WO2003015354A1

WO2003015354A1 - A communication network, a network element and method of congestion control therefor

Info

Publication number: WO2003015354A1
Application number: PCT/EP2002/007275
Authority: WO
Inventors: Liang Qiu Liu; Mehdi Witwit
Original assignee: Motorola Inc
Priority date: 2001-08-08
Filing date: 2002-07-01
Publication date: 2003-02-20
Also published as: GB2378613B; GB2378613A; GB0119291D0

Abstract

This invention relates to a communication network (300) using an unacknowledged transport layer data protocol (UTDP). A network element (305) of the communication network (300) has a receiver (311) for receiving data packets from other network elements in accordance with UDTP, e.g. using the User Datagram Protocol. it futhermore has a congestion processor (315) determining a congestion level for the network element (305) for UTDP data packets and means (317) for generating a transport layer congestion feedback message in response to the congestion level. The network element has a transmitter (319) for sending the transport layer congestion feedback message to network elements (301, 303) transmitting packets to the first network element (305). The message is transmitted using a higher layer signalling protocol, such as the GPRS Tunnelling Protocol (GTP) or Radio Access Network Application Protocol (RANAP). The invention is particularly applicable to cellular communication systems such as UMTS.

Description

A COMMUNICATION NETWORK, A NETWORK ELEMENT AND METHOD OF CONGESTION CONTROL THEREFOR

Field of the Invention

This invention relates to a communication network, a network element and method of congestion control therefor and in particular applicable, but not limited, to cellular communication systems.

Background of the Invention

Present day communications systems, both wireless and wire-line, have a requirement to transfer data between communications units. Data, in this context, includes speech and multimedia cornmunication. Such data transfer needs to be effectively and efficiently provided for, in order to optimise use of limited communication resources.

For data to be transferred across communications networks, a communication unit addressing protocol is required. The communication units are generally allocated addresses that are read by communications equipment such as bridges, gateways and/or routers, in order to determine how to transfer the data to the addressed unit. The interconnection between networks is generally known as internetworking (or internet).

Networks are often divided into sub-networks, with protocols being set up to define a set of rules that allow the orderly exchange of information. Currently, the two most popular protocol suites used to transfer data in communications systems are: connection-oriented Transfer Control Protocol (TCP) and Internet Protocol (TP) and connectionless User Datagram Protocol (UDP) over IP. In all but the simplest of communications systems, the TCP or UDP protocols often work as a complementary pair with IP. The IP protocol corresponds to data transfer in the network layer of the well-known OSI model and the TCP and UDP portion to data transfer in the transport layer of the OSI model. Their operation is transparent to the physical and data link layers and can thus be used on any of the standard cabling networks such as Ethernet, FDDI or token ring.

The Internet Protocol adds a data header on to the information passed from the transport layer. The resultant data packet is known as an Internet datagram. The header of the datagram contains information such as destination and source IP addresses, the version number of the IP protocol etc. An IP address is assigned to each node on the internet. It is used to identify the location of the network and any sub-networks.

When transmitted from the source node, each datagram is routed separately through the Internet and the received fragments are finally reassembled at the destination node, prior to forwarding the data to the respective communication unit. The TCP-IP version number helps gateways and nodes interpret the data packet correctly.

Due to the recent growth in communications, particularly in internet and wireless communications, there exists a need to provide packet data transfer techniques in a wireless communications domain.

An established harmonised cellular radio communication system is GSM (Global System for Mobile Communications). An enhancement to this cellular technology can be found in the Global Packet Radio System (GPRS), which provides packet switched technology on a basic cellular platform, such as GSM. A further harmonised wireless communications system currently being defined is the universal mobile telecommunication system (UMTS), which is intended to provide a harmonised standard under which cellular radio communications networks and systems will provide enhanced levels of interfacing and compatibility with other types of communications systems and networks, including fixed communications systems such as the Internet.

In a cellular communication system, each of the communication devices, including subscriber units, mobile station and mobile communication devices, communicates with typically a fixed base station. Communication from the communication device to the base station is known as uplink and communication from the base station to the communication device is known as downlink. The total coverage area of the system is divided into a number of separate cells, each predominantly covered by a single base station. The cells are typically geographically distinct with an overlapping coverage area with neighbouring cells. FIG. 1 illustrates a cellular communication system 100. hi the system, abase station 101 communicates with a number of communication devices 103 over radio channels 105. In the cellular system, the base station 101 covers users within a certain geographical area 107, whereas other geographical areas 109, 111 are covered by other base stations 113, 115. Some overlap areas can be covered by more than one cell.

