DYNAMIC AND JOINT IMPLEMENTATION OF NODE GROUPING AND GROUP CONFIGURATION SELECTION IN AN AD-HOC MOBILE COMMUNICATION NETWORK

Description

Technical field:

[0001] The invention lies in the field of mobile ad hoc wireless communication networks. Applications using ad hoc networks cover a very wide range: conferences, events, emergency operations (fires, earthquakes, natural disasters, etc.), exploration missions, and more.

[0002] The invention relates more specifically to the clustering and cluster configuration aspects implemented in these networks and enabling the nodes to adapt in real time to the various fluctuations relating to radio channel conditions, traffic, missions and the different movements of the nodes themselves or their surroundings.

[0003] Clustering refers to the grouping of nodes together, creating one or more clusters (or groups) of nodes. This involves selecting nodes from within the ad hoc network to form a group based on predefined criteria. For example, N is in the range [20-30] and g is in the range [4-5]. N can take various values, in the hundreds or thousands. The value of g depends on the waveform, the traffic to be handled, and more generally, the associated services. Grouping nodes into clusters, each associated with a cluster leader, offers several technical advantages. For instance, it limits data exchange and the size of routing tables, and reduces the impact of mobility on the entire topology.

[0004] Since spectral resources are limited, it is often necessary to select the cluster configuration, in particular the logical channel (a logical channel being an allocated time-frequency resource on which nodes can exchange information) on which the nodes of a cluster communicate, according to a number of criteria, for example to manage spatially the frequencies / logical channels, for example to manage spatial reuse (there may sometimes be fewer logical channels than clusters), while maintaining a sufficiently low level of interference to allow good quality of communications.

[0005] Configuration in this document means the assignment to a cluster of a set of cluster attributes including at least its logical channel (its color); other attributes can also be selectively assigned to cluster nodes during the configuration operation, such as, for example and not exhaustively, the scheduling of communications between cluster nodes and the allocation of resources (transmission power, type of coding and / or modulation). Previous technique:

[0006] We know, for example, the document FR3032848 "method for generating groups within an ad hoc radio network" which deals exclusively with clustering, or the document FR3004309 "distributed method for selecting a configuration in mobile networks" which deals with configuration.

[0007] These solutions are independent and do not address the joint problem. Simply combining the two solutions is not enough to make it work. FR3032848 "PROCESS FOR GENERATING GROUPS WITHIN AN AD HOC RADIO NETWORK" deals with the dynamic and distributed grouping of nodes in a mobile ad hoc network. However, it does not address the configuration of this set of nodes. The method in FR3004309 determines the configuration of a set of nodes already formed into a fixed cluster, and therefore does not allow for the management of dynamic node groupings.

[0008] It is therefore imperative to jointly understand the dynamic formation of clusters and the dynamic configuration of these clusters, taking into account the interactions between the dynamic formation and configuration of clusters.

[0009] The purpose of this invention is to provide a solution for jointly grouping nodes in a distributed and dynamic manner and for determining the configuration of the sets of nodes thus formed, while pooling the signaling necessary for both aspects. Summary of the invention:

[0010] To this end, according to a first aspect, the present invention describes a method for the joint dynamic implementation of node clustering and cluster configuration selection within a wireless ad hoc communication network of mobile nodes, a cluster comprising nodes, including a node designated as the cluster leader, sharing at least one common logical channel assigned to the cluster, said method being characterized in that: each node of a cluster is constrained to iterate the operating cycle comprising the following steps: i / triggering a neighborhood discovery step determining updated information including the identification of neighbor node(s) and the calculation of radio link quality indices between said neighbor nodes and said node; ii / collection by the cluster leader node of said updated information from the nodes of said cluster and updating of configuration metrics based on said information; iii / selection of a cluster configuration by the cluster leader node based on the updated configuration metrics, the selection of the configuration including at least the selection of the logical channel associated with the cluster and, if the selected configuration is different from the current configuration of the cluster, triggering, by the cluster leader node with each node of the cluster, the implementation of the new selected configuration;then iv / verification that a set of conditions relating to the configuration implemented in the cluster is indeed verified, said set of conditions determining whether the stability of the configuration implemented in the cluster is greater than a fixed threshold; and if and only if the set of conditions is verified, triggering of an evaluation of the cluster by the cluster leader node based on the information updated in step i / and in the case where said evaluation step determines that an action of merging the cluster with another cluster or of splitting the cluster or of leaving one or more nodes of the cluster to another cluster is to be carried out, implementation of said action following the triggering of said action by the nodes of the cluster;If the set of conditions is not met, step v is implemented directly: v / iteration of steps i to iv, whether a new configuration has been implemented or not, whether a merge, split or start action has been implemented or not.

