WO2007102096A1

WO2007102096A1 - Message distribution in a communication network

Info

Publication number: WO2007102096A1
Application number: PCT/IB2007/050599
Authority: WO
Inventors: Wilhelmus F. J. Fontijn
Original assignee: Koninklijke Philips Electronics N.V.; Sinitsyn, Alexander
Priority date: 2006-03-07
Filing date: 2007-02-26
Publication date: 2007-09-13

Abstract

A communication network (100), such as an ad-hoc peer-to-peer network, comprises a number of network elements (101-113). A network element (103) comprises a first receive processor (203) for receiving messages from a first network element (101). A probability processor (205) determines whether to forward messages to a second network element (109) in response to a stochastic process having a probability parameter. The probability of each message being forwarded to the second network element (109) is dependent on the probability parameter. A transmit processor (207) transmits the messages determined to be forwarded to the second network element (109). A second receive processor (209) receives missed message indications from the second network element (109) and the modification processor (211) modifies the probability parameter in response to the missed message indications. The invention may allow a Gossip protocol distribution of messages for automatic and local adaptation to the current conditions and network configuration.

Description

Message distribution in a communication network

The invention relates to message distribution in a communication network and in particular, but not exclusively, to message distribution in an ad-hoc peer-to-peer network.

Ad-hoc peer-to-peer (P2P) networks are becoming more and more widespread, e.g. in sensor networks and general-purpose area networks. In such networks, a non- centralized control and management philosophy is applied and each network element has only local knowledge about connected network elements. Furthermore, the network configuration and topology can change dynamically with new nodes joining and leaving the network and connections continuously appearing and disappearing.

One aspect of these networks is that distribution of a message throughout the system is uncertain and can result in a high resource usage. For example, it is known that propagating updates in an ad-hoc peer-to-peer network is problematic. A technique known as flooding is the simplest way to ensure that a message reaches all targets in a network. Flooding requires each node receiving the message to forward it to all connected nodes. This ensures that the message is transmitted over all connections and thus reaches all nodes in a connected network. However, the approach introduces significant network traffic overhead and is wasteful in terms of communication resource.

A popular method of reducing the traffic is to switch to a Gossip protocol for the propagation of messages. In a Gossip approach, the propagation of a message from one node to another is probabilistic so that messages are communicated on a connection only with a certain probability. The probability of the message being forwarded by a node is given by the gossip factor. In a Gossip distribution, the message will thus be distributed through a subset of the current connections. However, as the number of connections between nodes is typically relatively large, most nodes are likely to receive the message from at least one neighbor node. As the gossip factor increases, the probability of each node receiving the message increases but so does the traffic overhead at the same time. Indeed, the flooding distribution approach may be considered as a subset of the Gossip distribution approach with the gossip factor set to one. Typical values of the gossip factor are around 0.6-0.9 and for highly connected networks, the number of communicated messages can be typically reduced by a factor of up to 2, while still reaching most nodes in most cases.

However, research has shown that in most randomly connected networks, and certainly in most typical real world networks, the networks tend to form islands of connectivity. In other words, connected devices are typically arranged in clusters that are well connected within themselves but with relatively few connections present between the clusters. When using a Gossip protocol in such a case, there is a significant chance that whole clusters are not reached. As a result, the gossip factor needs to be relatively high to ensure that most nodes are reached in most cases, but this substantially reduces the benefit of gossip protocols.

Hence, an improved approach would be advantageous and in particular a system allowing increased flexibility, reduced complexity, reduced traffic overhead, increased probability of reaching the target destination, improved adaptation to current conditions and/or improved distribution of messages.

Accordingly, the invention seeks to preferably mitigate, alleviate or eliminate one or more of the above-mentioned disadvantages singly or in any combination. According to a first aspect of the invention, a network element for a communication network is provided, the network element comprising: means for receiving messages from a first network element; determining means for determining whether to forward messages to a second network element in response to a stochastic process having a probability parameter, a probability of each message being forwarded to the second network element being dependent on the probability parameter; forwarding means for transmitting the messages determined to be forwarded by the determining means to the second network element; means for receiving a missed message indication from the second network element; and modifying means for modifying the probability parameter in response to the missed message indication.