As a communication device moves from the coverage area of one cell to the coverage area of another cell, the communication link will change from being between the communication device and the base station of the first cell, to being between the communication device and the base station of the second cell. This is known as a handover. Specifically, some cells may lie completely within the coverage of other larger cells.

All base stations are interconnected by a fixed network. This fixed network comprises communication lines, switches, interfaces to other communication networks and various controllers required for operating the network. A call from a communication device is routed through the fixed network to the destination specific for this call. If the call is between two communication devices of the same communication system the call will be routed through the fixed network to the base station of the cell in which the other communication device currently is. A connection is thus established between the two serving cells through the fixed network. Alternatively, if the call is between a communication device and a telephone connected to the Public Switched Telephone Network (PSTN) the call is routed from the serving base station to the interface between the cellular mobile communication system and the PSTN. It is then routed from the interface to the telephone by the PSTN. A cellular mobile commumcation system is allocated a frequency spectrum for the radio communication between the communication devices and the base stations. This spectrum must be shared between all communication devices simultaneously using the system.

Information to be transmitted across the Internet is packetised, with packet switching routes established between a source node and a destination node. Hence, GPRS and UMTS networks have been designed to accommodate packet switched data to facilitate Internet services, such as message service, information service, conversational service and casting service.

FIG. 2 illustrates a simplified network architecture 200 for the packet switched part of a UMTS network. The base stations of UMTS are known as Node Bs 201, 203, 205, 207 and provide the radio communication with the subscriber units 209. The Node B's are connected to Radio Network Controllers (RNCs) 211,213 which are responsible for interfacing the Node Bs to the core network 215 and for the control of the radio interface resources. The packet switched part of the core network comprises Serving GPRS (General Packet Radio Service) Support Nodes (SGSNs) 217,219 and Gateway GPRS (General Packet Radio Service) Support Nodes (GGSNs) 221,223. SGSNs are network routing elements serving the attached mobile subscribers and dealing with mobility management, authentication, ciphering, generating charging records. GGSNs are similar to connect to a number of SGSNs but provide the interface to external networks via operator's IP backbone network tunnelling and routing packets of data to and from the correct SGSN and provide the interfaces to external packet data networks such as IP orX.25.

The communication between the RNCs, SGSNs and GGSNs in UMTS do not use TCP but uses the User Datagram Protocol (UDP) as the protocol for the Transport Layer.

UDP differs from TCP in several ways. TCP is a connection oriented protocol where a connection is set up at the beginning of a data flow and torn down at the end of the data flow. The connection setup will be used by all data packets of that data flow. In addition TCP prescribes end to end acknowledgement feed back where the ultimate destination returns acknowledge messages to the originating source to indicate that data packets have been correctly received. If a negative-acknowledgement or no acknowledgement is received instead, the source will retransmit the corresponding data packets. The TCP protocol is a very reliable protocol but has a very long delay.

UDP in contrast is a connectionless protocol where there is no setting up or tearing down of a connection for a data flow. It is an unacknowledged protocol with no feedback between destination and source. In effect a UDP data source will blindly transmit data packets without receiving any feedback or having any control over which route the data packet travels to its destination. Consequently it is a less reliable protocol than TCP but provides lower delay, higher flexibility and smaller overhead.

In order to keep the cost of the network down it is desirable to include as few network elements as necessary. Consequently it is advantageous to have as equally distributed load of each network element as possible. However, due to variations in the traffic there may be a significant variation in the traffic routed by one network element while other alternative network elements are under used. Furthermore, traffic may vary significantly with time and during some time periods the network may be used close to full capacity or even be under dimensioned for a peak traffic load. Consequently, it is important to have effective means for dealing with congestion avoidance and congestion control both of individual network elements and for the network as a whole.

However, unacknowledged protocols like UDP where the data source transmits data packets without consideration of the status of other network elements does not provide efficient congestion avoidance or control as there are no means for controlhng the source transmissions. The source will not expect any response after sending the previous data, therefore, and has no idea of what the situation at destination is. In other words, the source always assumes that the destination is capable of accepting the data it sends, and therefore congestion occurs when the volume of data traffic flows go above what the destination can handle. As a result, a need exists to provide a communication network using an unacknowledged protocol wherein at least some of the above mentioned disadvantage(s) may be alleviated.