[0011] The invention makes it possible to regulate and order the configuration and clustering functionalities, while allowing the pooling of the exploitation of topology information collected at the beginning of the cycle.

[0012] In some embodiments, such a process will further include at least one of the following features: said set of conditions includes one or more configuration-related conditions, among: ▪ no new configuration trigger was triggered in step iii of the current cycle; ▪ a time greater than a predefined minimum threshold has elapsed since the last configuration change; ▪ the current configuration has been implemented for the last θ stab cycles, where θ stab is a predefined integer value; in the cluster configuration step, the cluster leader node performs an operation to designate a new cluster leader from among the cluster nodes based on said updated information and the cluster leader node transmits to the cluster nodes, in the same message, the new selected logical channel and the indication of the new designated cluster leader;In the cluster evaluation stage, it is first determined by the cluster leader node, based on said updated information, whether the cluster meets a set of conditions defining the required characteristics of a cluster, and if it is determined that the cluster does not meet said conditions, a cluster split is evaluated by the cluster leader node; if it is determined that the cluster meets said conditions, a cluster merge operation or the departure of a subset of nodes from the cluster to another existing cluster is evaluated by the cluster leader node based on said updated information, and then triggered if appropriate;At the end of the operating cycle, the formula for a metric based on updated information used in step iii to select a configuration, and whose calculation at a given cycle is a function of a value calculated at the previous cycle, is modified by replacing said value with another value.

[0013] According to a second aspect, the invention describes a computer program intended to be installed in the memory of a mobile node of a wireless ad hoc communication network and to be executed on a microprocessor of the node, said computer program comprising software instructions which, when executed on the microprocessor, implement the steps, which are the responsibility of said node, of a process according to the first aspect of the invention.

[0014] According to a third aspect, the invention describes a mobile node of an ad hoc wireless communication network of mobile nodes in which one or more clusters comprising nodes, including a node designated as the cluster leader, share at least one common logical channel assigned to the cluster, said node being characterized in that: i / the node is adapted, if it is part of a cluster, to trigger a neighborhood discovery step determining updated information including the identification of neighbor node(s) and the calculation of radio link quality indices between said neighbor nodes and said node and to send this updated information to the cluster leader node; ii / if it is the cluster leader, the node is adapted to collect the updated information from the nodes of said cluster and update configuration metrics based on said information;iii / if it is the cluster leader: the node is adapted to select a cluster configuration based on updated configuration metrics, the configuration selection including at least the selection of the logical channel associated with the cluster and, if the selected configuration is different from the current cluster configuration, and the node is adapted to trigger, at each node of the cluster, the implementation of the new selected configuration; then iv / if it is the cluster leader: the node is then adapted to check if a set of conditions relating to the configuration implemented in the cluster is indeed met, said set of conditions determining if the stability of the configuration implemented in the cluster is greater than a fixed threshold;and if and only if the set of conditions is met, the node is suitable to trigger a cluster evaluation based on the information updated at operation i / and in the event that it determines, during said evaluation, that an action of merging the cluster with another cluster or splitting the cluster or leaving one or more nodes of the cluster to another cluster is to be carried out, to trigger said action with the nodes of the cluster; ; the node, if it is part of a cluster, is suitable to iterate operations i to iv, whether a new configuration has been implemented or not, whether a merge, split, or start action has been implemented or not.

