DYNAMIC CimiEKING MANAGEMENT

HEID OFTHE FϊϊSENTINVENTION: The p re sent invention generally re late s to communication networks and more specifically to the formation and management of plurality of nodes of a c o mmunic a tio n syste m into se If o rg a nize d c luste rs.

BACKGROUND OFTHE FBESENT1NVENηON: Today, in the communication field, element management within a group comprising a large number of such elements is critic a L For example, for Peer to Peer (P2P) or Ad -ho c networks, such as sensor networks, or me shed WM networks, one basic technology is to regroup a large numb e r o f the se elements into one ormore groups, also called here after clusters. Clustering contributes to performance enhancement of the elements as a group, and service quality improvement for one ormore applications carried outby the elements.

An example of such a network is a network comprising fire probes disseminated throughout a forest. Such probes may work as cooperative nodes in a wire Ie ss ne two rk fo r fire prevention by organizing the mse lve s into clusters of nodes. Wheneverone of the fire probes fails due to a battery failure or so me otherreasons, the related cluster needs to reorganize. Keep alive messages are constantly exchanged between nodes of a same cluster for the organization and management of said cluster. Another example maybe inse c ts mo nito ring through a networkof probes disseminated in a forest. One possible scenario for such networks is that the re isno pre-existing network infrastructure. The elements have therefore to establish communication links within the network to exchange information among themselves when carrying out a network application. A clustering method is thus needed to organize the networkof no d e s. Initial management methods included manually defined clustering or static

clustering. These methods were limited in e ffic ie nc y a s the cluster could not adapt easily to dynamic change.

Self organized clustering implies that the nodes can reorganize themselves following a triggering event with limited if no external intervention, the resultant c luster infra structure evolving overtime as a consequence. Nodes use exchanges of messages and follow clustering methods to (re)organize the c luste rs o f the network.

Self organized cluster management methods are known e.g. from US 7,171,476 B2 or from US 6,876,643 Bl. One major problem with the known management methods is that a large number of message exchange (message overhead) is needed to manage the clusters in a network. Furthermore, most methods are not scalable, Le. whe nevermore nodesjointhe same cluster, the dynamic management is no longer e ffic ie nt.

Anotherissue is that the cluster management needs to be robust to dynamic changes, such as the change of a cluster he ad, whatever the size of the clusteris. There is still a need today for an improved self organized cluster management, that is robust to dynamic changes in the nodes characteristics, and scalable in that the message overhead may be controlled as the number of nodes grows. There is also a need for improved self organized clusters in a network of no d e s.

StMMABy OFTHE FKBSBNTMEIBOD AND SYSTBM:

It is an objective of the present method and system to overcome disadvantages and/ormake improvements in the priorart.

Tb that extend, the present method relates to a method for self organizing a plurality of nodes of a communication network into atleastone clusterof nodes, the plurality of nodes comprising atleastone potentialclusterhead, and being able to communicate within the communication network over a plurality of broadcast communication channels, said method comprising for said atleast one potential c luster he ad the acts of:

-broadcasting an advertisement message over a first communication channel, said advertisement message comprising:

-first information related to the capacities of the potential cluster head with regards to a given application, -second information related to a second communication channel over which the plurality of nodes can exchange messages with said potential c luste r he a d ,

- receiving over the second communication channel a response message if any from a node, said response message comprising an indication that the node, based on the first information, hasselected the potentialclusterhead asthe head to a c luster comprising atleastsaid potentialclusterhead and said node.

In orderto establish a cluster, only one response message is needed. Sine e the response message is broadcasted, other nodes which have also selected the potentialclusterhead based on the first information do not need to send a similar message. All nodes having selected the potential cluster head will have the knowledge of the elected clusterhead through the first node that sent a response message.