Summary of the Invention

The invention seeks to provide a communication network using an unacknowledged transport layer data protocol with improved congestion avoidance, mitigation and control.

Accordingly there is provided a network element for a communication network as claimed in claim 1.

According to a second aspect of the invention, there is provided a communication network as claimed in claim 6.

According to a third aspect of the invention, there is provided a method of congestion control for a communication network as claimed in claim 17.

Further advantageous features are as claimed in the dependent claims.

Brief Description of the Drawings

An embodiment of the present invention is described below, by way of example only, with reference to the Drawings, in which:

FIG. 1 is an illustration of a cellular communication system according to prior art; FIG. 2 is an illustration of a simplified network architecture for the packet switched part of a UMTS network according to prior art;

FIG. 3 is an illustration of a communication network in accordance with an embodiment of the invention.

FIG. 4 is an illustration of the probability of dropping packets in response to the number of packets in a queue

FIG. 5 is an illustration of two network elements in accordance with this embodiment of the invention

Detailed Description of a Preferred Embodiment

The following description focuses on an embodiment compliant with a UMTS cellular communication network using the Internet Protocol and User Datagram Protocol but it will be apparent that the invention is not limited to this application. For example, the invention is equally applicable to GPRS systems.

FIG. 3 is an illustration of a communication network 300 in accordance with an embodiment of the invention.

The communication network comprises a number of network element 301, 303, 305, 307 , 309. In the present UMTS embodiment these network elements maybe RNCs, GGSNs, SGSNs or other suitable network elements.

A first network element 305 has a receiver 311 for receiving data from a number of transmitting network elements 301, 303 and a transmitter 313 for transmitting data to receiving network elements 307, 309. The network element in this embodiment performs a routing function wherein the data received from the transmitting network element 301, 303 are routed to the different receiving network elements 307, 309 dependent on their destination. Hence, the transmitting means also comprises scheduling and routing means operable to schedule the transmission of data packets on the connections to the appropriate receiving network elements 307, 309. In the present embodiment, the scheduling is performed by having a data packet queue for each connection or port corresponding to a receiving network element. The routing is performed by determining the destination address of incoming data packets and directing the data packet to the queue corresponding to receiving network element 307,309 which is the destination for the next hop.

The communication from the transmitting network elements 301, 303 to the first network element uses an Unacknowledged Transport Data Protocol (UTDP) where there is no acknowledgement of transmitted data packets being received at the destination.

The communication from the first network element to the receiving network element may use other data protocols but in the preferred embodiment it uses the same UTDP as the communication with the transmitting base stations.

Specifically, in the present embodiment the UTDP used is the User Datagram Protocol (UDP), which has been specified for UMTS in the 3G Partnership Project standardisation forum. Further details can be found in the standards 3G TS 23.060 V3.8.0 (2001-06) and 3G TS 23.060 V4.1.0 (2001-06).

A problem with a UTDP such as UDP, is that the network element transmitting a data packet does not receive any feed-back of whether the data packet has been correctly received. In case of a congested network or network element, packets maybe dropped and thus never reach the destination. However, the network element transmitting the information will not receive notification of the loss of the data packet and will thus not be able to take any action to compensate or relieve the effect. Hence, there is no flow control available, which in some systems provide an end-to-end flow control where the end destination feed back flow information to the original source. This feed-back is implemented at the user level i.e. at the application layer or the transport layer. However, this flow control is inefficient as it is only possible for end-to-end flow control on a per flow basis and thus the congestion relief of the hops between intermediate network elements is very limited.

In accordance with the embodiment of the invention, the first network element further comprises a congestion processor 315 for determining a congestion level for the network element for UTDP data packets. This congestion level is in the simplest embodiment determined by detecting how full the queues of the transmitting means 313 are, and if it is above a threshold the corresponding port (connection) of the network element is determined to be congested. The congestion level of the network element may thus relate to any measure of congestion of the network element such as individual congestion of one or more queues or a general measure of the total congestion factor. Hence, a network element may for example be considered congested if only one of the queues are congested or if the general level of a plurality or all queues are above a certain level. It will be clear to the person skilled in the art that any algorithm for determining congestion of any number, combination or weight of the queues of the network element can be used without subtracting from the invention.

