Overview of GPRS

Overview
GPRS represents an evolution of the GSM standard, allowing data transmission in packet mode and providing higher throughputs as compared with the circuit-switched mode. This evolution is usually presented under the designation of 2.5G to point out that it is a transition technology between 2G and 3G.

The GPRS network architecture reuses the GSM network nodes such as MSC/VLR, HLR, and BSS. New network nodes have been introduced for the transport of packet data. These nodes are the gateway GPRS support nodes (GGSN) and serving GPRS support nodes (SGSN). The subnetwork formed by the SGSNs and the GGSNs is called the GPRS core network. In order to reuse the GSM nodes, new interfaces have been defined between the GSM network nodes and the different elements of the GPRS core network. The GPRS logical architecture is described in Section 3.1.

The protocol layer has been split into two planes. On one side there is the transmission plane, which is mainly used for the transfer of user data. The signaling plane is used for the control and support of the transmission plane functions. Section 3.2 deals with the transmission and signaling planes.

GPRS has kept such main principles of the GSM radio interface as the notions of time slot, frame, multiframe, and hyperframe structures. It was indeed chosen by the operators and manufacturers involved in the system design to provide high-data-rate packet-switched services with minimized impacts on the GSM standard. The principles of the physical layer are given in Section 3.3.1. The details related to the physical layer are presented in Chapter 4.

One of the main GPRS characteristics is that a physical connection is established in uplink only when the MS needs to send continuous data to the network, and in downlink only when the network needs to send continuous data to the MS. This physical connection is released in one direction as soon as the sending entity has no more data to send. Different allocation schemes for radio resource (RR) management have been defined in order to multiplex several MSs on the same physical channel. An overview of the principles related to RR management is presented in Section 3.3.2. A complete description of RR management can be found in Chapter 5.

A logical entity called PCU has been introduced within the BSS to manage the GPRS functions over the radio interface. Section 3.4 deals with the BSS architecture and discusses the several possible locations of the PCU.

In a GPRS network, an RA identifies one or several cells. As soon as an MS enters a new RA, the network must be notified of this change in order to update its location. The SGSN is in charge of GPRS mobility management (GMM). An overview of GPRS mobility is proposed in Section 3.5. The details related to GMM are given in Chapter 7.

If data transmission in packet mode does not require the establishment of an end-to-end connection, it is necessary to establish a context between the mobile and the network in order to exchange packets. This context allows the network to identify the IP address of the MS, identify the access point with the external network, and define the QoS associated with data transmission in packet mode. The concept of PDP context is explained in Section 3.6. The details related to PDP context management are given in Chapter 7.

The BSS and the GPRS backbone network are connected via the Gb interface in order to exchange user data and signaling information. The principles of the Gb interface are given in Section 3.8. The details related to the Gb interface are given in Chapter 6.

When a context is established between the MS and the network, IP packet exchange may start at any time between the mobile and the network without establishing a connection beforehand. The packets are conveyed in the GPRS backbone network. An overview of the general architecture of the GPRS backbone network is presented in Section 3.9. A complete description of the user plane between the MS and external data packet network is given in Chapter 8.

3.1 GPRS Logical Architecture
Figure 3.1 shows the elements that are part of a GPRS network and their associated interfaces. A GPRS network is composed of the following network nodes:

SGSN. The SGSN is the node that is serving the MS; it is responsible for GMM. It delivers packets to the MSs and communicates with the HLR to obtain the GPRS subscriber profile. It manages the registration of the new mobile subscribers in order to keep a record of their LA for routing purposes. The SGSN can be connected to one or several BSSs.

GGSN. The GGSN provides interworking with external packet data networks (PDNs). It may be linked to one or several data networks. It is connected with SGSNs via an IP-based GPRS backbone network. The GGSN is a router that forwards incoming packets from the external PDN to the SGSN of the addressed MS. It also forwards outgoing packets to the external PDN. The PDN is the external fixed data network to which is connected the GPRS network. An example of a PDN is the Internet network.

HLR. The HLR is a database that contains, among other things, packet domain subscription data and routing information.

Mobile switching center/visitor location register (MSC/VLR). The MSC coordinates the setting up of calls to and from GSM users and manages GSM mobility. The MSC is not directly involved in the GPRS network. It forwards circuit-switched paging for the GPRS-attached MSs to the SGSN when the Gs interface is present.

