a greater variety of support services.

The GPRS system is primarily designed to provide best effort services to bursty data services in a spectrum efficient manner only. The design of the GPRS system is very good for providing this type of service and maintaining the required level of compatibility and interoperability with GSM. However, it is expected that 2.5G systems such as GPRS will eventually migrate to full 3G network deployments in an appropriate and cost-effective manner. It is therefore highly desirable to enhance these systems to include more advanced 3G functionality. One of the main attributes of 3G is to enable new service applications. These new traffic applications are supported by defining supported 3G traffic classes with different QoS requirement levels, including some 3G traffic classes with much more stringent delay requirements than the best effort traffic classes. The ETSI UMTS Phase2+ GPRS recommendations include the following service classes:

conversational class-maintaining conversational patterns with strict low latency and low error rate requirements. Example (c): voice service

Stream class-the temporal relationship between stream information units is maintained. Example (c): streaming audio, video

Interaction class-hold request response data transfer mode and data payload content. Example (c): web browsing

Background class-keeping data payload content and best effort traffic requirements. Example (c): background downloading of email messages

Conversational classes have the most stringent low latency requirements followed by streaming and interactive classes. The background class is inherently insensitive to delay.

The current GPRS system is best suited for best effort background traffic classes with very relaxed delay requirements. During idle periods in bursty data flows, the data transfer session is completely terminated (TBF is terminated and TFI is released). After the idle period ends with the arrival of additional data, the slow medium access procedure described above must be re-used to establish the data transfer. While secure and reliable data transfer is achieved, other traffic classes with more stringent QoS delay requirements cannot be efficiently provided in current GPRS systems. For example, consider a conversational class packet voice service. It is well known that voice activity detection combined with statistical multiplexing can greatly improve spectral efficiency. Therefore, it is preferable that the voice user releases the channel during silence and resumes access only at the beginning of the next talk spurt (talk spurt). The unused remaining capacity in these silent periods can be used to multiplex delay insensitive additional traffic (e.g., best effort data) with voice users, thereby improving overall network spectral efficiency. The current GPRS system cannot support this procedure because it would require the packet voice user to terminate their TBF during silence and release their TFI. The re-establishment of the data transfer connection using the current GPRS slow medium access procedure (new TBF and TFI) will not meet the voice communication QoS latency requirements.

It would be desirable to enhance the current GPRS system to be able to support these otherwise stringent delay requirement traffic classes to obtain a single integrated IP-based network that can provide all traffic from sessions to best effort data. Lower operating costs can also be achieved if all services are ultimately ported to such a platform. In order to achieve a particularly high efficiency for all these traffic classes, it is necessary to be able to multiplex several data sessions with different QoS delay requirements on the same set of channels. The key requirement is to enhance the slow GPRS medium access procedure to include fast in-session access capability. This is the goal of quality packet radio service.

Quality packet radio service fast session medium access procedure

To maximize spectral efficiency, all of the quality packet radio services are allocated uplink radio channel resources only if they have valid data to transmit. For example, in a packet voice session, the uplink channel is only allocated during a talk spurt. In all services, the mobile subscriber station must release the uplink when its session is in an inactive state. For services with strict low delay requirements, the mobile subscriber station may use an in-session network access procedure to request uplink channel resources when the session becomes active again due to data to be sent. The goal of the quality packet radio service RLC/MAC protocol design is to support network access in this session by providing the following capabilities:

fast uplink access during ongoing sessions

Fast resource allocation for uplink and downlink

The following new set of control channels are utilized to provide these capabilities to efficiently implement the in-session access procedure.

Fast packet common control channel

The new set of common control channels shown in fig. 8 is designed to provide these in-session network access capabilities. These channels support similar access and control functions as GPRS common control channels (e.g., PRACH, PAGCH), except that they are only used in quality packet radio service to implement in-session access. The structure of these common control channels meets the strict low delay requirements for in-session access and are referred to herein as "fast packet common control channels". Because the initial access between the mobile subscriber station and the cellular communication network has been established, a smaller amount of additional information is required to implement the in-session access, thereby allowing these stringent low latency requirements to be met. Specifically, for those services that are allowed to use the in-session access, the allocated uplink channel resources are released by releasing its allocated USF and packet data traffic channel during periods of inactive data in the session. However, the mobile subscriber station is allowed to maintain its uplink TFI. Thus, the mobile subscriber station can inform the base station subsystem of its identity and the specific TBF referred to by including the TFI in its in-session channel request message in order to identify the mobile subscriber station and the referred session very quickly, thereby enabling the base station subsystem to allocate the required uplink resources.