The present method also relates to a method for self organizing a plurality of nodesofa communication network into atleastone clusterof nodes, the plurality of nodes comprising at least one potential cluster head, and being able to communicate within the communication network over a plurality of broadcast communication channels, said method comprising forone node of said plurality of no d e s the a c ts o f:

-receiving from the at least one potential clusterhead an advertisement message over a first communication channel, said advertisement message comprising:

-first information related to the capacities of the potentialclusterhead with regards to a given application,

-second information related to a second communication channel over which the node can exchange messages with said potentialclusterhead,

- selecting, based on said first info mi a tion, the potential cluster head as the head to a cluster comprising said potentialclusterhead and said node,

-broadcasting over the second communication channel a response message comprising an indication that said node hasselected the p o te ntia 1 c luste r he a d a s the head to the c luste r.

The present system re late s to a system for self organizing a plurality of nodes of a communication network into at least one cluster of nodes, the plurality of nodes comprising at least one potential clusterhead and being able to communicate within the communication network over a plurality of broadcast communication channels, said potentialclusterhead being arranged to :

- broadcast an advertisement message over a first communication channel, said advertisement message comprising:

-first information related to the capacities of the potentialclusterhead with regards to a given application, -second information related to a second communication channel over which the plurality of nodes can exchange messages with said potential c luste r he a d , at least one node begin arranged to :

- send over the second communication channel a response message, said response message comprising an indication that said node, based on the first information, has selected the potential cluster head as the head to a cluster comprising atleastsaid potentialclusterhead and said node.

BEflEFDESCKIPπON OF THE DRAWINGS: The p re se nt syste m and method are explained in further detail, and by way of example, with reference to the accompanying drawings wherein:

FlG.1 shows an exemplary embo diment of the present system, FlG.2 shows an illustration of the message flows among nodes and cluster he ads in accordance with an embodiment of the present system,

FlG. 3 shows a flow diagram illustrating an embodiment of the present me tho d ,

FlG.4 shows a flow diagram illustrating another embodiment of the present method, and: FlG.5 shows an illustration of a change of status of a node in the network.

DEEULBD DESCHPDON OFTHEFraFEKRED EMBODIMENTS:

The following are descriptions of exemplary embodiments that when taken in conjunction with the drawings will demonstrate the above noted features and advantages, and introduce furtherones.

In the following description, forpurposes of explanation ratherthan limitation, specific details are set forth such as architecture, interfaces, techniques, etc., for illustration. Ho we ver, it will be apparentto those of ordinary skill in the artthatother embodiments that depart from these details would stillbe understood to be within the scope ofthe appended claims.

For example, the invention allows an improved se If o rg a nize d management ofnodesina communication network. The man skilled in the a it will no tic e that this is notthe sole embodiment possible, and thatthe present system and method may be implemented other content such as a wireless network, a Local Area Network (IAN) or mo re generally any type of network, wire less or wired line, over which a plurality of nodesmay exchange messages.

Moreover, for the purpose of clarity, detailed descriptions of well-known devices, systems, and methods are omitted so as not to obscure the description of the present system. In addition, it should be e xp re ssly und e rsto o d thatthe drawings are included for illustrative purposes and do not represent the scope ofthe present system.

By node or element, one may understand any electronic device capable of exchanging messages and data overa communication network. Such device may be a portable or mobile device, or a fixed one. The messages may be exchanged over a wireless network or not, or a combination of both. Bluetooth®, Infrared,

WiMAX or WM technologies may be used e.g. for exchanging messages between nodes in the example a wire less network. The man skilled in the art may readily apply the teachings of the present method and system to any existing communication networks. The nodes are also capable of broadcasting messages over a plurality of c ha nne Is within the c o mmunic a tio n ne two rk.

The nodes are also capable of carrying a given application, which is the reason behind the clustering. An example of such application may be data transfer, physical parameter monitoring, ... Some nodes may have higher capacities, Le. abilities, in regards to such an application, when compared to other nodes in the network. These nodes will be described later on as potential master nodes or potential cluster head. Once elected as the head to a cluster, these nodes willbe referred to as c luste r he a d s. The othernodes willbe called here afterregularnodes or member nodes once the cluster has been formed or once they have joined a c luste r.