BSS. The BSS ensures the radio connection between the mobile and the network. It is responsible for radio access management. The BSS is composed of two elements: the BTS and the BSC. The BTS integrates all the radio transmission and radio reception boards. The BSC is responsible for the management of the radio channels. The BSC has switching capabilities that are used for circuit-switched calls and can also be used for GPRS traffic.

EIR. The EIR is a database that contains terminal identities.
3.2 Transmission and Signaling Planes
3.2.1 Transmission Plane
The transmission plane consists of a layered protocol structure providing user data transfer, along with associated procedures that control the information transfer such as flow control, error detection, and error correction. Figure 3.2 illustrates the layered protocol structure between the MS and the GGSN.


Figure 3.2: Transmission plane MS-GGSN.
3.2.1.1 Air Interface
The air interface is located between the MS and the BSS. The protocols used on the air interface are as follows:

Radio link control/medium access control (RLC/MAC). RLC provides a reliable radio link between the mobile and the BSS. MAC controls the access signaling procedures to the GPRS radio channel, and the multiplexing of signaling and RLC blocks from different users onto the GSM physical channel.

GSM-RF layer. It is the radio subsystem that supports a certain number of logical channels. This layer is split into two sublayers: the radio frequency layer (RFL), which handles the radio and baseband part (physical channel management, modulation, demodulation, and transmission and reception of radio blocks), and the physical link layer (PLL), which manages control of the RFL (power control, synchronization, measurements, and channel coding/decoding).

A relay function is implemented in the BSS to relay the LLC PDUs between the air interface and the Gb interface.

3.2.1.2 Gb Interface
The Gb interface is located between the SGSN and the BSS. It supports data transfer in the transmission plane. The Gb interface supports the following protocols:

BSS GPRS protocol (BSSGP). This layer conveys routing and QoS-related information between the BSS and SGSN.

Network service (NS). It transports BSSGP PDUs and is based on a frame relay connection between the BSS and SGSN.

A relay function is implemented in the SGSN to relay the packet data protocol (PDP) PDUs between the Gb and Gn interfaces (IP PDUs in Figure 3.2).

3.2.1.3 Gn/Gp Interface
The Gn interface is located between two GSNs (SGSN or GGSN) within the same PLMN, while the Gp interface is between two GSNs in different PLMNs. The Gn/Gp interface is used for the transfer of packets between the SGSN and the GGSN in the transmission plane.

The Gn/Gp interface supports the following protocols:

GPRS tunnelling protocol (GTP). This protocol tunnels user data between the SGSN and GGSN in the GPRS backbone network. GTP operates on top of UDP over IP. The layers L1 and L2 of the Gn interfaces are not specified in the GSM/GPRS standard.

User datagram protocol (UDP). It carries GTP packet data units (PDUs) in the GPRS Core Network for protocols that do not need a reliable data link (e.g., IP).

Internet protocol (IP). This is the protocol used for routing user data and control signaling within the GPRS backbone network.

3.2.1.4 Interface Between MS and SGSN
This interface supports the following protocols:

Subnetwork-dependent convergence protocol (SNDCP). This protocol maps the IP protocol to the underlying network. SNDCP also provides other functions such as compression, segmentation, and multiplexing of network layer messages.

Logical link control (LLC). This layer provides a highly reliable logical link that is independent of the underlying radio interface protocols. LLC is also responsible for the GPRS ciphering.

3.2.2 Signaling Plane
The signaling plane consists of protocols for control and support of the transmission plane functions. It controls both the access connections to the GPRS network (e.g., GPRS attach and GPRS detach) and the attributes of an established network access connection (e.g., activation of a PDP address), manages the routing of information for a dedicated network connection in order to support user mobility, adapts network resources depending on the QoS parameters, and provides supplementary services.

3.2.2.1 Between MS and SGSN
Figure 3.3 shows the signaling plane between the MS and the SGSN. This plane is made up of the following protocols:

GMM. The GMM protocol supports mobility management functionalities such as GPRS attach, GPRS detach, security, RA update, and location update (see Section 3.5).

Session management (SM). The SM protocol supports functionalities such as PDP context activation, PDP context modification, and PDP context deactivation (see Section 3.6).


Figure 3.3: Signaling plane MS-SGSN.
3.2.2.2 Between Two GSNs
In the signaling plane, Gn/Gp interfaces are used for the transfer of signaling between the GSNs in the GPRS backbone network. Figure 3.4 shows the signaling plane between two GSNs.