[0015] In some embodiments, such a node will further include at least one of the following characteristics: said set of configuration conditions includes one or more of the following conditions: no new configuration trigger was triggered in step iii of the current cycle; a time greater than a predefined minimum threshold has elapsed since the last configuration change; the current configuration has been implemented for the last θ stab cycles, where θ stab is a predefined integer value; the node is adapted to, during cluster configuration selection, perform an operation to designate a new cluster leader from among the cluster nodes based on said updated information and to transmit to the cluster nodes, within the same message, the new selected logical channel and the indication of the new designated cluster leader.

[0016] According to a fourth aspect, the invention describes an ad hoc wireless communication network of mobile nodes, comprising a plurality of mobile nodes according to the third aspect of the invention. Brief description of the figures :

[0017] The invention will be better understood and other features, details and advantages will become clearer from the following description, given by way of non-limiting reason, and from the accompanying figures, given by way of example. There figure 1 represents a wireless communication network of ad hoc nodes in one embodiment of the invention; The figure 2 represents the steps of a configuration and clustering process in the network of the figure 1 in an embodiment of the invention;

[0018] Identical references may be used in different figures when they refer to identical or comparable elements. Detailed description:

[0019] Let us consider, with reference to the figure 1 A wireless ad hoc communication network 10 comprises N nodes 20, where N=8. Such an ad hoc network has no fixed infrastructure. The nodes 20, or at least some of them, can be mobile and move autonomously relative to each other. The network 10 comprises one or more clusters whose number and composition evolve dynamically through the implementation in the nodes of a prior art clustering algorithm, and the configuration of each cluster is selected from several possible configurations through the implementation in the nodes of a prior art configuration selection algorithm, the whole being implemented in the manner described below.

[0020] Each node 20, for example a cell phone, a laptop, comprises, as shown in more detail on the figure 1For one of the nodes, there is a radio frequency transceiver 21, a microcomputer 22, and a memory 23. The memory 23 stores the joint clustering and configuration selection algorithms. Each node in a cluster is adapted to exchange data with the other nodes in the cluster over a shared logical channel assigned by configuration, to implement predefined joint clustering and configuration functions.

[0021] Each node is adapted periodically, in addition to the determination of its internal state: analyze the spectral band and calculate channel quality metrics for all channels present; determine its neighboring nodes (neighborhood discovery) and information about each neighboring node such as a link quality index between them, their neighborhood, their cluster.

[0022] In one embodiment, the 20 nodes of the network are adapted to cyclically implement a process 100 schematically represented in figure 2 In one embodiment, each node 20 further includes in its memory 23 software instructions which, when executed on the microcomputer 22 of node 20, implement the steps, which are its responsibility, of the process 100.

[0023] With reference to the figure 2 The process according to the invention comprises three areas of action: a common area for configuration and clustering aspects, an area specific to configuration, and an area specific to clustering.

[0024] Let us describe an iteration θ of the operating cycle proposed according to process 100.

[0025] The network, at the start of this iteration, is in a given state, in terms of clusters formed and cluster configurations.

[0026] According to the invention, the 20 nodes of the network are forced to repeat this cycle iteratively. Consider any node 20, part of a cluster; a node of said cluster has previously been elected as the cluster leader. The cluster leader manages, for example, the cluster addressing and the distribution of control traffic within the cluster. Note that upon node initialization, it is common practice to require the node to search for a cluster to join, even if it means becoming the leader of its own cluster.

[0027] The cyclic process according to the invention comprises three successive phases ordered as follows: i ) a phase of information dissemination, ii ) a configuration selection phase based on the information obtained during the information dissemination phase, and finally iii ) a phase of grouping nodes into clusters based on information obtained during the information dissemination phase.