The first network element further comprises means 317 for generating a transport layer congestion feedback message in response to the congestion level. In the simplest form this simply comprises in a message indicating that the first network element is congested if the congestion processor indicates that a queue has exceeded a threshold. As the protocol for the transport layer is an unacknowledged protocol, it does not contain signalling means for feeding back transport layer information. However, in accordance with the embodiment of the invention, the first network element comprises a message transmitter 319 for sending the transport layer congestion feedback message to the transmitting network elements using a higher layer signalling data protocol than the transport layer. Hence, the first network element composes a message containing congestion information and sends this messaging back to the transmitting network elements as a signalling message at a higher layer protocol.

Upon receiving the signalling message, the transmitting network element 301 ,303 decodes it and detects that it is a message relating to the transport layer. It then takes the appropriate action to compensate or alleviate the congestion. In the simplest form, this will include sending packets to other network elements rather than the first network element 305 in order to route data packets to their destination along a route not involving the congested first network element 305. It may also include slowing down the transmission of this particular class of traffic at the transmitting network element by adjusting the weights using suitable load sharing algorithms. Furthermore, It may also, for example, drop packets according to a suitable algorithm.

Specifically in the preferred embodiment wherein the UDTP is UDP and GTP tunnelling protocols are used over UDP as specified in 3GPP, the transport layer congestion feedback message is communicated using the signalling protocol of a GPRS Tunnelling Protocol (GTP-C) or the Radio Access Network Application Protocol (RANAP). These protocols are well known to the person skilled in the field and are specified for use in UMTS by 3GPP.

In the preferred embodiment, the congestion feedback message is sent to all transmitting network elements but in other embodiments it may be sent to only a selected set of transmitting network elements, for example the specific network elements which have sent data packets meeting specific criterion within a certain time interval e.g. only to the network elements which have sent data packets to the congested queue of the first network element within e.g. the last second.

A specific example of a more advanced method of detecting a congestion level is the Random Early Detection algorithm. This algorithm monitors the current load of a queue and determines the probability of dropping packets in response to the input and output rate. FIG. 4 is an illustration of the probability of dropping packets in response to the number of packets in a queue or buffer.

FIG. 4 illustrates that when the number of packets awaiting transmission in a queue is less than a given minimum threshold, the probability of dropping packets within a given time interval is zero assuming the current input and output rates. Furthermore, FIG. 4 illustrates that when the number of packets in the queue is above a maximum threshold the probability of dropping packets is virtually 1. Between the thresholds, the probability of dropping data packets is a continuous, incremental function of the loading of the queue. The probability of dropping packets is an indication of the congestion level of the queue and thus the network element and in this embodiment the congestion feedback message may contain the probability of packet droppings or alternatively and or additionally the loading of the queue i.e. how full the queue buffer is.

In another embodiment, the dropping of data packets may be graduated depending on the service class of a communication. For example, the data may be divided into four categories of service:

1. Background service

2. Interactive service

3. Streaming service

4. Conversational Service.

In this example the delay sensitivity increases from a lower sensitivity of a background service to a higher sensitivity of a conversational service and hence the routing should preferably prioritise the services with higher service number according to this list.

By varying the maximum and minimum threshold values among the different queues, one can alter the reaction to the congestion in an individual queue in comparison to other queues. FIG. 5 is an illustration of two network elements 301, 305 in accordance with this embodiment of the invention. For simplicity, only the components relating to transmission from one network element to one other element are shown. The transmitting network element 501 comprises four queues 501, 503, 505, 507 one for each service class and a transmitter scheduler 509, which is responsible for scheduling data packets from the different data packets in order to meet the quality of service required for the different service classes. The scheduler achieves this by weighting the packets in the different queues in accordance with the relative priority of each service class. In a congested network element it will not be advantageous to transmit all data packets from one service class before transmitting any data packets from other service class queues, as this could result in some service classes not receiving any service. A better approach is to weight the different queues such that data packets are transmitted from all queues but more data packets are transmitted from higher prioritised services or e.g. from services having higher delay sensitivity.

The data packets on the output of the transmitting scheduler 509 are transmitted to the corresponding first network element 305. This network element 301 also has a separate queue 511, 513, 515,517 for each of the four service classes. Thus each data packet transmitted from queue 501 is of service class 1 and will therefore end up in queue 511. Similarly, queue 513 corresponds to queue 503, queue 515 to queue 505 and queue 517 to queue 507. The first network element 305 similarly has a first network element scheduler 519, which operates in a similar fashion to the transmitting scheduler 509.