For simplification purposes, the method willbe illustrated here after for one given application. The method maybe readily transposed to clustering involving two or mo re applications. Nodes which are potential master nodes for a first application could be regularnodesfora second orthird application. FlG.1 shows an illustration of an embodiment of the present system. A group of network nodes 115 have self-organized into a cluster 110. Other nodes 125 and 135 have self organized into clusters 120 and 130 respectively. Eke h cluster willhave a clusterhead, respectively 116, 126, and 136.

The nodes from a communication network as seen in FlG.1 may be of three typ e s:

- a "regular" node 145 which have not joined a c luste rye t and which is bound to become a cluster member. It may become a cluster member when electing a potential cluster head 146 as cluster head. The formed cluster, like clusters 110, 120 and 130 in the illustration of FIG.1, then comprise sat least said regular node and the potential cluster head which have become respectively a cluster member and a

cluster head. Regular nodes 145 may also become a cluster member when joining an already existing cluster. The election mechanism, as described lateron, is based on advertisement messages (AM) sent by the potential cluster heads and/or the existing c luste r he a d s thro ug ho ut the network, - potential cluster heads 146. Potential cluster heads are determined upon the ir sp e c ific capacities in relation to the given application. How these capacities are established is external to the present clustering mechanism. These capacities may be seen as having a direct consequence on the performance of such a node with regards to the given application. For nodes such as electronic portable or mobile devices, the capacities may be forinstance the battery reserve that will allow the device to carry the application overa period of time longerthan other nodes. The capacity may also be, but not limited to , the data transferthe device is capable of, its processor characteristics, ...

- cluster heads 116, 126 or 136. These cluster heads correspond to potential cluster heads which have been elected previously. They are the head or master nodes to clusters (respective Iy 110, 120 and 130 in FlG.1) comprising the cluster he ad and at least one cluster member such as cluster member 115, 125 and 135 respectively.

When the clustering mechanism is initiated, all nodes are eitherregularnodes or potential cluster heads. All nodes will exchange periodical Advertisement Messages (AM) and Response Messages (RM) to form and maintain clusters using different broadcasting channels, as explained here after.

A first communication channel, called here after the network channel or global channel, is used to exchange messages between all nodes during the clusters formation. This channel is also used by the cluster heads and potential clusterheads to exchange messages.

More specifically, the globalchannelforthe whole group is used by: 1. clusterheads and potential clusterheads to communicate information, including the advertising about the ircapacitiesin respect to the application, 2. new nodes coming to the network and which want to join a cluster (regular

nodes) orstart a new one (potentialclusterheads), and;

3. former cluster members from a cluster and which have lost their cluster head. Such nodes may indeed be seen as new nodes.

The communication at each cluster level will use a dedicated communication channel, called here after the clusterchanneL In other words, each c luster is further characterized by a distinct cluster channel which is defined by each potential c luster he ad prior to being elected as cluster head. Tb do so , a potential cluster head will se Ie c t its o wn cluster channel, distinct from the global channel or the cluster channels already in use. The potential cluster head will advertise information related to this cluster channel over the global channel and wilUisten to response messages if any on this cluster c ha nne L The global channel and cluster channels, all of them broadcast channels, are control channel for the purpose of c luster formation and maintenance.

Regular nodes wishing to either join an existing cluster or elect a potential cluster head to form a new cluster, will listen to advertisements on the global channel, will elect an appropriate (potential) cluster head and respond using the cluster channel corresponding to the appropriate (potential) clusterhead. These nodes will then become members of the corresponding cluster.

The cluster formation and maintenance will now be illustrated in relation to FlGs.2 and 3.

