GB2527298A - Method of assigning communication channel to communication means in a mesh network - Google Patents

Abstract

A method of assigning communication channels in a mesh network 10 comprising multiple nodes 100, 110, 120, 130, 140 comprises the steps of identifying, for an uninitialized node 140, at least one neighbour node; determining possible channel assignments for the uninitialized node 140 that permit a connection 201, 202 to at least one neighbour node; calculating, for each possible channel assignment, an optimisation value; and implementing a possible channel assignment having an optimisation value closest to a predetermined target optimisation value. The optimisation value preferably quantifies a topological aspect of the network such as the number of nodes connected to only one node, the number of links between the uninitialized and master nodes and the number of wired links between the uninitialized and master nodes. If a subset of channel assignments each have an optimisation value closest to the target, a sub-optimisation value, preferably relating to a different topological aspect than the optimisation value, may be calculated for each channel in the subset. The channel assignment with a sub-optimisation value closest to a target will then be implemented. Taking this approach allows one to assign channels taking into account both the latency and the robustness of the network.

Description

Intellectual Property Office Application No. GB1410679.3 RTIVI Date:28 October 2014 The following terms are registered trade marks and should be read as such wherever they occur in this document: Wi-Fi Bluetooth Intellectual Property Office is an operating name of the Patent Office www.ipo.govuk

I

Method of assigning communication channel to communication means in a mesh network

Field of the Invention

The invention belongs to the field of wireless mesh networks for which nodes operate over multiple communication channels.

More specifically, the invention relates to a method for assigning a communication channel or channels to each node in a wireless mesh network, so as to improve robustness of the network to disruptions, e.g. shadowing conditions, and other aspects of network performance.

Description of the Related Art

In a mesh network, each node can assume the role of a relay node, repeating received data in order to either extend network coverage or improving robustness by providing multiple copies of the data to destination nodes. This is especially true for networks that contain wireless links, where the ambient environment (e.g. fixed or moving obstacles) increases the likelihood of transmission errors or even complete shadowing.

By "wireless network," it is understood to mean any means of communication in computer networks where information packets are transmitted over a radio-wave link, for instance radio-Ethernet, Wi-Fi, Bluetooth, etc. In such networks, it may be desirable or even necessary to provide several radio communication modules in each node. Indeed, such a configuration potentially extends available bandwidth through radio channel aggregation, and/or increase spatial coverage when the radio technology does not easily allow omnidirectional data transmission. This is the case, for instance, with 60 GHz millimetre-wave technology, where a typical antenna system generates, at best, a hemispherical radiation pattern. Here, multiple radio modules provide a multitude of antennas, compounding the radiation pattern of each antenna and realizing improved spatial coverage.

In particular, 60 GHZ millimetre-wave technology is subject to high attenuation from physical obstacles such as the human body. As a result, the quality of communications may be degraded or subjected to interference, in particular masking conditions.

To mitigate the effect of fading and interference, and to ensure successful data transmission, it is desirable for the destination node to receive a particular data packet multiple times, potentially using several relay nodes to achieve transmission redundancy. This may be achieved by employing a mesh network topology, wherein at least some of the nodes in the network are connected to two or more other nodes.

Typically, the source node transmits using an antenna setting generating a wide beam in order to reach a set of neighbour nodes. When receiving, the node uses an antenna setting which scans for incoming signals in a narrow cone. This avoids some interference due to multiple receptions from the multiple paths that may be generated by the passive signal reflections in the environment.

When a node is equipped with several radio modules, simultaneous transmission by multiple radio modules over the same radio channel must be avoided. Otherwise, interference generated by the simultaneous transmission is likely to prevent correct reception.

Thus, if several radio modules of a single node or group of nodes are configured to transmit and receive on the same radio channel, then transmission must be performed sequentially. That is, the first radio module transmits during a first time slot, followed by the second radio module transmitting during a second, subsequent time slot, and so on.

To optimize bandwidth usage, it is more advantageous to assign, as much as is possible, a different radio channel to each radio module, and transmit simultaneously on radio modules operating on non-overlapping radio channels.

For instance, four radio channels are defined in the 60 GHz frequency band dedicated to millimetre-wave radio technology (e.g. the 802.llad and 802.1 5.3c standards).

When a radio channel is assigned to one radio module, it is preferable to avoid changing this assignment during operation, as the switching of channels is time consuming and thus detrimental to network performance. For example, in the 802.15.3c standard, the time to switch a radio module from one radio channel to a new radio channel can be on the order of 1 OOps.