The invention may allow improved performance in a communication network. In particular, message distribution may be improved in a communication network such as an ad-hoc and/or a peer-to-peer network. In particular, a Gossip protocol distribution approach may be improved. The network element may adapt the Gossip procedure to reflect the currently experienced conditions and may particularly adapt the message propagation characteristics so as to suit the current network topology and configuration. Specifically, the network element may adapt its operation to reflect whether it is likely to be a critical bridge element between different network clusters in a clustered topology. The adaptation may be performed in response to easily provided local information, and no centralized control or management is necessary. The invention may facilitate operation and management and/or reduce the complexity of the network.

The missed message indication may be an indication that the second network element has not received a specific message from any network element.

In an optional feature of the invention, an increasing probability parameter value corresponds to an increased probability of each message being forwarded, and the modifying means is arranged to increase the probability parameter value in response to receiving a missed message indication.

This may allow an improved network and in particular an improved message propagation in a communication network. In particular, it may allow the probability of messages being transmitted over critical connections to be increased and thereby improve the probability of the message reaching its intended targets (with only a marginal increase in the number of messages). For example, the feature may improve the probability of messages being propagated across critical bridges between different clusters.

In an optional feature of the invention, the message comprises a message sequence indication. This may facilitate and/or improve operation. In particular, it may facilitate and/or enable the second network element to determine that a message has been missed. The message sequence indication may be a sequential message number.

The forwarding means may be specifically arranged to include the message sequence indication in the message and/or the original source of the message may include the message sequence indication, and the network element may be arranged to forward the received message sequence indication.

The message sequence indication may be, for example, a message number or a time stamp. Specifically, if global time is available, or if message intervals are much larger than clock inaccuracies, time stamping may be used. In another optional feature of the invention, an increasing probability parameter value corresponds to an increased probability of each message being forwarded, and the modifying means is arranged to decrease the probability parameter value in response to not receiving a missed message indication for a message. This may allow an improved network and in particular an improved message propagation in a communication network. In particular, it may allow the probability of messages being transmitted over non-critical connections to be reduced, thereby reducing the traffic overhead of propagating messages. Specifically, it may allow a gossip factor for the connection to the second network element to adapt to the current conditions.

The reduction of the probability parameter may be limited by a lower threshold corresponding to a minimum allowable value of the probability parameter.

In a further optional feature of the invention, the network element also comprises means for storing a first message and for transmitting the first message to the second network element following a modification to the probability parameter if a missed message indication is received for the first message.

This may allow improved performance and increase the probability of the message reaching the second network element and/or reduce the traffic overhead associated therewith. The first message may be transmitted to the second network element following the receipt of the missed message indication despite the determining means previously determining that the first message should not be transmitted to the second network element.

In an optional feature of the invention, the forwarding means is arranged to transmit messages to a plurality of network elements; the determining means is arranged to determine whether to forward messages to different network elements in response to a stochastic process having different probability parameters for different network elements; and the modifying means is arranged to modify the probability parameter for each network element in response to a missed message indication from that network element.

This may allow improved performance and in particular an improved local adaptation to the specific and individual conditions associated with the different connections supported by the network element.

According to a second aspect of the invention, a communication network comprising a first network element, a second network element and a third network element is provided; the second network element comprising: means for receiving messages from the first network element; determining means for determining whether to forward messages to the third network element in response to a stochastic process having a probability parameter, a probability of each message being forwarded to the second network element being dependent on the probability parameter; forwarding means for transmitting the messages determined to be forwarded by the determining means to the third network element; means for receiving a missed message indication from the third network element; and modifying means for modifying the probability parameter in response to the missed message indication.

In an optional feature of the invention, the third network element is arranged to only transmit missed message indications to the second network element for messages which have not been received from any network element.

This may allow improved performance.

In another optional feature of the invention, the first network element is arranged to include a message sequence indication in the messages, and the third network element is arranged to transmit a missed message indication to the second network element in response to a detection of a missing message in the message sequence.

This may facilitate and/or improve operation. In particular, it may facilitate and/or enable the second network to determine that a message has been missed. The message sequence indication may be a sequential message number.

The forwarding means may be specifically arranged to include a locally generated message sequence indication in the message and/or the original source of the message may include the message sequence indication, and the network element may be arranged to include this message sequence indication simply by forwarding the received message.