In a preliminary act 300, the cluster formation is initiated. In this exemplary embodiment of the present method, it is assumed that no c luste r e xists already, and the nodes in the network are eitherregularnodes 215 orpotential clusterhead 216 as seen in FRJ.2. O the r sc e na rio s, such as a regular node joining an existing cluster, ora clusterhead sending maintenance messages, will be illustrated lateron.

A parameter, here after called the Cluster Member Index (CMl), may be defined to keep trackof a node current status. The C MI may be a number with three values to represent the three types of members, for instance C IUSTEK_MEMBEK,

POTENTIAL_CIUSIER_HEAD and C EUSTER_HEAD. The CMIof each node, whetherat the initial act 300, or at any later stage of the cluster maintenance, may be set as

follows:

-for a potential cluster head: any member which "wishes" to become a c luster he ad, based e.g. on its capacities with regard to the given application, will set its CMIto a value corresponding to PO TENT1AL_C UMER_HEAD, -fora regularnode: a regularnode may set its CMIas C IIMERJVIEMBER if it does not wish to be a potential cluster head, e.g. its capacities with regards to the given application are limited. This may be the case if the b a tte ry is re a c hing a low level threshold, orits data transferrate is limited,

- for cluster head: a potential cluster head will become a cluster he ad only when elected by regular nodes. Rom that point on, a potential cluster head will change itsCMIto the C LIJSIER_HEA.D value. When a clusterhead isno longersought by its c luster members, e.g. it no longer receives response messages from any cluster member, it will switch back to potential cluster head, and change its CMI back to the value POTENηAL_CIJUSIER_HEA.D. h a further act 310, a potential clusterhead 216 will send an advertisement message (AM) over the global channel as seen in FlG.2. In order to advertise the other nodes 215 about its capacities in relation to the given application, in other words to enable regular nodes 215 to elect it as cluster head, the potential cluster head 216 will send in the AM: -a first information related to the capacities of this p o te ntia 1 c luste r he a d 216 with regard s to the given application. The first information maybe a directorindirect indication of the potentialclusterhead capacities. Forinstance, the first information maybe limited to an indication that the node sending the message dec lares itself as a potentialclusterhead. The first information may also correspond to more detailed indications of the c a p a c itie s, like data tra nsfe r, b a tte ry c ha rg e left, ...

-a second information related to a cluster communication channel over whichthe nodesmay exchange messages with this potentialclusterhead.

The second information may correspond for instance to a Channel ID (identification) tha t ind ic a te s the cluster channel over which cluster members can communicate with the potentialclusterhead once it is chosen as the clusterhead.

As mentioned before, the cluster channel is chosen by the potential cluster head among available channels. The cluster channel may be chosen e.g. randomly to avoid duplicated channels among clusters.

FIG. 2 illustrates the advertisement messages AM sent out over the global channelby a plurality of potentialclusterheads 216, 226, 236 and 246. ban additional embodiment of the present method, further information may be included in the AM:

- the sc o p e DG O f the AM. The sc o p e de fine s ho w far a way fro m the p o te ntial c luster he ad the AM is valid. Beyond that distance DG the AM may no longerbe read by a node. The distance may be measured in number of hops from the potential clusterhead. Initially, when the c luste r is no t fo rme d yet, the scope of the AM may be setto a large numberand willbe adjusted one e the clusterhead received one or more response messages from cluster members, as explained later on,

- an advertisement cycle TW that is the time period between two advertisement messages AM. The advertisement cycle maybe adjusted based on the overhead control as explained later in an additional embodiment of the present method. Other parameters such as remaining life of battery for instance may be taken into account in the d e finitio n o f TADV.

Once act 310 has been carried out, the potential cluster head will wait for response messages (RM) from regular nodes, which will signal through the RM that they have elected the potentialclusterhead as clusterhead.