Figure 3.4: Signaling plane GSN-GSN.
The signaling plane between two GSNs is made up of the following protocols:

GTP for the control plane (GTP-C). This protocol tunnels signaling messages between SGSNs and GGSNs, and between SGSNs, in the GPRS core network (see Section 3.9).

UDP. This protocol transfers signaling messages between GSNs.

3.2.2.3 Interface with Signaling System No. 7
The various GSNs of the GPRS backbone network use a Signaling System No. 7 (SS7) network to exchange information with GSM SS7 network nodes such as HLR, MSC/VLR, EIR, and SMS-GMSC. The SS7 network provides facilities to quickly exchange messages between GPRS backbone network nodes irrespective of data transmission through the GPRS PLMN network.

In the GSM/GPRS backbone network, we found the following protocols for SS7 signaling:

Message transfer part (MTP). The three MTP layers allow signaling messages to be exchanged through the SS7 network.

Signaling connection control part (SCCP). The SCCP protocol layer allows the service to be used in connected mode and messages to be exchanged between different PLMNs by using an international gateway for SS7 address translation between an SS7 global address (based on the E.164 numbering plan) and an SS7 local address.

Transaction capabilities application part (TCAP). The TCAP protocol layer allows dialogs to be structured in an independent manner from any application.

Mobile application part (MAP). The MAP protocol layer allows mobile mobility to be managed within different equipment nodes of the NSS across SS7 networks.

As new equipment nodes have been introduced in GSM networks to support the GPRS feature, new interfaces were defined with the HLR, MSC/VLR, EIR, and SMS-GMSC. Table 3.1 lists the new interfaces with SS7 network.

Table 3.1: New Interfaces with the SS7 Network Interface Name
Localization
Mandatory/Optional

Gr
SGSN-HLR
Mandatory

Gc
GGSN-HLR
Optional

Gf
SGSN-EIR
Optional

Gd
SGSN-SMS GMSC or SGSN-SMS IWMSC
Optional

Gs
SGSN-MSC/VLR
Optional


Gr Interface
The Gr interface is defined between the SGSN and HLR. It allows the SGSN to retrieve or update GPRS subscription and GPRS location information in the HLR during location-management or authentication procedures. The MAP protocol has been modified to take into account this interface. Figure 3.5 shows the signaling plane on the Gr interface.


Figure 3.5: Signaling plane on the Gr interface.
Gc Interface
The Gc interface is defined between the GGSN and the HLR. The GGSN contacts the HLR in order to determine the SGSN address where the MS is located and if the MS is reachable. The MAP protocol has been modified to take into account this interface. Figure 3.6 shows the signaling plane on the Gc interface. (Note: If a GGSN does not have a SS7 MAP interface, it will interface to a GSN performing a GTP-MAP protocol-conversion in order to retrieve the needed information from the HLR via the Gc interface.)


Figure 3.6: Signaling plane on the Gc interface.
Gf Interface
The Gf interface is defined between the SGSN and EIR. It is used by the SGSN to contact the EIR database during the identity check procedure. It allows the SGSN to check the IMEI against the EIR. The MAP protocol has been modified to take into account this interface. Figure 3.7 shows the signaling plane on the Gf interface.


Figure 3.7: Signaling plane on the Gf interface.
Gs Interface
The Gs interface is defined between the MSC/VLR and the SGSN. It allows for the coordination of circuit-switched and packet-switched paging in the SGSN as well as location information of any MS attached to both circuit and packet services. This interface is only present in a network that operates in network operation mode I (see definition in Section 3.5.3.1). The BSS application part+ (BSSAP+) allows mobility functionality to be managed on the Gs interface. Figure 3.8 shows the signaling plane on the Gs interface.


Figure 3.8: Signaling plane on the Gs interface.
Gd Interface
The Gd interface is defined between an SGSN and an SMS-GMSC or an SMS-IWMSC. The progress of a short message intended for delivery to an MS requires in circuit and packet modes a gateway function Short Message Service-gateway MSC (SMS-GMSC) between the mobile network and the network that provides access to the SMS center. An SMS to be delivered to an MS is routed from the SMS-GMSC toward the SGSN on the Gd interface if this SMS is to be sent over GPRS.

The progress of a short message originated by the MSs requires in circuit and packet modes a PLMN interworking server SMS-IWMSC (inter-working MSC) that provides access to the SMS center. An SMS originated by an MS is routed from the SGSN toward the SMS-IWMSC on the Gd interface if the SMS is to be sent over GPRS.