In a further optional feature of the invention, the missed message indication indicates a number of missed messages, and the modification means is arranged to modify the probability parameter differently for different numbers of missed messages.

This may allow improved performance and in particular an improved and/or faster adaptation of the operation to the experienced conditions.

In a further optional feature of the invention, the communication network is a peer-to-peer network.

The invention allows particularly advantageous performance in a peer-to-peer network and may specifically allow network elements to automatically adapt to the current conditions without requiring centralized control and management. An improved message distribution may be achieved and in particular critical connections may be automatically identified and compensated.

In yet another optional feature of the invention, the communication network is an ad-hoc network.

The invention allows particularly advantageous performance in an ad-hoc network and may specifically allow network elements to automatically and dynamically adapt to the current conditions while they change. An improved message distribution may be achieved and in particular critical connections may be automatically identified and compensated.

According to another aspect of the invention, a method of operation for a network element of a communication network is provided, the method comprising the steps of: receiving messages from a first network element; determining whether to forward messages to a second network element in response to a stochastic process having a probability parameter, a probability of each message being forwarded to the second network element being dependent on the probability parameter; transmitting the messages determined to be forwarded by the determining means to the second network element; receiving a missed message indication from the second network element; and modifying the probability parameter in response to the missed message indication.

According to a further aspect of the invention, a computer program product is provided for executing the method described above. These and other aspects, features and advantages of the invention are apparent from and will be elucidated with reference to the embodiments described hereinafter.

Embodiments of the invention will be described, by way of example only, with reference to the drawings, in which:

Fig. 1 illustrates a communication network in accordance with some embodiments of the invention;

Fig. 2 illustrates an example of the network element in accordance with some embodiments of the invention; Fig. 3 illustrates a method of operation for a network element in accordance with some embodiments of the invention; and

Fig. 4 illustrates a method of operation for a network element in accordance with some embodiments of the invention.

The following description focuses on embodiments of the invention that are applicable to an ad-hoc peer-to-peer communication network. However, it will be appreciated that the invention is not limited to this application but may be applied to many other communication networks. Fig. 1 illustrates a communication network 100 in accordance with some embodiments of the invention. The communication network 100 is an ad- hoc peer-to-peer communication network without any centralized management or control. Furthermore, network elements or nodes may dynamically leave and join the communication network, and connections between the different network elements may continually change with new connections being established and existing connections disappearing. Specifically, the communication network 100 may be a wireless local area network, such as an ad-hoc mode of WiFi access networks or geographically spread sensor networks such as those deployed to monitor seismic activity, crop conditions, bird nesting colonies, logistics, battle theatre networks, etc.

In Fig. 1, the network is formed by a number of substantially identical network elements 101-113 and dynamic connections established therebetween. The connections between network elements are illustrated by double-edged arrows.

In some circumstances, it is advantageous to distribute a message to all network elements in the communication network. For example, an update patch may need to be distributed to all network elements 101-113. The message may be originally distributed from a single network element, such as a first network element 101.

A frequently used technique is known as flooding and includes the message being forwarded by all network elements on all connections. This will result in a reliable distribution to all network elements 101-113 but will also lead to a high traffic overhead. For example, in the system of Fig. 1, most network elements 101-113 will receive the message from three or more connected neighbors.

A more efficient technique is the Gossip technique wherein each network element 101-113 only forwards the message with a certain probability. In such a technique, the probability of all network elements receiving the message can be close to one, provided that the network elements are sufficiently interconnected and the probability of the network elements communicating the message is sufficiently high. For well-connected networks, a probability factor of around 0.6 - 0.9 tends to lead to a very reliable distribution.