In a further act 320, the potential clusterhead 216 receives over the second communication channel, Le. the c luste r c ha nne 1, a response message RM from a regular node 215 in the network. This response message comprises an indication that this node, based on the first information in the AM of the potential clusterhead has selected the potentialclusterhead asthe head to a cluster comprising atleastsaid potentialclusterhead and said node.

Indeed, upon receiving the AM from the potential cluster head, any regular nodes 215 may elect this potential cluster head as cluster head, using the first info rmatio n in the AM o f the p o te ntial c luste r he a d and its o wn se t o f e Ie c tio n rule s.

The set of rules for election of a cluster he ad may vary from one regular node to the other. The distance from the potential cluster head may be for instance of importance to some nodes, on top of the first information of the AM.

In an additional embodiment of the present method, the response message may include the distance from the elected cluster head. The distance from the clusterhead can be obtained from the AM received by the regularnode sending the RM using forinstance a TTL(Tlme Tb live) parameterin the AM, Le. a distance parameterindicating the distance from the potentialclusterhead. Rυm the RMs, a clusterhead may determine the clustersize C HJSIER_S]ZE, Le. the distance from said clusterhead to its farthest cluster member. This distance maybe seen as an uplink distance from the potentialclusterhead to the cluster member.

In an alternative embodiment of the present method, a do wnlink distance maybe defined from the clusterhead to the potentialclusterhead using a TTLin the RM. Upon choosing the clusterhead, a regularnode about to become a cluster me mb e r will re sp o nd to the cluster channel indie a ted in the AM from the potential clusterhead. The newly cluster member may then switch from listening to the global channel to listening to the c luste r c ha nne 1, a nd thus may respond only to that cluster channeLThe cluster member will thus receive a 11 AMs se nt b y the clusterhead onthe cluster channel as well as other response messages sent by cluster members from the same c luste r, a s the cluster channel is a broadcast channeL

In a further embodiment of the present invention, a selective suppression of the response messages may be carried out to reduce the message overhead. Every regular node which has elected the potential cluster head may schedule its response message with a random delay. The delay may randomized based on the advertisement cycle TXDV from the advertisement message so that all these nodes get a chance to send a RM be fore the ne xt AM fro m the potentialclusterhead.

As a 11 re sp o nse messages from the nodes in the same cluster a re sent on the c luste r c ha nne 1, and will be received by all members from this same cluster. The selective suppression may be applied as follows: upon receiving a RM from any

fellow cluster member (which have also elected the same potential cluster head) and whose distance to the elected cluster head islargerthan its own distance to said elected cluster he ad, the scheduled RM will be simply suppressed orcancelled. C EUSiliK_S]ZE will the n b e determined from the distance parameterin the RM which reports the largest distance from the elected cluster head to the furthest cluster me mb e r.

Inorderto facilitate the choice of a clusterhead, any regularnode may store a "AM table" for all the potential cluster heads it has received an AM from. That table may for example comprise the following data for each of these potential clusterheads:

- potentialclusterhead ID,

- first information in the AM related to the capacity of said potential cluster head,

- distance from the potentialclusterhead to itself. h a further act 330, once a response message has been received by the potentialclusterhead, it changes its status to clusterhead. This may be carried out through the C MI which is switched to the CLLJSTER_HEAD value.

In a further embodiment of the present method, the newly elected cluster head may start to send in an act 340 two advertisement messages both to the globalchanneland its own cluster channel every advertisement cycle TW:

1. a c luster advertisement message to itsown cluster channel with a scope Dc set to C LLBIiK-SIZE, the scope defining how farfrom the clusterhead the message may go. This message maybe seen as a keep alive message, to inform the cluster nodes that it is still an active clusterhead, 2. A clusterhead will send the cluster advertisement message to the global channel with a large scope DG. The use of the global channel allows communication with o the r c luste r he ads, aswellas with a ny o the r re g ula r no d e s wishing to jo in the existing cluster. This message comprises the same information as the AM sent in act 310 and may use the same large scope as the one from the AM in act 310. The two AMs on the clusterand global channel respective Iy are illustrated as

being sent simultaneously and with the same periodicity. The AM on the global channelmay be sent eve iy multiple ofthe advertisement time TADV, a s this AM is use d for any nodes wishing to join the existing cluster. These two AMs may also be sent a sync hro no usly in a n a lte ma tive e mb o d ime nt. The present method allows a simple and straightforward maintenance of a cluster. C luster members will response to the keep alive AM on the cluster channel in the same way they replied to the AM on the cluster channel during the cluster formation. These acts may be seen as acts 310 and 320. In other words, cluster members will reply to a cluster head for cluster maintenance in the same way regular node s respond to a potential cluster he ad once they have elected it during the c luster formation.

When a regular node, referred to here after as new comer, wishes to join an existing cluster, it will listen to the global channel for any AM. The advertisement messages on the globalchannelmay come from: - cluster heads to existing clusters, which keep on advertising on the global channel, as explained here before,

- potential c luster he ads advertising to create a new cluster. In both instances, this new comer will elect a cluster he ad, whether potential or not, based on the first information in the AM and its own election rule sand send a response message on the cluster channel correspond ing to the second information in the AM. This response message willbe one ofthe response messagesthe cluster head may receive as in act 320 on the clusterchanneL The additional embodiment ofthe suppression may also be carried out in the same way, as the new comerwill listen to the clusterchannelonce a clusterhead elected. New potential cluster head will start with advertising to the global channel with theirown predetermined time period, asdescribed in relation to acts 300 to 330. Regularnodes, which are not cluster members, will listen to the ne w p o te ntia 1 c luste r heads as well as other AMs from existing potential cluster heads and cluster heads and may chose to elect said potentialclusterhead as clusterhead. Cluster members, having hopped out of a cluster for any given reason, may

also listen on the global channel to the AMs of potential cluster heads as well as other AMs from existing potentialclusterheads and c luste r he a d s, just like a regular no d e Io o king to jo in a c luste r .

For some reasons, there maybe instances when a cluster member no longer receives any AM from its cluster channel for a predetermined time period (e.g. two or mo re advertisement cycles), in an additional embodiment of the present method, it will then switch to the global channel to elect a new potential cluster head or c luste r he a d a nd jo in a c luste r c o nse q ue ntly , just like the new comer illustra te d he re above. When a cluster head no longer receives any RM on its cluster channel for a predetermined time period (e.g. two ormore advertisement cycles), it will switch backto potentialclusterhead, e.g. change its CMI to POTENTlAL-CIUSrEILHEAD. It will sto p se nd ing AM o n its c luste r c ha nne 1, and will c o ntinue se nd ing AM o nly to the global channel with the same ora longertime p e rio d , just like any new potential c luste r he a d .

In an additional embodiment of the present method, the advertisement period may vary for a potential cluster he ad depending on the reception or not of RMs. Forinstance, a potentialclusterhead may start to send one AM to the global c ha nne Ie very advertisement cycle fora predetermined time period TiNrr(e.g. three a dve rtise me nt c yc Ie s, 3 TADV). The p o te ntia 1 c luste r he a d will liste n to its o wn c luste r channel for any response from cluster members. If no response is received afterTiNrr, it will remain as a potentialclusterhead and will send one advertisement message to global channel at a second predetermined time period Tp larger than the initial one TTNγγ( e.g. every two advertisement cycles 2 TADV). h a further embodiment of the present method, the potentialclusterhead may change its status to regular node if no RM is received after a third predetermined time Tt. This node may nevertheless, beyond TT, switch backto the status of potential clusterhead at regular intervals to checkif it can form a cluster. The change may also happen through externalmeansand/orfo Uo wing a change in its capacities (e.g. remaining battery life below a given level).

ha further embodiment of the present method, the first information in the AM maybe limited to the CMl The CMImaybe seen as an indication of the capacities of potential c luster he ad and clusterheads with regard to the application.