In particular, the following fast packet common control channels are implemented in quality packet radio service:

uplink F-PACH (fast packet access channel)-use as F-PRACH (fast packet random access channel) or F-PDACH (fast packet dedicated access channel).

Downlink F-PCCH (faster packet control channel)Use as F-PAGCH (fast packet access grant channel) or F-PPCH (fast packet poll channel).

In a quality packet radio service system, these channels may be located on specific TDMA time slots of certain selected carrier frequencies. Fig. 8 shows the structure of these channels implemented in the first time slot of each TDMA frame. Each uplink F-PACH has a corresponding downlink F-PCCH paired with it.

Fast packet access channel

The structure of the F-PACH (fast packet access channel) is similar to the PRACH in GPRS. The messages are sent in individual bursts and are not interleaved across bursts within several TDMA frames. The difference between the two is that F-PACH is only used for in-session access and never for initial network access. The F-PRACH and F-PDACH on the same physical channel may be time multiplexed depending on the traffic requirements in each cell site. The specific characteristics of these channels are as follows:

the F-PRACH is used for contention resolution together with the replicated Aloha random access protocol described below. The information in the message sent on the F-PRACH includes the TFI and other identifying information of the requested service. For a given TFI, the base station subsystem already has the information required to determine its requirements, such as the required resources and allocation priorities, in order to make the traffic data channel allocation.

F-PDACH is used for contention-free fast dedicated access. It is therefore useful for, and reserved for, future defined services that do not allow any QoS delay variability, for example:

a. downlink channel state measurement-allows the base station subsystem to allocate a greater number of downlink time slots to higher quality channels, thereby improving system throughput through dynamic bandwidth allocation.

b. In future EDGE (improved data rate GSM service) systems, dynamic data rate allocation can also be implemented by allowing advanced PSK (phase shift keying) modulation to be performed on higher quality channel states, thereby increasing the availability of peak data rates.

c. A pilot tracking signal for implementing a smart antenna to improve system throughput.

d. Timing information for maintaining synchronization during inactive data in a session. Fast packet control channel

The F-PCCH (fast packet control channel) serves two main functions: transmitting an access permission message to a mobile subscriber station requesting network access in a session; and transmitting a polling message to the specific mobile subscriber station. The F-PAGCH and F-PPCH on the same physical channel may be time multiplexed depending on the traffic requirements in each cell site. The specific characteristics of these channels are as follows:

the F-PAGCH is used to send channel assignment messages in response to access requests received on the paired P-PACH. The assignment message specifies the slot number, USF, packet data traffic channel, and other parameters, such as the access probability parameter for the duplicate Aloha random access procedure described below.

The F-PPCH is used to poll different mobile subscriber stations for access queries and measurement reports when needed.

Fast access RLC/MAC protocol in session

The RLC/MAC protocol for in-session uplink access in quality packet radio service utilizes the F-PCCH described above, as shown in fig. 10. The following assumptions are made to obtain a version of the RLC/MAC protocol for providing in-session access that meets all QoS delay requirements of the supported traffic classes:

4 service classes of conversational, streaming, interactive and background classes defined in the ETSI UMTS Phase2+ GPRS proposal will be supported.

2. For all services, radio channel resources are allocated only if they have valid data to transmit. When the mobile subscriber station is in an inactive state, the mobile subscriber station must release uplink radio resources during its session.

3. Only the F-PRACH control channel is used to initiate fast in-session access. The channel assignment is sent by the base station subsystem to the mobile subscriber station using the F-PAGCH control channel paired with the F-PRACH. The F-PDACH control channel is only used to transfer low bit rate measurement data from the mobile subscriber station to the base station subsystem.