However, most practical ad-hoc communication networks tend to cluster. For example, the system of Fig. 1 has a structure with a first cluster comprising the network elements 101-107 that are well-connected, and with a second cluster comprising the network elements 109-113 that are well-connected. However, there are only two connections between the two clusters, namely the connection between network elements 103 and 109 and the connection between network elements 105 and 111. Accordingly, if both network elements 103 and 105 decide not to transmit the update message originally distributed from the first network element 101 on these connections, the entire second cluster will not receive the message. In other words, the network elements 103 and 105 form critical bridges between the two clusters. Thus, the Gossip technique tends to lead to sub-optimal performance in clustered configurations. It would be advantageous to identify the nodes that function as bridges between clusters and modify the operation for these. In a centralistic system, one could try to build a connectivity trap tree (routing table) and thereby determine the bridge nodes. However, this will generate additional network traffic offsetting the gains from the Gossip protocol. Moreover, for dynamic ad-hoc networks (which reconfigure constantly), the connectivity map needs to be updated frequently and the network traffic overhead created by it will be incurred repeatedly. Finally, while the centralistic approach may work for networks of a limited size (say, less than 100 nodes), the cost and complexity of maintaining this approach in large networks is prohibitive. In the example of Fig. 1, the network elements are arranged to locally adapt their performance so as to reflect the current conditions in the communication network. The adaptation is based on locally available or provided information, and each individual network element can perform the adaptation autonomously and independently without considering the adaptation performed by other network elements. A global improvement in the performance of the Gossip technique can be achieved by local adaptation and without requiring any centralized operation, control or management. The approach is thus particularly advantageous in ad-hoc and/or peer-to-peer communication networks.

Fig. 2 illustrates an example of a network element in accordance with some embodiments of the invention. The network element is particularly a second network element 103 of the communication network 100 of Fig. 1 and will be described with reference thereto. It will be appreciated that all other network elements 101, 105-113 of Fig. 1 may be substantially identical to the second network element 103 and may operate in the same way.

The network element 103 comprises a network interface 201 which is arranged to communicate with other network elements of the communication network. For example, the network interface 201 may comprise a wireless transmitter and receiver operating in accordance with the technical specifications for a wireless local area network. In the specific example, the network interface 201 has currently established connections with four other network elements 101, 105, 107, 109 including a network element 109 in the second cluster. The network interface 201 is coupled to a first receive processor 203 which is arranged to receive messages from other network elements. In the specific example, the first network element 101 is the source of an update message which is to be distributed to all network elements of the communication network 100. Accordingly, the first network element 101 transmits the message to the second network element 103 (it will be appreciated that, due to the Gossip protocol used, the message may not be received directly from the first network element 101 but may be received via one or more intermediate network elements 107). When the update message is received by the network interface 201, it is fed to the first receive processor 203. The first receive processor 203 is coupled to a probability processor 205 which is arranged to determine whether the update message should be further propagated by the second network element 103. In the described example, the probability processor 205 is arranged to individually evaluate each connection of the second network element 103. Thus, the probability processor 205 independently determines whether the update message should be forwarded on the first connection, the second connection, etc. The decision for each connection may be independent and the parameters used to determine whether to forward the message can be different for the different connections.

The probability processor 205 determines whether to forward the message in response to a stochastic process. Thus, the determination of whether to forward the message on a given connection is probabilistic. The probability of forwarding the message on a given connection is determined by a probability parameter (such as a Gossip factor) and, in the example, the probability of the message being forwarded increases for higher values of the probability parameter.

As a simple example, the probability parameter may have a value between zero and one and can directly indicate the probability of transmitting the message. A typical range of values is from 0.6 to 0.9. In such an example, the probability processor 205 can generate at random a number between zero and one (with a uniform probability distribution) for each connection, and if the random number is lower than the probability parameter for the connection, the update message is transmitted on the connection and otherwise the update message is not transmitted on the connection.

The probability processor 205 is coupled to the transmit processor 207 which is further coupled to the first receive processor 203. If the probability processor 205 determines that the update message should be forwarded on a specific connection, the transmit processor 207 is informed of this. In response, the transmit processor 207 retrieves the update message from the first receive processor 203 and transmits it on the specific connection via the network interface 201.

Thus, the second network element 103 comprises functionality for supporting update message distribution using a Gossip protocol. The second network element 103 further comprises a second receive processor

209 which is coupled to the network interface 201. The second receive processor 209 is arranged to receive messages from other network elements. Specifically, the second receive processor 209 can receive missed message indications from other network elements.

A missed message indication is an indication that the network element has not received a specific message from any network element.

Specifically, an update file that is to be distributed to all network elements in the communication network can be divided into a number of individual update messages. A message sequence indication is included in the individual messages, for example, the update messages may simply be numbered sequentially. This may allow the individual network element to monitor whether it receives all the update messages associated with the update file. If a network element detects that it has not received a specific message, it can generate a missed message indication and transmit this to another network element to indicate that the message has not been received. Specifically, the network element can generate a missed message indication and transmit this to the network element from which a subsequent update message was received.