[0028] In the first phase (that of information dissemination), an information request is triggered, by querying one or more nodes or links and asking them to evaluate a communication quality index for a node or a link between two nodes of a cluster, and / or to evaluate a communication quality index for links between nodes of different clusters.

[0029] In one embodiment, this first phase is implemented, for example, using steps 101 and 102 and is triggered by the cluster leader. In step 101, a node 20, upon receiving the request, calculates the current local topology of the cluster to which node 20 belongs at the beginning of the considered cycle θ, by exchanging information with its geographical neighbors in the same cluster, in order to identify which nodes are present in the cluster. It then determines data enabling the evaluation of the quality of the identified communication links within the cluster. Generally, the link quality evaluation is performed using one or more of the following: bandwidth, load, delay, n-hop neighborhood, loss rate, received signal strength, SNR, SINR, and / or BER... or a quality index obtained based on one or more of these elements.Depending on the embodiment, one or more of the following elements will be further determined: the links between these nodes, the number of hops per link (this data will be used to establish the routing tables), the weights associated with the identified nodes, etc. In step 102, node 20 performs inter-cluster neighborhood discovery tasks, enabling the identification of the list of nodes (and / or cluster leader nodes) of neighboring clusters, along with their cluster identifier and weight, as well as indices of the quality of the associated communication links as defined above. These indices are derived in particular from measurements or exchanges carried out by the nodes. Depending on the embodiment, this step is performed only by the cluster leader nodes (nodes 20) or by any node 20 or a subset of nodes 20 belonging to the cluster. These steps 101 and 102 are part of the so-called common sector mentioned above.They correspond to the dissemination, between nodes and up to the cluster leader node, of information about the topology, including the evaluation of the quality of the cluster's communication links. They are implemented, for example, through known neighborhood discovery protocols, involving the exchange of "hello" type messages indicating the identifiers of nodes, clusters, node weights, and the number of nodes in the cluster as known to the node sending the "hello" message.

[0030] As an example, it is assumed here that a weight is assigned to each node, dynamically updated according to one of these performance criteria specific to that node or a combination of criteria: proximity to the geographic center of the cluster, stability, degree of mobility assessed on the historical power of received levels or geographic coordinates, for example, remaining energy, capacity, transmission power. These criteria correspond to metrics to be calculated.

[0031] Next, in the cluster configuration phase, the cluster leader node in each cluster triggers a configuration selection operation for the cluster based on the information received in the first phase, specifically based on a comparison of the received quality indices and given constraints for the cluster (e.g., minimum quality of service thresholds, minimum throughput, maximum latency). This operation is implemented by executing the configuration algorithm stored in its memory.

[0032] For example, in one embodiment, the configuration phase includes steps 103 and 104. In step 103, the cluster leader node (20) calculates updated metrics based on the updated information received in the first phase. These metrics might include a metric indicating whether each node in the cluster is connected to at least one other node in the cluster, a metric associated with the election of the new cluster leader, a metric called "utility" indicating the ratio of the number of active links to the maximum number of links (i.e., if all nodes in the cluster were directly connected to all other nodes in the cluster), and a cluster throughput metric. In step 104, the cluster leader node (20) selects the cluster configuration based on the metrics calculated in step 103. If the configuration currently implemented in the cluster is the one selected, then step 106 is implemented directly.If the newly selected configuration differs from the configuration currently implemented in the cluster, it is transmitted by the cluster leader to the nodes of that cluster (for example, it includes the list of logical channels or the choice of the cluster leader), and step 105 is then implemented. This configuration, when distinct from the configuration previously implemented in the cluster, specifies, for example, one or more logical channels for the cluster that are distinct from those previously configured. It also includes, for example, the choice of a new cluster leader, which is deemed "better" than the current one (based on cluster leader selection criteria, for example, corresponding to the highest weight). In some embodiments, other configuration parameters are selected from among: the transmission power of the cluster nodes, the coding method, the modulation, etc.In one embodiment, the new cluster configuration and the designation of the new cluster leader are transmitted in a single message to the other nodes of the cluster by the current cluster leader. These steps, 103 and 104, are part of the configuration area mentioned above.