For each node in the mesh network, there is at least one other node that is a "neighbour node." Two nodes are neighbour nodes and thus "visible" to each other if radio communications are possible between them. This means that at least one radio communication module of the first node and at least one radio communication module of the second node transmit and receive over the same radio frequency channel, and that the radiation patterns of the associated antennas permit communications between them. Thus, the number of neighbour nodes for any given node in a network directly depends on the allocation of radio channels to each radio communication module of each node.

As a consequence, the connectivity and functionality of the network can vary widely according to this radio channel assignment. The problem that arises here is the difficulty in assigning communication channels so as to provide a network connectivity that is robust to transmission errors. By "robust," it is meant that when a wireless link between two nodes is disturbed by interference or otherwise rendered non-functional, there remains at least one other path by which each of the two nodes in question may still communicate with each other.

Additionally, it is desirable to maximize the performance of the network by minimizing network latency and improving synchronization accuracy for any information traveling to and from the master node from the slave nodes. Each intermediate node in a transmission path will introduce some amount of network jitter and timing error when reproducing the timing information from the master node. This error is amplified by each intermediate node, thus the greater the number of intermediate nodes involved in the transmission of information, the greater the introduced error.

As a consequence, while a small inter-frame gap is desirable to optimize bandwidth usage, the inter-frame gap must ultimately be adapted to the synchronization accuracy for the transmission path in use. A longer transmission path reduces transmission accuracy and obligates the use of a large inter-frame gap, while conversely a shorter transmission path increases transmission accuracy and permits the use of small inter4rame gaps.

A solution to this problem has been proposed in US patent application publication US 2007/0002794. This document presents a method of allocating radio channels in a wireless mesh network having multiple nodes. The allocation is performed in a sequential manner, node after node, and according to a connectivity constraint which consists in having, for a given node, at least one common channel with all of its neighbour nodes. The channel allocation is then refined according to the computation of throughput available for each communication link.

There are several limitations for this solution. First, the requirement of one common channel for a given node and its neighbour node might not be satisfied, especially in the case of a system with a limited number of radio modules per node and a limited number of available radio channels overall. Moreover, the selected criterion for radio channel allocation is based on throughput calculation, so it does not take into account the robustness of the network to shadowing or other interference conditions. Lastly, this method does not take into account the connection conditions between each node and a master node.

Instead, in a typical wireless mesh network one node is designated as a master node and thereby placed in charge of managing the network. For instance, where the medium access is organized with the TDMA (Time Division Multiple Access) channel-access method, the master node periodically transmits a beacon frame determining the beginning of a data transmission sequence where each transmission time slot is allocated to one unique node of the network. Based on the reception of this beacon frame, non-master nodes (AKA "slave" nodes) are informed about and synchronised with the TDMA sequence. The slave nodes can also relay the beacon frame, enabling other slave nodes to be synchronized even if they are not directly connected to the master node.

As a consequence, for such a system it is advantageous to limit the number of intermediate nodes between the master node and any particular slave node. So far, there is no solution for providing channel allocation in a multiple-channel wireless network which takes into account the connectivity of the slave nodes with the master node.

SUMMARY OF THE INVENTION

In a first aspect, therefore, the invention is directed towards a method of assigning wireless communication channels in a mesh network comprising multiple nodes and at least one wireless link between two of said nodes.

According to the invention, the method comprises the steps of identifying, for an uninitialized node not connected to the mesh network at least one neighbour node to which at least one wireless communication channel has been assigned; determining at least one possible channel assignment for the uninitialized node, said at least one possible channel assignment permitting the establishment of at least one direct wireless link between said uninitialized node and said at least one neighbour node; calculating, for each possible channel assignment, an optimisation value; and implementing the possible channel assignment having an optimisation value closest to a predetermined target optimisation value.

This is advantageous in that when the uninitialized node is introduced into the wireless mesh network, the communication channel over which it will communicate with its neighbour nodes is selected so as to optimize the connectivity of the newly-introduced uninitialized node. By calculating an optimisation value for each possible channel assignment, the possible channel assignments may be assessed and the one best fulfilling the needs of the user may be rapidly implemented.

In a preferred embodiment, the optimisation value quantifies a topological aspect of the mesh network.

The channel assignment is thereby based on the operational characteristics of the network, improving the degree to which the connectivity of the network is so optimized.

Most preferably, the optimisation value quantifies a topological aspect of the mesh network corresponding either to the number of nodes in the mesh network connected to only one other node, the number of links between the uninitialized node and a master node of said mesh network, or the number of wired links between the uninitialized node and said master node.

In this way, the optimisation that is performed upon introduction of the uninitialized node may be tailored to the requirements of the particular network in question. The resulting mesh network will thus either be optimized for maximum robustness, and therefore maximum resilience in the face of interference and transmission errors; or for a minimum number of transmission links between the introduced uninitialized node and the master node, and thus minimized latency.