The second network element 103 comprises a modification processor 211 which is coupled to the second receive processor 209 and the probability processor 205. The modification processor 211 is arranged to modify the probability factor used by the probability processor 205 in response to receiving a missed message indication. Specifically, if a missed message indication is received that indicates that a connected neighbor of the second network element 103 has not received a specific message from the second network element 103 or from any other network element, this is an indication that the connectivity to the neighbor is relatively limited and that the distribution of the update message to this network element depends highly on it being forwarded by the second network element 103. Accordingly, the modification processor 211 increases the probability parameter so as to increase the probability that the second network element 103 will indeed forward the update message to the network element.

As a specific example, if the network were divided into two clusters with only one connection between the clusters, the above procedure would typically result in a probability property (gossip factor) of 1 on the bridge nodes, thereby ensuring that each message would cross the bridge.

As a specific example, the second network element 103 may receive messages MO, Ml, M2, M3, M4 but, due to the probabilistic determination, may transmit only messages MO, M2, M4 to a third network element 109 being part of the second cluster. Due to the current configuration of the communication network 100, the messages Ml and M3 can only reach the third network element 109 if they are forwarded by the network element 105 to the network element 111, as this is the only other connection between the first and the second cluster. Thus, if the probability evaluation of network element 105 results in only message Ml being forwarded to network element 111 (and from there through intermediate network elements 113 to the third network element 109), the third network element 109 will never receive update message M3.

However, the third network element 109 will receive update message M4 from the second network element 103. When it receives update message M4, it checks if all the previous update messages have been received and thus identifies that update message M3 has not been received. Accordingly, it generates a missed message indication for message M3 and transmits this to the second network element 103 from which it has received update message M4. In response to receiving the missed message indication, the modification processor 211 increases the probability factor in such a way that the probability of messages being transmitted on the connection between the second network element 103 and the third network element 109 is increased.

In the above example, it is assumed that the distribution delay for propagating a message throughout the communication network is lower than the interval between consecutive messages. Thus, if a message with a higher sequence number than the expected number is received, the individual network element can instantly determine that a message has been missed. It will be appreciated that, in some embodiments, the distribution delay may be higher than the message interval and that the individual network element may apply a delay before it determines that the message has been missed.

As another example, in some embodiments, the network elements 101-113 may be arranged to transmit the missed message indication if the message is not received within a given time interval (even if it is received after this time interval). The time interval may be, for example, a time interval from receipt of the previous message or a time interval that ends when a subsequent message is received.

This may allow improved performance in many embodiments. For example, if a part of the network is well connected but by a route that is long/slow and there are bridges from A to B that reach this part, a message may reach network element B from A before the message has reached A via the long route. This may serve as an incentive to increase the probability parameter for the bridge network element and decrease it on the long route. Thus, a network element detecting that it is on the far side of a bridge may autonomously adapt its probability parameter for the bridge connection. In some embodiments, the modification processor 211 can be arranged to decrease the probability parameter in response to not receiving missed message indications.

Thus, if no missed message indications are received for a specific connection, the probability parameter may gradually be reduced in order to reduce the number of messages being communicated in the system. Another indication that the probability parameter may be decreased may be if a specific message is received from multiple other nodes.

As a specific example, the network element 105 is connected to five other network elements of the first cluster and, accordingly, the probability parameter for connections to this network element 105 can be gradually reduced from the nominal probability in order to reflect the large number of possible paths to this network element 105. It will be appreciated that the increase and decrease of the probability parameter may be subject to specific limits so as to ensure an acceptable operational interval.

For example, the modification process 211 may limit the values of the probability parameter to be between 0.5 and 1. It will also be appreciated that the probability parameter may be modified gradually, for example, in relative steps and with a certain maximum rate determined, for example, in response to time or number of transmitted update messages. For example, the probability parameter may be updated in response to a missed message indication for the message as defined by the formula:

ProbPar = 1 - (0.9 * (1 - ProbPar))

Thus a low-pass filtering/averaging may be introduced. It will be appreciated that the dynamic characteristics of the updating of the probability parameter can be optimized for the individual embodiment.