[0033] In a step 105, falling under the common sector, the newly selected configuration for the cluster is then implemented within the cluster, by each node 20 in the cluster.

[0034] In the node grouping phase, the grouping of nodes into clusters is updated, which can lead to the exit or entry of nodes into an existing cluster, the dissolution of existing clusters or the creation of new clusters depending on the verification of a set of a priori operating constraints by the topology data and quality indices obtained during the diffusion phase.

[0035] For example, in one embodiment, the clustering phase includes steps 106 to 109.

[0036] In step 106, the cluster leader checks if a set of conditions is met.

[0037] These conditions are, for example, one of (or a combination of some of) the following: no configuration change was triggered at step 104 of cycle θ; a time greater than a predefined minimum threshold has elapsed since the last configuration change; the current configuration has been implemented since the last θ stab cycles, where θ stab is a predefined integer value. Step 106 aims to allow the network to stabilize before triggering a clustering operation. This prevents a clustering change based on the new configuration being triggered until it has been verified that it is better or until the new configuration is stably installed.

[0038] If the condition in step 106 is met, then in step 107, the cluster leader checks if a set of constraints is satisfied; if these constraints are not met, the cluster is divided.

[0039] This set of constraints includes, for example, one or more of the following conditions, and the verification takes into account the information obtained in steps 101, 102, or metrics calculated based on this information: Constraints on the current cluster topology: the number of nodes updated following steps 101, 102 is less than a predefined maximum number of nodes associated with the cluster; the diameter of the cluster; the distance to the cluster leader; the connectivity of the cluster; a required level of quality of service.

[0040] If a constraint from step 107 is not met, then a cluster split operation is evaluated and then triggered in step 109 by the cluster leader node (20), in a known manner, to the affected nodes, if the cluster split conditions are met. The triggering is performed, depending on the case, by the cluster leader node or by another node, for example, if a failure of the cluster leader or its absence from the cluster is detected. Then step 110 is implemented.

[0041] If the constraints of step 107 are met, then the functions relating to the cluster evaluation itself are triggered in step 108 by the cluster leader node 20, leading for example to a merge with a neighboring cluster.

[0042] In this case, the cluster evaluation in step 108 is implemented using a clustering algorithm, which determines a dynamic and distributed grouping of nodes based on one or more of the following parameters: node location, residual energy, mobility, current communication quality, topology, frequency spectrum, desired quality of service (in terms of throughput, latency, jitter, loss rate, and security), and the maximum and / or minimum cluster size (in number of nodes and / or distance between nodes). Clustering includes identifying the nodes to be grouped together and identifying specific nodes, such as gateway nodes, that provide communication links to neighboring clusters.

[0043] Step 108 results, for example, in the cluster leader initiating a merger of two neighboring clusters with the relevant nodes, or in a subset of nodes leaving the cluster for another existing cluster. Then step 110 is implemented.

[0044] If the condition of step 106 is not met, then any triggering of the clustering algorithm on the cluster in question is prevented and step 101 of the following cycle (cycle θ+1) will be the next step implemented.

[0045] In step 110, implemented following steps 108 and 109, some of the metrics are updated by the 20 nodes, so that they are not skewed by the change in clustering.

[0046] For example, consider a metric that measures the evolution of cluster utility (ratio of the number of links to the maximum number of links). A cluster change accompanied, relative to the previous cluster, by an increase in the maximum number of links or a decrease in the number of links associated with a cluster change (for example, merging a cluster of 10 interconnected nodes – utility equal to 1 – with a single-node cluster that is only connected to the previous cluster by one node results in a new cluster with a utility (equal to 91 / 110) less than 1) would lead to a decrease in the metric measuring the evolution, which could trigger a configuration change; however, the decrease in the metric at such a time would not reflect a decrease in the quality of the configuration, but simply a change in clustering.Thus, in one embodiment, in step 110, the formula for one (or more) metric used in step 103 to decide whether or not to trigger a configuration change, and whose calculation at a given cycle depends on a value calculated in the previous cycle, is modified by replacing that value with another value. For example, the utility value of the cluster considered before the cluster change in cycle θ is replaced, at the end of cycle θ, by the utility value of the new cluster. This prevents the configuration from being considered bad in the next cycle due to a negative utility change metric, which would lead to another configuration change, which, if considered bad, would result in a cluster split, and so on, leading to an unstable situation.