The MAP protocol has been updated as a consequence of the signaling exchange between the SGSN and the SMS-GMSC or the SMS-IWMSC. Figure 3.9 shows the signaling plane on the Gd interface.


Figure 3.9: Signaling plane on the Gd interface.


3.3 Radio Interface
3.3.1 Physical Layer Principles
3.3.1.1 Packet Data Channel
The GPRS physical layer is based on that of the GSM (see Chapter 1). The access scheme is TDMA, with eight basic physical channels per carrier (TS 0 to 7).

A physical channel uses a combination of frequency- and time-division multiplexing and is defined as a radio frequency channel and time slot pair. The physical channel that is used for packet logical channels is called a packet data channel (PDCH). PDCHs are dynamically allocated in the cell by the network. The PDCH is mapped on a 52-multiframe, as shown in Figure 3.10. The 52-multiframe consists of 12 radio blocks (BO to B11) of 4 consecutive TDMA frames and 4 idle frames (frames 12, 25, 38, and 51), amounting to a total of 52 frames.


Figure 3.10: Time slots and TDMA frames.
3.3.1.2 Packet Data Logical Channel
GPRS, like GSM, uses the concept of logical channels mapped on top of the physical channels. Two types of logical channels have been introduced, namely traffic channels and control channels. Three subtypes of control channels have been defined for GPRS: broadcast, common control, and associated. In addition, the GSM common control channels (BCCH, CCCH, and RACH) may be used to access the network and establish packet transfer.

The different packet data logical channels are:

Packet broadcast control channel (PBCCH). The presence of PBCCH in the cell is optional. The PBCCH broadcasts information relative to the cell in which the mobile camps and information on the neighbor cells. This information is used by the mobile in order to access the network. When there is no PBCCH in the cell, the information needed by the mobile to access the network for a packet transfer is broadcast on BCCH.

Packet common control channel (PCCCH). The PCCCH is present in the cell only if the PBCCH is present in the cell. When it is not present in the cell, the common control signaling for GPRS is handled on the GSM common control channels (CCCH). PCCCH is composed of packet random access channel (PRACH), used for random access, packet paging channel (PPCH), used for paging, and packet access grant channel (PAGCH), used for access grant. The PRACH is used by the MS to initiate uplink access to the network. The PPCH is used by the network to page the mobile in order to establish a downlink packet transfer. The PAGCH is used by the network to assign radio resources to the mobile for a packet transfer.

Packet data traffic channel (PDTCH). The PDTCH is used to transfer user data during uplink or downlink packet transfer. The PDTCH is a unidirectional channel, either uplink (PDTCH/U) for a mobile-originated packet transfer or downlink (PDTCH/D) for a mobile-terminated packet transfer. A PDTCH is a resource allocated on one physical channel by the network for user data transmission.

Packet associated control channel (PACCH). The PACCH is a unidirectional channel that is used to carry signaling during uplink or downlink packet data transfer. The uplink PACCH carries signaling from the MS to the network and the downlink PACCH carries signaling from the network to the mobile. The PACCH is dynamically allocated on a block basis.

Packet timing advance control channel (PTCCH). The PTCCH is a bidirectional channel that is used for TA update. The PTCCH is an optional channel. The PTCCH when present is mapped on frames number 12 and 38 of the 52—multiframe.

Table 3.2 lists the various GPRS logical channels.

Table 3.2: GPRS Logical Channels Logical Channel
Abbreviation
Uplink/ Downlink
Task

Packet broadcast control channel
PBCCH
DL
Packet system broadcast information

Packet paging channel
PPCH
DL
MS paging for downlink transfer establishment

Packet random access channel
PRACH
UL
MS random access for uplink transferestablishment

Packet access grant channel
PAGCH
DL
Radio resources assignment

Packet timing advance control channel
PTCCH
UL/DL
Timing advance update

Packet associated control channel
PACCH
UL/DL
Signaling associated with data transfer

Packet data traffic channel
PDTCH
UL/DL
Data channel


3.3.1.3 Multislot Classes Definition
In order to provide higher throughputs, a GPRS MS may transmit or receive in several time slots of the TDMA frame. The multislot capability is indicated by the multislot class of the GPRS MS. This multislot class is defined by several parameters such as the maximum number of time slots supported by the MS per TDMA frame in uplink and in downlink. The multislot class of the MS is sent to the network during the GPRS attach procedure. A detail of multislot classes is given in Section 4.2.1.