When the first network element approaches congestion it will begin to drop packets according to a dropping algorithm. Similarly to the scheduling, it is preferable to drop packets across the different service classes in relation to the relative delay sensitivity requirements of the different services. For example the overall required packet loss rate for service class 1 may be 0.01 and for service class 2 may be 0.001.. By dropping packets from both even when the service requirements are met a graceful deterioration maybe possible by a balanced packet dropping across the different classes of service. A simple dropping algorithm is to drop packets depending on the load of each queue such that if the appropriate service class queue is loaded above a given threshold packets are dropped and if below the threshold the data packets are accepted. In this case the threshold would be different for the different service classes in accordance with the quality of service requirements. However, any suitable dropping algorithm can be used without detracting from the scope of the current invention.

In accordance with the embodiment of the invention, the first network element comprises a congestion processor 315 for determining a congestion level for the network element for UTDP data packets, means 317 for generating a congestion feedback message in response to the congestion level and a transmitter 319 for sending the transport layer congestion feedback message to the transmitting network element 305 using a higher layer signalling data protocol than the transport layer.

The congestion processor 315 generates a congestion level relating to each of the plurality of service classes and the congestion feedback message includes information relevant to each service class individually. Hence, in the simplest embodiment the queue load for each of the service class queues are measured and the information included in the congestion feedback message. The transmitting network element 301 receives this information and from this information and knowledge of the dropping algorithm used in the first network element 305 it is able to adjust the weights the different service class queues for the scheduling algorithm so that a more balanced load of the individual queues in the first network element can be achieved. Thus if the congestion feedback message reports that the service class 2 queue is almost full whereas the service class 3 queue is almost empty, it can adjust the weights such that more data packets from its service class 3 queue are scheduled for transmission and less data packets from its service class 2 queue are scheduled. This will substantially reduce the probability of the queues of the first network element overrunning due to short term peaks or bursts. It will result in a more balanced load of the network elements and consequently of the communication network and therefore result in increased throughput and less risk of congestion and resultant data loss.

Furthermore, the congestion feedback message received from one network element may be forwarded to another network element so that this network element can participate in congestion avoidance. For example, the transmitting network element may further forward the congestion feedback message received from the first network element to network elements sending data packets to the transmitting network element. Similarly, the first network element may include information in the congestion feedback message from any congestion feedback messages received from receiving network elements.

According to an embodiment of the invention, the network will comprise network elements, which are operable to act upon congestion feedback messages and network elements, which are not. This situation may arise when the congestion feedback message is introduced as a feature in an existing network. In this embodiment the first network element may receive some data flows, which are responsive to the congestion feedback message and some which are not. In accordance with this embodiment of the invention, the communication network comprises means for detecting an unresponsive data flow, which is unresponsive to the congestion feedback messages, and means for dropping packets of the unresponsive data flow upon congestion. Typically, these means will be part of the first network element and implemented by detecting if any changes results from the transmission of congestion feedback messages. If data packets form a specific data flow continues to arrive at a rate which substantially remains unchanged, it is considered that this is due to the flow being unresponsive. In this case the dropping policy is changed such that relative more packets are dropped from this flow than if it had been designated as a data flow responsive to the congestion feedback messages.

The description of the communication network herein is given by way of example only, and it will be clear to the person skilled in the art that many variations can be considered without detracting from the invention. Specifically, the transmitting, receiving and first network elements may be identical, similar or substantially different. Furthermore, a given network element may be interconnected to the first network element such that it acts both as receiving and transmitting network element. It will also be clear that the separation of the functionality of the network elements into the described means are for illustration only and that these may be separated or combined in any suitable manner without detracting from the invention.

Specifically, the transmitter for sending the transport layer congestion feedback message to the transmitting network elements using a higher layer signalling data protocol than the transport layer may be the same as the transmitter for transmitting data to receiving network elements.

The network elements may be implemented in any suitable manner to provide suitable apparatus. As such the required adaptation may be implemented in the form of processor-implementable instructions stored on a storage medium, such as a floppy disk, hard disk, PROM, RAM or any combination of these or other storage media. Furthermore, theindividual parts of the network elements may be implemented in the form of hardware, firmware, software, or any combination of these.