[0047] Then the cycle is iterated, whether a new configuration has been implemented or not, whether an action of merging or splitting or leaving node(s) from the cluster to another existing cluster has been implemented or not.

[0048] The invention thus defines an operating cycle which also governs the impact of the grouping of nodes on the configuration of this set and vice versa.

[0049] In one embodiment, the cycle is implemented in parallel with the useful traffic (for example, out of 10 consecutive packets, n (n is a value that can be adapted according to the results sought, for example n=2) will be allocated to the implementation of the cycle, 8 to the useful data).

[0050] The pooling of signaling which allows for example to pool the information collected in steps 101, 102 and the role of cluster leader, present for the configuration, and to use it for the grouping of nodes.

[0051] The invention establishes, in one embodiment, new quality indices for the configuration, taking into account the dynamics of the grouping of nodes within a cluster (steps 103 and 110).

[0052] The proposed approach defines a cycle that allows the two functionalities to operate sequentially, with one implementation that prioritizes configuration and can block node grouping operations. It also introduces new metrics and utility functions that take into account the dynamics of node grouping in relation to configuration. The solution shares signaling, which notably allows the cluster leader role to be shared during configuration and used for node grouping.

[0053] The solution described with reference to the figures is: distributed: it does not need a central entity managing all the information; dynamic: it allows adaptation to changes, such as topology; complete: it deals with all aspects of the problem, including its implementation and the necessary signaling.

Claims

1. Method for dynamically implementing joint node clustering and cluster configuration selection within an ad hoc wireless communication network (10) of mobile nodes (20), a cluster comprising nodes, including a node designated as cluster head, which share at least one common logical channel assigned to the cluster, said method being characterised in that: each node (20) of a cluster is required to iterate the operating cycle comprising the following steps: i / triggering a neighbourhood discovery step which determines updated information comprising the identification of neighbouring node(s) and the computation of indices of the quality of the radio links between said neighbouring nodes and said node (20); ii / collection by the cluster head node of said information updated by the nodes of said cluster and updating of configuration metrics according to said information; iii / selection of a cluster configuration by the cluster head node according to the updated configuration metrics, the selection of the configuration comprising at least the selection of the logical channel associated with the cluster and, if the selected configuration is different from the current configuration of the cluster, triggering, by the cluster head node in each node of the cluster, the implementation of the new selected configuration; then iv / verification that a set of conditions relating to the configuration implemented in the cluster is correctly met, said set of conditions determining whether the stability of the configuration implemented in the cluster is greater than a fixed threshold; and if, and only if, the set of conditions is met, triggering of an evaluation of the cluster by the cluster head node according to the information updated in step i / and in the case where said evaluation step determines that an action of merging the cluster with another cluster or of dividing the cluster or of moving one or more nodes of the cluster to another cluster is to be carried out, implementation of said action following the triggering of said action in the nodes of the cluster; if the set of conditions is not met, step v is implemented directly; v / iteration of steps i to iv, regardless of whether a new configuration has been implemented or not, and regardless of whether a merging or dividing or moving action has been implemented or not.

2. Method for dynamically implementing joint node clustering and cluster configuration selection according to claim 1, in which said set of conditions relating to the configuration comprises one or more of: - no new configuration trigger has been triggered in step iii of the current cycle; - a time greater than a predefined minimum threshold has elapsed since the last configuration change; - the current configuration has been implemented since the last θstab cycles, where θstab is a predefined integer value.