Fig. 9 shows the overall access procedure:

at the start of a new data session, the mobile subscriber station starts processing by sending a packet channel request message on the PRACH in step 901, using the normal GPRS initial access procedure 91 (one-step procedure or two-step procedure). During this initial access procedure and channel resource allocation performed by the base station subsystem, the mobile subscriber station sets up a TBF and obtains a TFI in step 903 and obtains a USF and a packet data traffic channel in step 904. For the conversational, streaming and interactive traffic classes, an open TBF is established, whereas for the background traffic class, only a closed TBF is allowed. Furthermore, in order to prioritize random access contention resolution in sessions, flows and sessions of interactive traffic classes, class specific access probabilities are assigned for the replicated Aloha (Aloha) random access procedure in step 905, as described below.

At step 906, the RLC layer enables link-level retransmission of background (process 92), interactive and streaming traffic data (process 93) during RLC block data transfer from the mobile subscriber station to the base station subsystem. Link-level retransmissions of session traffic data are prohibited to minimize transmission delay for the traffic class.

At the end of each valid data burst period during the session (no data is available for transmission until the next valid data burst period):

a. for the background traffic class, the mobile subscriber station releases its TFI, USF and packet data traffic channel in step 907.

b. For the conversational, streaming and interactive traffic classes, the mobile subscriber station maintains its TFI but releases its USF and packet data traffic channel in step 911.

At the beginning of the next valid data burst period in the session:

a. for the background traffic class, the mobile subscriber station goes through the entire GPRS initial access procedure in step 909 (starting by sending a channel request message on the PRACH in step 908) to obtain a new uplink channel assignment: TFI, USF and packet data traffic channel.

b. For the conversational, streaming and interactive traffic classes, the mobile subscriber station starts the fast-conversational access procedure by sending a packet channel request message containing its TFI to the base station subsystem together with its assigned traffic class access probability parameter in the replicated Aloha random access procedure using the F-PRACH control channel in step 912. In step 913, the mobile subscriber station obtains a new USF and packet data traffic channel assignment from the base station subsystem using the assignment message received on the F-PAGCH control channel paired with the F-PRACH.

The base station subsystem may request the mobile subscriber station to send information like measurement reports by sending these requests on the fast packet polling channel. The mobile subscriber station may send a response back to the base station subsystem on the assigned packet data traffic channel or the assigned F-PDACH, steps 910, 914, respectively.

Duplicating Aloha random access protocol

Fig. 11 shows the general access and acknowledgement periods for a procedure in a session. Each fast uplink access request on the uplink F-PRACH control channel occupies one TDMA time slot. The downlink F-PAGCH assignment channels paired with each F-PRACH control channel in a given logical frame are multiplexed in the next downlink logical frame. Thus, a full access and allocation period takes two logical frames or 40 millisecond duration. If the contention is successful in the first access channel request, the minimum delay incurred to complete the in-session access procedure is 40 milliseconds. Each unsuccessful contention period increases the delay by 40 milliseconds. Thus, for very fast access, especially for traffic classes with more stringent low delay requirements, a higher probability of contention success is desired. In order to increase the contention success probability of the F-PRACH control channel random access, a duplicate Aloha protocol is proposed. The protocol operates in the following manner:

1. whenever a mobile subscriber station has a packet channel request message ready to initiate access in a session, the mobile subscriber station randomly selects k time slots from n consecutive F-PRACH time slots and transmits the same request message burst on each of the k time slots. Here, k/n is the traffic class specific access probability assigned by the base station subsystem to the mobile subscriber station in the initial access session access procedure described earlier.

2. After sending the channel request, the mobile subscriber station listens for an uplink assignment message from the base station subsystem on the F-PAGCH control channel paired with the k F-PRACH channels it uses to send the channel request.

3. Upon correctly receiving the access request message, the base station subsystem ignores any duplicate requests from the same mobile subscriber station. The base station subsystem transmits the uplink allocation message to the mobile subscriber station on the F-PAGCH paired with the first F-PRACH control channel that correctly received the access request message.

4. If none of the k request message transmissions from the mobile subscriber station were successfully received (i.e., the mobile subscriber station did not receive any assignment message in the expected F-PAGCH slot), step 1 is repeated again. Step 1 is allowed to repeat only up to K times before the access attempt is aborted.