In an embodiment wherein there exists a subset of possible channel assignments each having an optimisation value closest to the target optimisation value, a sub-optimisation value is calculated for each possible channel assignment in said subset, the channel assignment having a sub-optimisation value closest to a pre-determined target sub-optimisation value being thereby implemented.

This is advantageous in that the assessment of the possible channel assignments is performed over an additional iteration, thereby further refining the determination of the channel assignment having the best possible connectivity.

Preferably, the sub-optimisation value quantifies a topological aspect of the mesh network different from that quantified by the optimisation value.

The utilization of a sub-optimisation value further refines the selection of the channel assignment. More particularly, this enables the selection of a channel assignment when several possible channel assignments are substantially equal as concerns the optimisation value.

Most preferably, the sub-optimisation value corresponds either to the number of nodes in the mesh network connected to only one other node, the number of links between the uninitialized node and a master node of said mesh network, or the number of wired links between the uninitialized node and said master node.

This is advantageous in that, similar to what is described above, the sub-optimisation value is tailored to the conditions of use of the mesh network and to the desired performance and connectivity characteristics. In particular, this According to a second aspect, there is provided a method of assigning wireless communication channels in a mesh network comprising multiple nodes and at least one wireless link between two of said nodes.

According to the invention, the method comprises the steps of identifying, for an uninitialized node not connected to the mesh network, at least one neighbour node to which the at least one wireless communication channel has been assigned; determining a plurality of possible channel assignments for the uninitialized node, each of said possible channel assignments permitting the establishment of at least one direct wireless link between said uninitialized node and said at least one neighbour node; for each possible channel assignment, calculating an optimisation value corresponding to the number of nodes in the mesh network connected to only one other node; determining a subset of possible channel assignments each having an optimisation value closest to zero; for each possible channel assignment in said subset, calculating a sub-optimisation value corresponding to the number of links between the uninitialized node and a master node of said mesh network; and implementing a possible channel assignment having an optimisation value closest to a predetermined target optimisation value.

In this way, the uninitialized node is incorporated into the wireless network quickly, in a fashion that results in the most robust and efficient mesh network possible under the circumstances.

According to a third aspect, the invention is drawn to a mesh network configured according to the method described above.

Such a mesh network embodies the advantages of the method described above, in particular the rapidity with which it is configured and its adaptiveness and robustness in the face of changing ambient conditions.

According to a fourth aspect, there is provided device for assigning communication channels in a mesh network comprising multiple nodes and at least one wireless link between two of said nodes.

According to the invention, the device comprises a means for identifying, for an uninitialized node not connected to the mesh network at least one neighbour node to which at least one wireless communication channel has been assigned; a means for calculating, for each possible channel assignment, an optimisation value; and a means for implementing a possible channel assignment having an optimisation value closest to a predetermined target value.

A device so configured will yield a mesh network embodying the advantages discussed above.

Preferably, said device is a wireless node for a mesh network.

This is advantageous in that the advantages of the device are applied to ameliorate the peculiar difficulties in operating a wireless mesh network.

According to a fifth aspect, the invention is drawn to a computer program comprising instructions for implementing a method as described above.

By so implementing the method, the method is implemented with a maximum of efficiency and automation, minimizing the effect of its execution upon the overall performance of the mesh network.

The present invention thus provides a method for assigning communication channels to communication modules attached to or within the nodes of a mesh network. This assignment is performed individually by each node joining the network, taking into account two network connectivity criteria. First, the method takes into consideration the connectivity of the new, uninitialized nodes with the already-initialized nodes: it shall be able to communicate with at least one other neighbour node. Second, the method takes into consideration at least one of two other criteria: giving priority to minimizing the distance between the uninitialized node and the master node; and/or giving priority to the robustness of the network by providing multiple connections to a maximum number of neighbour nodes.

A predefined or arbitrary communication channel assignment is not likely to result in an optimized configuration, instead resulting in a situation where some nodes are isolated with few or even no connection(s) to other nodes. The advantage of the method of this invention is that a new node does not need to be aware of the overall network topology to assign a channel to each of its communication modules. Based on the local network connectivity information received from its neighbour nodes, the new node is able to assign communication channels to its communication modules in such a way as to optimize the connectivity of the mesh network. It can perform this operation before sending any message requesting to join the network. The algorithm for communication channel allocation may be repeated later if the environment has changed or in case of a request from one of the other nodes.

BRIEF DESCRIPTION OF THE DRAWINGS

Other particularities and advantages of the invention will also emerge

from the following description.