Thus, the described system allows a local adaptation to be performed in the individual network element based on locally available information. The approach may allow a much improved Gossip distribution which automatically adapts to the current conditions and configuration and in particular adapts to taking critical connections into account. Specifically, it allows critical bridge nodes between clusters to identify themselves transparently and to adapt the operation accordingly.

In some embodiments, the network elements may further comprise functionality for temporarily storing update messages. If the network element receives a missed message indication from the neighbor network element, it may proceed to retrieve the missed message and forward this to the network element.

In the specific example described above, the second network element 103 may store all the update messages and when the missed message indication is received from the third network element 109 indicating that message M3 has not been received, the second network element 103 can immediately transmit this to the third network element 109, thereby immediately restarting the propagation of this message in the second cluster. To this end, the missed message indication may include an identification of the message that has been missed. In some embodiments, a missed message indication may indicate a number of missed messages and the modification processor 211 can modify the probability parameter differently for different numbers of missed messages. For example, the receiving node can report all messages it missed and the probability parameter for the connection may then be increased proportionally. Hence, if two messages have been missed, the probability parameter can be increased by a step with a step size double the step size used if only one message had been missed.

In some embodiments, a training mode may be implemented, wherein the probability parameters are initially set to very low values which are then automatically increased until a high coverage is achieved.

Figs. 3 and 4 illustrate methods of operation for a network element in accordance with some embodiments of the invention. The methods are applicable to the second network element 103 of Figs. 1 and 2 and will be described with reference thereto. Specifically, Fig. 3 illustrates elements of the method of operation of the first receive processor 203, the probability processor 205 and the transmit processor 207, and Fig. 4 illustrates elements of the method of operation of the second receive processor 209 and the modification processor 211. The methods may be performed continuously and in parallel.

In step 301, the first receive processor 203 receives an update message from the first network element 101. Step 301 is followed by step 303 wherein the probability processor 205 determines whether to forward the update message to the third network element 109 in response to a stochastic process having a probability parameter. The probability of each message being forwarded to the second network element is dependent on the probability parameter. Step 303 is followed by step 305 wherein the transmit processor 207 transmits the messages determined to be forwarded by the probability processor 205 to the third network element 109.

The method then iterates, i.e. step 305 is followed by step 301 wherein the next message is awaited. In step 401, the second receive processor 209 receives a missed message indication from the third network element 109.

Step 401 is followed by step 403 wherein the modification processor 211 modifies the probability parameter in response to the missed message indication.

The method then iterates, i.e. step 403 is followed by step 401 wherein the next missed message indication is awaited.

It will be appreciated that the described system provides a number of advantages including one or more of the following:

It improves application of Gossip protocols in real world networks.

It allows transparent identification of bridge nodes with almost no additional traffic.

It allows automatic adjustment to dynamically changing network topologies.

The application of localized adjustments ensures low additional message overhead.

No global network routing knowledge is needed. Automatic restart of died-out gossip is possible without additional traffic.

Extra reduction of network traffic may be achieved by allowing a decreased average Gossip factor (probability parameter) while still reaching most of the nodes most of the time.

It automatically takes alternative routes into account. It will be appreciated that embodiments of the invention have been clarified in the above description with reference to different functional units and processors. However, it will be apparent that any suitable distribution of functionality between different functional units or processors may be used without detracting from the invention. For example, illustrated functionality to be performed by separate processors or controllers may be performed by the same processor or controllers. Hence, references to specific functional units are only to be considered as references to suitable means for providing the described functionality rather than being indicative of a strictly logical or physical structure or organization. The invention can be implemented in any suitable form including hardware, software, firmware or any combination of these. The invention may be optionally implemented at least partly as computer software running on one or more data processors and/or digital signal processors. The elements and components of an embodiment of the invention may be physically, functionally and logically implemented in any suitable way. Indeed, the functionality may be implemented in a single unit, in a plurality of units or as part of other functional units. As such, the invention may be implemented in a single unit or may be physically and functionally distributed between different units and processors.