The access probability parameter (k, n) may be selected to meet the access delay requirements of different QoS traffic classes. One possible choice of these parameters may be:

1. conversation class (k, n) ═ 2, 8)

2. Class (k, n) ═ 1, 8)

3. Interactive class (k, n) ═ 1, 16)

That is, the access attempt in the conversational traffic class session randomly selects the 2 access bursts in the next two logical frames, while the access attempt in the interactive traffic class session randomly selects the 1 access burst in the next 4 logical frames. Fig. 12 shows a duplicate Aloha access and allocation period in which the mobile subscriber station 1 uses the access probability parameter (2, 4) and the mobile subscriber station 2 uses the access probability parameter (1, 4).

Access delay performance analysis

Duplicate Aloha (Aloha) is designed to achieve access delays better than standard Aloha (Aloha) random access contention resolution protocols. Of course, by sending multiple copies of the request message, the contention success probability without any radio transmission errors may be increased. Furthermore, it should be noted that hostile mobile wireless transmission environments may result in large transmission error rates, thereby requiring retransmission of incorrectly received channel request messages caused by these errors. Replication of Aloha (Aloha) also has the ability to mitigate these effects. Our mathematical analysis shows that the average access delay performance of the k 2 replica Aloha protocol is 20% to 40% lower than the corresponding average delay of the standard Aloha protocol. This analysis also suggests that using more than k-2 transmitted copies in duplicate Aloha (Aloha) does not yield significant additional performance gains. This is because a larger number of transmission copies increases the amount of contention traffic, thereby offsetting the advantage obtained by redundant transmission. Therefore, it seems that k-2 duplicate Aloha random access should be adopted in quality packet radio service to provide more stringent low QoS latency services.

Summary of the invention

The quality packet radio service enhances the slow GPRS media access procedure to include fast in-session access capability. For all of the quality packet radio services, uplink radio channel resources are allocated only if they have valid data to transmit, and a new set of common control channels is designed to provide these in-session network access capabilities. These channels support similar access and control functions as GPRS common control channels (e.g., PRACH, PAGCH), except that they are only used in quality packet radio service to implement in-session access.

Claims (26)

1. A method for providing low latency network access to a mobile subscriber station operating in a cellular communication network providing packet switched data communications, the method comprising:

allocating a radio channel to serve a mobile subscriber station in response to a service request received from the mobile subscriber station; and

allocating radio channel resources on said allocated radio channel to a mobile subscriber station in response to a service request received from said mobile subscriber station indicating valid data to be exchanged with said mobile subscriber station.

2. The method of claim 1, wherein the allocated radio channel has an uplink channel and a downlink channel, the allocating radio channel resources comprising:

releasing the uplink channel in response to a communication session performed in the mobile subscriber station entering an inactive state.

3. The method of claim 2, wherein the step of releasing radio channel resources comprises:

the assigned uplink channel uplink state flag and packet data traffic channel are released.

4. The method of claim 3, wherein the step of releasing radio channel resources further comprises:

maintaining the mobile subscriber station uplink temporary flow identifier.

5. The method of claim 2, wherein the step of allocating radio channel resources further comprises:

reallocating radio channel resources on the allocated radio channel to the mobile subscriber station in response to the communication session becoming active again as a result of having data to transmit.

6. The method of claim 5, wherein:

the step of allocating a radio channel comprises:

implementing a virtual connection supporting unidirectional transport of logical link control packet data units on a packet data physical channel between the mobile subscriber station and a base station subsystem in the cellular communication network,

maintaining a temporary flow identifier indicating an active state of the virtual connection; and

the step of allocating radio channel resources further comprises:

maintaining at least one uplink state flag indicating an identity of a particular data transfer performed in the communication session.

7. The system of claim 6, wherein the mobile subscriber station transmits data indicating its identity and the particular virtual connection referenced by including the temporary flow identifier in its in-session channel request message, the step of allocating radio channel resources enabling the base station subsystem to allocate uplink resources to serve the virtual connection.

8. The system of claim 1, wherein the allocated radio channel has an uplink channel and a downlink channel, the step of allocating radio channel resources comprising:

transmitting messages between said mobile subscriber station and said cellular communication network over a fast packet access channel; and

an access grant message is sent to the mobile subscriber station requesting network access in the session over a fast packet control channel, and a polling message is sent to the mobile subscriber station.

9. The system of claim 8, wherein the step of transmitting over a fast packet access channel transmits the message in individual bursts of a single data frame.