3.3.1.4 Cell Reselection
In GPRS as in GSM, the mobile performs cell reselection. However, there are some differences compared with GSM.

In GPRS the mobile performs cell reselection when it is in idle mode but also during packet transfer. The cell reselection is either performed by the mobile autonomously or optionally controlled by the network. Unlike GSM, there is no handover in GPRS but only cell reselections. So when there is a reselection during a packet transfer, this latter is interrupted and it has to be started again in the new cell. There is an interruption of the packet transfer during the reselection phase.

Although the GPRS cell reselection algorithms used by the mobile are based on the same principles as those used in GSM, they have been slightly modified in order to provide more flexibility. These algorithms are described in Section 5.3.

3.3.1.5 Radio Environment Monitoring
The MS performs different types of radio measurements that are reported to the network and used by it for RLC. These estimations are also used by the mobile itself to compute its transmission power (open-loop power control; see Section 4.1.3.1), for cell selection and cell reselection.

The mobile performs the following types of measurement:

Received signal level (RXLEV) measurements. The RXLEV measurements are performed on both serving cell and neighbor cells for the purpose of cell reselection. During packet-transfer mode (see Section 3.3.2.1), the serving cell RXLEV measurement can also be used for downlink coding scheme adaptation (see Section 4.1.2.1), network-controlled cell reselection, and downlink and uplink power control.

Quality (RXQUAL) measurements. The RXQUAL is computed from the average BER before channel decoding. During packet-transfer mode the mobile estimates the quality of the downlink blocks it receives. In packet idle mode, no quality measurements are performed. The RXQUAL can be used by the network for network-controlled cell reselection, dynamic coding scheme adaptation, and downlink power control. The RXQUAL is the current GSM quality indicator.

Interference measurements. These measurements have been introduced for GPRS. They correspond to a received signal level measurement performed on a frequency that is different from a beacon frequency. The interest is in having an estimation of the interference level on the PDTCH. It can be used by the network to optimize the mobile RR allocation, to select a more appropriate coding scheme, to trigger a network-controlled cell reselection, and for power control or for network statistics.

The measurements are either used by the MS for its own purposes or by the network. The RXLEV and RXQUAL measurements are also performed at the BSS side for each MS. They are used for network-controlled cell reselection, uplink power control (closed loop, see Section 4.1.3.1), and dynamic coding scheme adaptation.

3.3.1.6 Principles of Power Control
Power control can be used in order to improve the spectrum efficiency while maintaining radio link quality and reducing the power consumption in the MS. Power control in GPRS is more complicated than for a circuit-switched connection, since there is not necessarily a continuous two-way connection.

Note that power control can be performed in both uplink and downlink directions. Uplink power control and downlink power control are described in Sections 4.1.3.1 and 4.1.3.2, respectively.

3.3.1.7 Channel Coding
The GPRS user data is sent on radio blocks encoded with one of four channel coding schemes (CS1, CS2, CS3, CS4). The GPRS signaling is sent with the CS1 channel coding scheme. This scheme provides the highest protection level against error transmission, while the CS4 channel coding scheme provides the lowest protection level. The more the channel coding scheme provides an efficient protection level against error transmission, the more the useful data throughput decreases due to redundant information added to source data. The channel coding scheme used on uplink and downlink depends on the radio quality between the network and the MS. Table 3.3 gives the throughputs associated with each coding scheme.

Table 3.3: Throughput Associated with Coding Scheme Coding Scheme
Throughput (Kbps)

CS-1
9.05

CS-2
13.4

CS-3
15.6

CS-4
21.4


3.3.2 RR Management Principles
3.3.2.1 RR Operating Modes
At the RR level, the MS behavior is dependent on two operating RR states. These states, packet idle mode and packet transfer mode, allow the RR activity of the MS to be characterized.

Packet Idle Mode
When the MS is in packet idle mode, no radio resources are allocated. Leaving packet idle mode occurs when upper layers request the transfer of uplink data requiring the assignment of uplink resources from the network. It also occurs at the reception of a downlink resource assignment command from the network for a downlink transfer.

In case of downlink transfer, the mobile switches from packet idle mode to packet transfer mode when it receives the downlink assignment command from the network. In the case of uplink transfer, the mobile leaves packet idle mode when it requests the assignment of uplink resources to the network. However, switching to packet transfer mode is not instantaneous. The mobile switches to packet transfer mode only when it has been uniquely identified at the network side; this will be explained in more detail in Chapter 5. Thus there is a period between packet idle mode and packet transfer mode during which the mobile is in a transitory state.