It will be understood that the invention among others tends to provide the following advantages singly or in any combination:

• Throughput will be increased for dynamic traffic resulting in increased overall network capacity

• Data Packet traffic will be smoothed to reduce sensitivity to peaks caused by short term bursty traffic.

• Queue buffer sizes can be reduced.

• Differentiated quality of service can be achieved.

Claims

1. A network element for a communication network operable to communicate with a plurality of other network elements comprising means for receiving data packets from other transmitting network elements in accordance with an Unacknowledged Transport Data Protocol (UTDP); means for transmitting the data packets to other receiving network elements; means for determining a congestion level for the network element for UTDP data packets; means for generating a transport layer congestion feedback message in response to the congestion level; and means for sending the transport layer congestion feedback message to at least one of the other transmitting network elements using a higher layer signalling protocol than the transport layer.

2. A network element as claimed in claim 1 wherein the Unacknowledged Transport Data Protocol is a User Datagram Protocol.

3. A network element as claimed in claim 1 or 2 wherein the transport layer congestion feedback message contains information specific to a service class.

4. A network element as claimed in any of the claims 1 to 3 wherein the network element comprises separate queues for different classes of service and separate transport layer congestion feedback information is generated for separate queues and included in the transport layer congestion feedback message.

5. A network element as claimed in any of the claims 1 to 4 wherein separate transport layer congestion feedback information is generated for separate traffic flow and included in the transport layer congestion feedback message.

6. A communication network comprising at least a first network element as claimed in any of the previous claims.

7. A commumcation network as claimed in claim 6 further comprising at least a first transmitting network element

8. A communication network as claimed in claim 7 wherein the first transmitting network elements comprises a scheduler operable to alter the number of UTDP data packets scheduled for transmission to the first network element in response to the transport layer congestion feedback message received from the first network element.

9. A communication network as claimed in claim 7 or 8 wherein the first transmitting network elements comprises means for forwarding the transport layer congestion message to a network element transmitting UTDP data packets to the first transmitting network element.

10. A communication network as claimed in any of the claims 7 to 9 as dependent on claims 3 or 4 wherein the first transmitting network element is operable to modify the relative prioritisation of the transmission of UTDP data packets belonging to different service classes in response to the transport layer congestion feedback messages corresponding to the different service classes.

11. A communication network as claimed in any of the claims 7 to 10 comprising means for detecting an unresponsive data flow which is unresponsive to the transport layer congestion feedback message.

12. A communication network as claimed in claim 11 further comprising means for dropping packets of the unresponsive data flow upon congestion.

13. A cellular communication system comprising at least one communication network as claimed in any of the previous claims 6 -12.

14. A cellular communication system as claimed in claim 13 wherein the communication system is a UMTS commumcation system

15. A cellular communication system as claimed in claim 13 or 14 wherein the higher layer signalling protocol is a General Packet Radio Service Tunnelling Protocol (GTP)

16. A cellular communication system as claimed in claim 13 or 14 wherein the higher layer data protocol is a Radio Access Network Application Protocol (RANAP)

17. A method of congestion control for a communication network comprising the steps of: receiving data packets from other transmitting network elements in accordance with an Unacknowledged Transport Data Protocol (UTDP); transmitting the data packets to other receiving network elements; determining a congestion level for the network element for UTDP data packets; generating a transport layer congestion feedback message in response to the congestion level; and sending the transport layer congestion feedback message to at least one of the other transmitting network elements using a higher layer signalling protocol than the transport layer.

18. A method of congestion control as claimed in claim 17 wherein the Unacknowledged Transport Data Protocol is a User Datagram Protocol.

19. A method of congestion control as claimed in claim 17 or 18 wherein the transport layer congestion feedback message contains information specific to a service class.

20. A method of congestion control as claimed in any of the claims 17 to 19 wherein separate queues are held for different classes of service and separate transport layer congestion feedback information is generated for separate queues and included in the transport layer congestion feedback message.

21. A method of congestion control as claimed in any of the claims 17 to 20 separate transport layer congestion feedback information is generated for separate traffic flows and included in the transport layer congestion feedback message.

22. A method of congestion control as claimed in any of the claims 17 to 21 wherein the higher layer signalling protocol is a General Packet Radio Service Tunnelling Protocol (GTP)

23. A method of congestion control as claimed in any of the claims 17 to 21 wherein the higher layer data protocol is a Radio Access Network Application Protocol (RANAP)