A regular node, the CMIof which is set to C EUSIER_MEMBER, may select from the AMs received on the global channel, for example using the AM table, the potentialclusterhead orclusterhead which is the closest.

As mentioned before with regard to the capacities with regards to the application, the CMIis set extemalto the present clustering mechanism.

The advertisement and response messages, AM and RM respectively, are considered as overhead traffic for cluster formation and maintenance. Measurements may be defined to evaluate the overhead of the present cluster formation, one example being the number of message SNRECHVED received by any node in the network during a given period of time TPREDπERMINED. This number of messages takes into account the facts that : - each cluster member will receive on the cluster channel AM from its cluster head, and some RM from fellow cluster members,

- potential cluster heads receive on the global channel only AM from all other potentialclusterheads and clusterheads,

-a cluster head will receive AM from all potential cluster heads, other cluster heads, and RM from its own cluster members.

Tb control overhead, each cluster he ad and potentialclusterhead may monitor the number of message NRECHVED received during TPREDπERMINED. A predetermined threshold maybe set. Once this threshold isreached, the advertisement eye Ie Tk>vmay be adjusted to bring down the number of messages NRBCHVED to below the threshold leveL Once the NRBCHVED of a potentialclusterhead orclusterhead hasreached the predetermined threshold, such potentialclusterhead orclusterhead may increase its advertisement cycle TAΌV to decrease the overhead. The adjustment of the advertisement cycle may alternatively be carried out independently among the potentialclusterheads and clusterheads. As explained earlier, the status of a network node, as defined by the CMI

parameter, is set external to the clustering mechanism, e.g. based on the node capacities with regards to the given application. The CMI of potential cluster heads and c luster members may be set and changed by external means based on the given application requirements. In a further emb o dime nt of the present method, a (potential) cluster he ad may be allowed to be both a (potential) cluster head and a cluster member of another c luster at the same time, In order to do so, a fourth status may be defined through an additional CMIvalue CIIJSTERJXJPLEX.

Such a membermay: - advertise to both global channel (as in act 310) and cluster channel (a sin act

340forthe keep alive message) asa (potential) cluster he ad,

- listen to itsown c luste r c ha nne 1 fo r ie sp o nse message, and

-listen to another cluster channel and send RM to that cluster channel of the clusterhead it has chosen from its AM table as explained before. FIG.4 showsan additionalembodimentofthe presentmethod and corresponds to the actcarried outbya regularnode orclustermember.

In a preliminary act 400, the cluster formation is initiated. In this exemplary embodimentofthe presentmethod, itisassumed thatno cluster exists already, and the nodes in the network a ie either regular nodes 215 orpotential clusterhead 216 as seen in EtG.2.

In a further act 410, a regular node receives from the at least one potential clusterhead an advertisement message over a first communication channel, ie. the global channel, said advertisement message comprising the first and second information as described in relation to E(G.3. In an additional act 420, the regular node will select, based on the first information comprised in the one ormore AM it has received, a potentialclusterhead asthe head to a cluster comprising said potentialclusterhead and said node. ha furtheract430, the regularnode wϋlbroadcast overthe cluster channel (as indicated from the second information) a response message RM to the elected cluster head. The regularnode becomes a cluster member to the cluster comprising at least

itself and the elected cluster he ad. The newly clustermemb erwill also switch to listening to the clusterchanneL

In a further act 440, the cluster member may reply to further AM (keep alive messages) from the clusterhead during cluster maintenance. Other cases, like a cluster member leaving its cluster, or a new regular node available to join a new clusteroran existing clusterhave been illustrated before.