During packet idle mode, the MS listens to its PCH and the CBCH. This last one is the PBCCH when present in the cell; otherwise it is the BCCH.

Packet Transfer Mode
When the MS is in packet transfer mode, it is clearly identified at the network side and uplink or/and downlink radio resources are allocated.

Switching from packet transfer mode to packet idle mode occurs when the network releases all downlink and uplink resources. This transition can also occur in the case of an abnormal condition during packet transfer mode (e.g., radio link failure) or when the mobile decides on a cell reselection toward a new cell.

During packet transfer mode, the mobile transmits and receives data. Figure 3.11 summarizes the transition between the different RR states.


Figure 3.11: Transition between RR operating modes.
3.3.2.2 Temporary Block Flow
A temporary block flow (TBF) is a physical connection between the RR entity in the MS and the RR entity at the network side to support the unidirectional transfer of LLC protocol data units over PDCH. A TBF is characterized by one or several PDCHs allocated by the network to an MS for the duration of the data transfer. Once the data transfer is finished, the TBF is released.

When the mobile must send continuous data to the network, it requests the establishment of an uplink TBF by sending signaling information over CCCH or PCCCH. When the network wants to send data to the mobile, it assigns a downlink TBF between the two RR entities.

A downlink TBF supports the transfer of data from the network to the mobile, while an uplink TBF supports the transfer of data from the mobile to the network. One uplink and one downlink TBF can be supported at the same time between the two RR entities. These two TBFs are defined as concurrent TBFs.

The number of TBFs per mobile and per direction is limited to one. However TBFs belonging to different mobiles can share the same PDCH.

Note As a packet transfer session is composed of a lot of requests, responses, and acknowledgments, many consecutive uplink and downlink TBFs are established for the same session (e.g., Web browsing).


Each TBF is identified by a temporary flow identifier (TFI) assigned by the network. So in case of concurrent TBFs, one TFI identifies the uplink TBF and another one the downlink TBF. The TFI is used to differentiate TBFs sharing the same PDCHs in one direction.

Figure 3.12 gives an example of TBF mapping for different mobiles onto PDCHs. A downlink TBF, identified by a TFI equal to 2, has assigned resources on the PDCH numbers 2, 3, and 4, while another downlink TBF identified by a TFI equal to 1 has assigned resources on the PDCH numbers 4, 5, and 6. An uplink TBF identified by a TFI equal to 4 has assigned resources on the PDCH numbers 3 and 4.


Figure 3.12: Example of mapping of TBFs with their respective TFI onto PDCHs.
3.3.2.3 Allocation Modes on the Uplink
Several MSs may be multiplexed on the same PDCH. In order to share the uplink bandwidth between several mobiles mapped on the same PDCH, different allocation schemes have been defined to allocate an uplink radio block instance to a particular mobile.

Three allocation schemes exist for medium access control:

Dynamic allocation;

Extended dynamic allocation;

Fixed allocation.

The second scheme is optional for the mobile while the others are mandatory. On the network side, either fixed allocation or dynamic allocation must be implemented. Extended dynamic allocation is optional for the network.

Dynamic Allocation
In principle, dynamic allocation allows uplink transmission to mobiles sharing the same PDCH, on a block-by-block basis. During the uplink TBF establishment, an uplink state flag (USF) is given to the MS for each allocated uplink PDCH. The USF is used as a token given by the network to allow transmission of one uplink block.

Whenever the network wants to allocate one radio block occurrence on one uplink PDCH, it includes, on the associated downlink PDCH, the USF in the radio block immediately preceding the allocated block occurrence. When the mobile decodes its assigned USF value in a radio block sent on a downlink PDCH associated with an allocated uplink one, it transmits an uplink radio block in the next uplink radio block occurrence, that is, the B(x) radio block if the USF was detected in the B(x- 1) radio block.

The principle of dynamic allocation is illustrated in Figure 3.13.


Figure 3.13: Principle of dynamic allocation.
The USF is included in the header of each downlink RLC/MAC block. The USF coding (3 bits) enables eight mobiles to be multiplexed on the same uplink PDCH.

Dynamic allocation implies the constant monitoring (radio block decoding) of the downlink PDCHs associated with the allocated uplink PDCHs.