3. Method for dynamically implementing joint node clustering and cluster configuration selection according to any one of the preceding claims, in which, in the cluster configuration step, the cluster head node performs an operation of designating a new cluster head from among the nodes of the cluster according to said updated information and the cluster head node transmits to the nodes of the cluster, within the same message, the new logical channel selected and the indication of the new designated cluster head.

4. Method for dynamically implementing joint node clustering and cluster configuration selection according to any one of the preceding claims, in which, in the step of evaluating the cluster: - it is determined in advance by the cluster head node, according to said updated information, whether the cluster meets a set of conditions defining the required characteristics of a cluster, and if it is determined that the cluster does not meet said conditions, a cluster division is evaluated by the cluster head node, and then triggered if appropriate; - if it is determined that the cluster meets said conditions, an operation of merging the cluster or moving a subset of nodes from the cluster to another existing cluster is evaluated by the cluster head node according to said updated information, and then triggered if appropriate.

5. Method for dynamically implementing joint node clustering and cluster configuration selection according to any one of the preceding claims, in which, at the end of the operating cycle, the formula of a metric as a function of the updated information which is used in step iii to select a configuration and whose calculation in a given cycle is a function of a value calculated in the preceding cycle is modified by replacing said value with another value.

6. Computer program product intended to be installed in the memory of a mobile node (20) of a wireless ad hoc communication network (10) and to be executed on a microprocessor of the node, said computer program comprising software instructions which, when executed on the microprocessor, implement the steps, incumbent on said node, of a method according to any one of claims 1 to 5.

7. Mobile node (20) of a wireless ad hoc communication network (10) of mobile nodes in which one or more clusters comprising nodes, including a node designated as cluster head, share at least one common logical channel assigned to the cluster, said node being characterised in that: i / the node (20) is suitable, if it is part of a cluster, for triggering a neighbourhood discovery step which determines updated information comprising the identification of neighbouring node(s) and the computation of indices of the quality of the radio links between said neighbouring nodes and said node, and for sending this updated information to the cluster head node; ii / if it is a cluster head, the node (20) is suitable for collecting the information updated by the nodes of said cluster and updating configuration metrics according to said information; iii / if it is a cluster head: the node (20) is suitable for selecting a cluster configuration according to the updated configuration metrics, the selection of the configuration comprising at least the selection of the logical channel associated with the cluster and, if the selected configuration is different from the current configuration of the cluster, and the node is suitable for triggering, in each node of the cluster, the implementation of the new selected configuration; then iv / if it is a cluster head: the node (20) is then suitable for verifying whether a set of conditions relating to the configuration implemented in the cluster is correctly met, said set of conditions determining whether the stability of the configuration implemented in the cluster is greater than a fixed threshold; and if, and only if, the set of conditions is met, the node is suitable for triggering an evaluation of the cluster according to the information updated in operation i / and, if it determines during said evaluation that an action of merging the cluster with another cluster or of dividing the cluster or of moving one or more nodes of the cluster to another cluster is to be carried out, for triggering said action in the nodes of the cluster; the node (20), if it is part of a cluster, is suitable for iterating operations i to iv, regardless of whether a new configuration has been implemented or not, and regardless of whether a merging or dividing or moving action has been implemented or not.

8. Node (20) according to claim 7, wherein said set of conditions relating to the configuration comprises one or more conditions relating to the configuration from: - no new configuration trigger has been triggered in step iii of the current cycle; - a time greater than a predefined minimum threshold has elapsed since the last configuration change; - the current configuration has been implemented since the last 8stab cycles, where θstab is a predefined integer value.

9. Node (20) according to any one of claims 7 to 8, suitable for, at the time of the cluster configuration selection, performing an operation of designating a new cluster head from among the nodes of the cluster according to said updated information and for transmitting to the nodes of the cluster, within the same message, the new logical channel selected and the indication of the new designated cluster head.

10. Ad hoc wireless communication network (10) of mobile nodes, comprising a plurality of mobile nodes (20) according to any one of claims 7 to 9.