Although the present invention has been described in connection with some embodiments, it is not intended to be limited to the specific form set forth herein. Rather, the scope of the present invention is limited only by the accompanying claims. Additionally, although a feature may appear to be described in connection with particular embodiments, one skilled in the art will recognize that various features of the described embodiments may be combined in accordance with the invention. In the claims, use of the verb "comprise" and its conjugations does not exclude the presence of other elements or steps. Furthermore, although individually mentioned, a plurality of means, elements or method steps may be implemented by e.g. a single unit or processor. Additionally, although individual features may be included in different claims, these may be combined advantageously, and the inclusion in different claims does not imply that a combination of features is not feasible and/or advantageous. Also the inclusion of a feature in one category of claims does not imply a limitation to this category but rather indicates that the feature is equally applicable to other claim categories as appropriate. Furthermore, the order of features in the claims does not imply any specific order in which the features must be worked, and in particular the order of individual steps in a method claim does not imply that the steps must be performed in this order. Rather, the steps may be performed in any suitable order. In addition, singular references do not exclude a plurality. Thus references to "a", "an", "first", "second", etc. do not preclude a plurality. Reference signs in the claims are provided merely as clarifying examples and shall not be construed as limiting the scope of the claims in any way.

Claims

CLAIMS:

1. A network element (103) for a communication network (100), the network element (103) comprising: means (203) for receiving messages from a first network element (101); determining means (205) for determining whether to forward messages to a second network element (109) in response to a stochastic process having a probability parameter, a probability of each message being forwarded to the second network element (109) being dependent on the probability parameter; forwarding means (207) for transmitting the messages determined to be forwarded by the determining means to the second network element (109); means for receiving a missed message indication from the second network element (109); and modifying means (211) for modifying the probability parameter in response to the missed message indication.

2. The network element (103) of claim 1, wherein an increasing probability parameter value corresponds to an increased probability of each message being forwarded, and the modifying means (211) is arranged to increase the probability parameter value in response to receiving a missed message indication.

3. The network element (103) of claim 1, wherein the message comprises a message sequence indication.

4. The network element (103) of claim 1, wherein an increasing probability parameter value corresponds to an increased probability of each message being forwarded, and the modifying means (211) is arranged to decrease the probability parameter value in response to not receiving a missed message indication for a message.

5. The network element (103) of claim 1, furthermore comprising means for storing a first message and for transmitting the first message to the second network element (109) following a modification to the probability parameter if a missed message indication is received for the first message.

6. The network element (103) of claim 1, wherein the forwarding means (207) is arranged to transmit messages to a plurality of network elements; the determining means (205) is arranged to determine whether to forward messages to different network elements in response to a stochastic process having different probability parameters for different network elements; and the modifying means (211) is arranged to modify the probability parameter for each network element in response to a missed message indication from said network element.

7. A communication network (100) comprising a first network element (101), a second network element (103) and a third network element (109); the second network element (103) comprising: means (203) for receiving messages from the first network element (101); determining means (205) for determining whether to forward messages to the third network element (109) in response to a stochastic process having a probability parameter, a probability of each message being forwarded to the third network element (109) being dependent on the probability parameter; forwarding means (207) for transmitting the messages determined to be forwarded by the determining means to the third network element (109); means (209) for receiving a missed message indication from the third network element (109); and modifying means (211) for modifying the probability parameter in response to the missed message indication.

8. The communication network (100) of claim 7, wherein the third network element (109) is arranged to only transmit missed message indications to the second network element (103) for messages which have not been received from any network element.

9. The communication network (100) of claim 7, wherein the first network element (101) is arranged to include a message sequence indication in the messages, and the third network element (109) is arranged to transmit a missed message indication to the second network element (103) in response to a detection of a missing message in the message sequence.

10. The communication network (100) of claim 7, wherein the missed message indication indicates a number of missed messages, and the modification means (211) is arranged to modify the probability parameter differently for different numbers of missed messages.

11. The communication network (100) of claim 7, wherein the communication network is a peer-to-peer network.

12. The communication network (100) of claim 7, wherein the communication network is an ad-hoc network.

13. A method of operation for a network element ( 103) of a communication network, the method comprising the steps of: receiving (301) messages from a first network element (101); determining (303) whether to forward messages to a second network element (109) in response to a stochastic process having a probability parameter, a probability of each message being forwarded to the second network element (109) being dependent on the probability parameter; transmitting (305) the messages determined to be forwarded by the determining means to the second network element (109); receiving (401) a missed message indication from the second network element (109); and modifying (403) the probability parameter in response to the missed message indication.

14. A computer program product for executing the method of claim 13.