As explained previously, the USF allows the sending of one block in the next uplink occurrence. However, dynamic allocation can also be used in such a way that the decoding of one USF value allows the mobile to send four consecutive uplink blocks on the same PDCH. The choice between one block or four blocks is indicated during the TBF establishment by the network to the mobile.

The concept of USF granularity has been introduced in order to indicate the number of uplink radio blocks to be sent upon detection of the assigned USF by the MS. The USF granularity is conveyed to the MS during the uplink TBF establishment. Thus it involves the MS transmitting either a single radio block or a sequence of four consecutive radio blocks starting on the B(x) radio block if the USF was detected in the B(x - 1) radio block.

Extended Dynamic Allocation
The extended dynamic allocation scheme offers an improvement over the dynamic allocation scheme. Some RR configurations are not compliant with all MS multislot classes in the dynamic allocation scheme. In the dynamic allocation scheme, the MS must decode all USF values on all downlink PDCHs associated with the allocated uplink PDCHs.

The mobile monitors its assigned PDCHs starting from the lowest numbered one (the one that is mapped on the first allocated time slot in the TDMA frame), then it monitors the next lowest numbered time slot, and so on. Whenever the MS detects its assigned USF value on a PDCH, it transmits one radio block or a sequence of four radio blocks on the same PDCH and all higher-numbered assigned PDCHs. The mobile does not need to monitor the USF on these higher PDCHs. This is of particular interest in some RR configurations that are not compliant with all MS multislot classes in the dynamic allocation scheme.

Let us take the example of a class 12 MS, which is defined by:

A maximum number of four receive time slots per TDMA frame;

A maximum number of four transmit time slots per TDMA frame;

A total number of transmit and receive time slots per TDMA frame less than or equal to five.

The network cannot allocate four uplink PDCHs to a MS multislot class 12 with the dynamic allocation. Indeed, the MS must decode the USF fields on the four associated downlink PDCHs. This means that the MS would have to receive on four time slots to be able to transmit on four time slots. That gives a total number of eight received and transmit time slots, which is not compliant with a multislot class 12 MS. In the case of extended dynamic allocation, the network can allocate four uplink PDCHs without exceeding a total number of five receive and transmit time slots.

Fixed Allocation
Fixed allocation enables a given MS to be signaled predetermined uplink block occurrences on which it is allowed to transmit. The network assigns to each mobile a fixed uplink resource allocation of radio blocks onto one or several PDCHs.

The network allocates uplink radio blocks using bitmaps (series of zeros and ones). A 0 indicates that the mobile is not allowed to transmit, and a 1 indicates a transmission occurrence. The bitmaps are sent during the establishment of the uplink TBF. If more uplink resources are required during the uplink TBF, the network sends a bitmap in downlink on the PACCH.

A fixed allocation TBF operates as an open-ended TBF when an arbitrary number of octets are transferred during the uplink TBF. When the allocated bitmap ends, the MS requests a new bitmap if it wishes to continue the TBF.

A fixed allocation TBF operates as a close-ended TBF when the MS specifies the number of octets to be transferred during the uplink TBF establishment.

Comparison of Allocation Schemes
Fixed allocation allows for an efficient usage of the MS multislot capability, as the downlink monitoring is limited to one time slot (listening of the PACCH). The advantage of dynamic allocation is that the management of uplink resources is much more flexible.

From an implementation point of view, dynamic allocation is easier to implement on the network side. An efficient management of bitmap in fixed allocation is not so easy compared with USF handling. In fact, the management of resource allocation with anticipation (as needed in fixed allocation) is made more complex by the bursty nature of packet transfer. Most of the time the duration of the uplink TBF is completely unknown at the network side at the beginning of the TBF. This is why it is very difficult to manage radio block allocation at the beginning of the TBF. If the bitmap allocation is too short, the reaction time of the network for reallocation of new bitmaps will increase the duration of the TBF. If the bitmap allocation is too large compared with the TBF duration, block bitmap will be reallocated at the network side in order to avoid waste of uplink resources. Dynamic allocation is very easy to manage, since the USF allocation needs to anticipate only a few block periods.






3.4 BSS Architecture
3.4.1 PCU
In order to introduce GPRS within the BSS, the PCU concept has been defined. The PCU stands for a logical entity that manages packet toward the radio interface. The PCU communicates with the channel codec unit (CCU), positioned in the BTS.