10. The system of claim 8, wherein the transmitting over the fast packet control channel transmits the message in individual bursts of a single data frame.

11. The system of claim 8, wherein the step of allocating radio channel resources multiplexes multiple data streams with different quality of service latency requirements on multiple data traffic channels.

12. The system of claim 8, wherein the step of transmitting over a fast packet control channel transmits control channel measurements and timing measurements to the cellular communication network, the method further comprising:

performing at least one of the following channel management functions: dynamically allocating traffic slots according to channel quality conditions to improve general packet radio service network throughput through dynamic bandwidth allocation; for mobile data networks employing a set of different multi-level modulation schemes, dynamically varying the peak transmission rate according to channel quality conditions; timing information is transmitted during inactivity of the communication session.

13. The system of claim 8, wherein the step of allocating radio channel resources transmits multiple copies of each message in order to reduce average delay caused by both traffic contention and mobile radio channel fading degradation.

14. A system for providing low latency network access to a mobile subscriber station operating in a cellular communication network providing packet switched data communications, the system comprising:

channel allocation means for allocating a radio channel to serve a mobile subscriber station in response to a service request received from the mobile subscriber station; and

fast packet channel allocating means for allocating radio channel resources on said allocated radio channel to a mobile subscriber station in response to a service request received from said mobile subscriber station indicating valid data to be exchanged with said mobile subscriber station.

15. The system of claim 14, wherein the allocated radio channel has an uplink channel and a downlink channel, the fast packet channel allocating means comprising:

channel releasing means for releasing the uplink channel in response to a communication session performed in the mobile subscriber station entering an inactive state.

16. The system of claim 15, wherein the channel releasing means comprises:

means for releasing the assigned uplink channel uplink state flag and packet data traffic channel.

17. The system of claim 16, wherein the channel releasing means further comprises:

means for maintaining the mobile subscriber station uplink temporary flow identifier.

18. The system of claim 15, wherein the fast packet channel assignment means further comprises:

channel reassignment means for assigning radio channel resources on said assigned radio channel to said mobile subscriber station in response to said communication session again becoming active due to the presence of data to be transmitted.

19. The system of claim 18, wherein:

the channel allocation device comprises:

temporary flow identifier means for implementing a virtual connection supporting unidirectional transport of logical link control packet data units on a packet data physical channel between the mobile subscriber station and a base station subsystem in the cellular communication network,

data flow management means for maintaining a temporary flow identifier indicating an active state of the virtual connection; and

the fast packet access channel apparatus further comprises:

means for maintaining at least one uplink state flag indicating an identity of a particular data transfer performed in the communication session.

20. The system of claim 19 wherein the mobile subscriber station sends data indicating its identity and the particular virtual connection referenced to the fast packet channel assignment means by including the temporary flow identifier in its in-session channel request message, the fast packet access channel assignment means enabling the base station subsystem to assign uplink resources to serve the virtual connection.

21. The system of claim 14, wherein the allocated radio channel has an uplink channel and a downlink channel, the fast packet channel allocating means comprising:

fast packet access channel means for transmitting messages between said mobile subscriber station and said cellular communication network; and

fast packet control channel means for transmitting an access grant message to a mobile subscriber station requesting network access in a session and transmitting a polling message to the mobile subscriber station.

22. The system of claim 21, wherein the fast packet access channel means transmits the message in individual bursts of a single data frame.

23. The system of claim 21 wherein the fast packet control channel means transmits the message in individual bursts of a single data frame.

24. The system of claim 21, wherein the fast packet channel assignment device multiplexes multiple data streams having different quality of service latency requirements on multiple data traffic channels.

25. The system of claim 21, wherein the fast packet control channel means transmits control channel measurements and timing measurements to the cellular communication network to enable the cellular communication network further comprises:

means for performing at least one of the following channel management functions: dynamically allocating traffic slots according to channel quality conditions to improve general packet radio service network throughput through dynamic bandwidth allocation; for mobile data networks employing a set of different multi-level modulation schemes, dynamically varying the peak transmission rate according to channel quality conditions; timing information is transmitted.

26. The system of claim 21 wherein the fast packet channel assignment means transmits multiple copies of each message in order to reduce average delay caused by both traffic contention and mobile radio channel fading degradation.