A clustermembermay decide to change cluster affiliation as its election rules will Ie ad to the decision of hopping o ut o f its c urre nt c luste r. The "ex"-clustermember will resume its listening of the global channel, as all potential cluster heads and c luster he ads keep on sending AM at regular intervals on the global channel,, ban alternative embodiment, it may select a potential clusterhead based on its AM table (selecting forexample the next in line) even if the table hasnotbeen updated since the regular node, while a cluster member, was not listening to the global channeL FlG.5 shows an illustration of a change of status of a node in the network. A potential c luster he ad 510, once elected by at least one regularnode, will switch its status to clusterhead 520. A cluster head 520, which no longer receives RM from cluster members, will switch back to the potential cluster head status after a predetermined time. In an additional embodiment of the present method, a potential cluster he ad

510 may choose to stay as potential clusterhead status a swished, or may choose to switch to the status of a regularnode 530 (and may become a cluster member) if its capacities has changed so asnotbeing able to serve as potential clusterhead, or may choose to leave the system completely at wflL . A regularnode 530, once its cap a cities imp roved to having the potential and will to be a potential cluster head, may change its status to potential cluster head through external means.

The nodes of the communication network comprise one or mo re processors with p o rtio ns to carry out the present method. The plurality of nodes of the present

communication network forms a system wherein the present method may be carried out.

Obviously, readily discernible modifications and variations of the p ie se nt syste m are possible in lightofthe above teachings. It is the ie fo re to be understood thatwithin the scope of the appended claims, the present invention may be practiced otherwise than as specifically described herein. For example, while described in terms of hardware/ software components interactive Iy cooperating, it is contemplated that the invention described herein may be practiced entirely in software. The software may be embodied in a carrier such as magnetic or optical disks, or a radio frequency or audio frequency carrierwave.

Thus, the foregoing discussion discloses and describes merely exemplary embodimentsofthe present invention. Aswϋlbe und e rsto o d b y tho se skilled in the art, the p re sent invention maybe embodied in o the r sp e c ific forms without departing from the spirit or essential characteristic s thereof. Accordingly, the disclosure of the present invention is intended to be illustrative, but not limiting of the scope of the present invention, as well as other claims. The disclosure, including any readily discernible variants of the teachings herein, defines, in part, the scope of the foregoing claim terminology such that no inventive subjectmatterisdedicated to the public.

Forexample, several parallel clustering maybe carried outwhen the nodesfrom the communication network are capable of running more than one application. Furthermore, the nodes may be various electronic devices, even within the same cluster. A node may change status through the action of external means, such as a network operator, or based on itself, e.g. when the monitoring of a node parameter triggers a status change. A low battery may force a clusterhead to lower its activity, and switch to cluster member for the rest of the battery life. In case of a mobile cluster head, the clusterhead, showing a lower data transfer as a consequence to lower network coverage, may change its status to cluster member. Conversely, any improved networkcoverage may change the statusbackto potentialclusterhead. The section headings included herein are intended to facilitate a review but are not intended to limit the scope of the present system. Accordingly, the

specification and drawings are to be regarded in an illustrative mannerand are not intended to limit the scope ofthe appended claims.

In interpreting the appended claims, it should be understood that: a)the word "comprising" doesnot exclude the presence of otherelements or acts than those listed in a given claim; b) the word "a" or "an" preceding an element doesnot exclude the presence of a plurality of such elements; c) any reference signs in the claims do no t limit the ir sc o p e ; d) several "means" may be represented by the same item or hardware or software implemented structure or function; e) any of the disclosed elements may be comprised of hardware portions (e.g., including discrete and integrated electronic circuitry), software portions (e.g., computerprogramming), and any combination thereof; f) hardware portions may be comprised of one or both of analog and digital p o rtio ns; g) any of the disclosed devices orportions thereof may be combined together orseparated into furtherpo rtio ns unless specific ally stated otherwise; h) no specific sequence of acts or steps is intended to be required unless specifically indicated; and i) the term "plurality of an element includes two or more of the claimed element, and does not imply any particular range of number of elements; that is, a plurality of elements can be as few as two elements, and can include an immeasurable numberof elements.