The PCU is in charge of RLC/MAC functions such as segmentation and reassembly of LLC frames, transfer of RLC blocks in acknowledged or unacknowledged mode, radio resource assignment, and radio channel management. The CCU handles GSM layer 1 functions such as channel decoding, channel encoding, equalization, and radio channel measurements.

As shown in Figure 3.14, the PCU can be located either at the BTS site, the BSC site, or the SGSN site. When the PCU is located at the BSC or SGSN site, it is referred to as being a remote PCU.


Figure 3.14: Remote PCU position. (From- [1].)
If the PCU is located at the BSC, it could be implemented as an adjunct unit to the BSC. When the PCU is located at the SGSN side, the BSC is transparent for frames transmitted between the PCU and the CCU. This PCU location implies the implementation of a signaling protocol between the BSC and the PCU (e.g., for time slot management, access on CCCH management). A protocol is also needed when the PCU is located at the BTS side.

In the case of remote PCU, GPRS traffic between the PCU and the CCU are transferred through the Abis interface. The Abis interface in GSM is based on TRAU frames carrying speech data and having a fixed length of 320 bits (every 20 ms). This corresponds to a throughput of 16 Kbps per Abis channel.

As the PCU is supporting functions such as RLC block handling (retransmission, segmentation, and so forth) and access control, it needs to know the GSM radio interface timing. This implies in the case of a remote PCU the design of a synchronous interface between the PCU and the CCU. The PCU must be able to determine in which radio frames is sent an RLC/ MAC block. In case of a PCU located at the BTS side, the PCU easily knows the radio interface time.

The remote PCU solution requires the sending of in-band information between the PCU and the CCU (for transmission power indication, channel coding indication, and synchronization between the CCU and the PCU).

The advantages and drawbacks of each solution are listed next.

PCU at the BTS Side
Advantages:

There is an internal interface between the PCU and the CCU.

There is a low round-trip delay (the round-trip delay is the time between the transmission of a block and the reception of the answer).

There is no waste of Abis bandwidth due to retransmission of RLC blocks.

Drawbacks:

There is a likely impact on the existing BTS hardware (when migrating from a circuit-switched network to a GPRS one). This is an important drawback considering the number of BTSs in the field.

PCU at the BSC Side
Advantages:

There is an internal interface between the PCU and the BSC.

Drawbacks:

There are likely hardware impacts on the current BSC (however, the number of BSCs is lower compared with the number of BTSs).

There is a greater round-trip delay.

A synchronous protocol is needed between the PCU and the CCU.

Abis bandwidth is wasted in case of RLC blocks retransmission.

PCU at the SGSN Side
Advantages:

There is no hardware impact on the current GSM network (BTS, BSC).

There is a smooth introduction of GPRS in the network.

Drawbacks:

There is a greater round-trip delay (longer TBF establishment).

A synchronous protocol is needed between the PCU and the CCU.

Abis bandwidth is wasted in case of RLC blocks retransmission.

A protocol is needed between the BSC and the PCU.

3.4.2 Transmission Plane
When the PCU function is not implemented in the BTS, a new protocol (L1/L2) between the PCU and the BTS is introduced. This protocol ensures transmission of RLC/MAC blocks from the PCU to the CCU, and allows in-band signaling for control of the CCU from the PCU and synchronization between both entities. Figure 3.15 shows the BSS transmission plane for a remote PCU.


Figure 3.15: BSS transmission plane for a remote PCU.
When the PCU is located in the BTS, a protocol (L1/L2/L3) is needed between the BSC and the BTS for the transmission of the LLC frames. The BSS transmission plane is shown in Figure 3.16.


Figure 3.16: BSS transmission plane when the PCU is in the BTS.
3.4.3 Signaling Plane
Depending on the PCU location, the signaling plane within the BSS will not be the same.

3.4.3.1 PCU in the BTS
The RR layer is at the BTS side. The (L1/L2/L3) protocol is used for the transmission of signaling between the BSC and the BTS. Figure 3.17 shows the signaling plane when the PCU is located in the BTS.


Figure 3.17: Signaling plane when PCU is in the BTS.
3.4.3.2 PCU at BSC or SGSN Side
The signaling plane implies the usage of the (L1/L2) protocol for RR signaling transmission. Figure 3.18 shows the signaling plane when the PCU is located at BSC side or SGSN side.


Figure 3.18: Signaling plane when PCU is at BSC or SGSN side.