In the accompanying drawings, given by way of non-limiting examples: -Figure 1 represents a network composed of three initialized nodes and one fourth node to be initialized and joined to the network; -Figure 2 represents the network with a first radio channel allocation for the fourth node; -Figure 3 represents the network with a second radio channel allocation for the fourth node; -Figure 4 represents another network topology with the introduction of a wired link; -Figure 5 represents the beacon frame format; -Figure 6 is a flow chart of the algorithm executed by a slave node joining the network; -Figure 7 is a flow chart of the algorithm executed by a slave node already inserted in the network; -Figure 8 is a flow chart of the algorithm executed by the master node; and -Figure 9 is a functional block diagram of a node according to an embodiment of the invention.

DETAILED DESCRIPTION OF THE EMBODIMENTS

Figure 1 represents a mesh network 10 composed of three initialized nodes and one new node whose addition to the network is sought. In this example it is considered that each node is equipped with two radio communication modules, and that each radio communication module covers a hemispherical space emanating therefrom with a limited range.

Here, the network 10 is controlled by one master node 100 equipped with two radio communications modules 101 and 102. The master node 100 as such regularly transmits a beacon frame.

In this embodiment, the master node 100 simultaneously transmits this beacon through both radio communications modules 101 and 102. These modules operate on two different and non-overlapping radio channels. The choice of radio channels for the master node 100 can be done arbitrarily among the available radio channels, La by arbitrarily selecting a channel not already in use by another system. In Figure 1, the radio communication module 101 operates on radio channel 1 and the radio communication module 102 operates on radio channel 2.

At this point, it is supposed that a first slave node 110 has joined the network. For that, slave node 110 performs an antenna discovery operation to detect and to receive the beacon frame from the master node 100. In this antenna discovery operation, the node activates one of its radio communication modules and selects a first radio channel. After scanning the space with the reception antenna, the node records if the beacon frame has been received or not. Then it repeats this operation for each available radio channel and for each radio communication module.

After this operation, the node can allocate a radio channel to its radio communication module. In this example, it is considered that slave node 110 is able to receive the beacon frame only with its radio communication module 111 operating on radio channel 1. Radio communication module 112 is configured to operate on radio channel 2.

In the event of beacon reception with both radio communication modules, the node can assign radio channels so as to give priority to the configuration providing the best reception quality (e.g. the highest received signal level). This action creates a wireless link 161 between the master node 100 and the slave node 110.

Without this antenna discovery, the slave node 110 would have arbitrarily assigned radio channels to its radio communication modules 111 and 112, running the risk of not being able to communicate with master node 100.

The slave node 110 can then join the network by sending a join request message during a contention time slot identified in the beacon frame. The master node 100, upon receiving the join request message, can then grant the access to the network with a specific message to that effect. Also, a time slot is allocated to the slave node 110 to repeat the beacon frame received from the master node 100.

As for the master node 100, the slave node 110 retransmits the beacon frame through its two radio communication modules simultaneously.

To continue the process, it is supposed that a second slave node 120 joined the network in the same manner, detecting only one wireless link 162 possible with slave node 110 and no wireless link available with master node 100.

The consequence for node 120 is to select radio channel 2 for its radio communication module 121 and radio channel 1 for its radio communication module 122.

To join the network, slave node 120 sends a join request message relayed by slave node 110 for the master node 100. An access-granting message from the master node 100 is also relayed by the slave node 110 for the slave node 120. Lastly, a third slave node 130 joins the network in the same manner as node 110, assigning radio channel 2 to its radio communication module 131 and radio channel 1 to its radio communication module 132, and creating the wireless link 163 with the master node 100.

At this point, a fourth slave node 140 is introduced to the network, starting with the antenna discovery phase. When the radio communication module 141 of slave node 140 scans on radio channel 2, it detects the master node 100 and the slave node 110; and when scanning on radio channel I it detects the slave nodes 120 and 130. It is supposed that no node can be detected with its communication module 142.

Two options are thus possible with regard to slave node 140. The first option is illustrated by Figure 2, wherein the slave node 140 assigns the radio channel 2 to its radio communication module 141, creating the wireless link 201 with the master node 100 and the wireless link 202 with the slave node 110. For slave node 140, this creates a direct connection to the master node 100 without any intermediate nodes, reducing network latency for communications between the master node 100 and the slave node 140.

The second option is illustrated by Figure 3, where the slave node 140 assigns the radio channel 1 to its radio communication module 141, creating the wireless links 301 and 302 between it and the slave nodes 120 and 130, respectively.

The advantage of this network topology is that the network is robust to a broken link; for example, should the wireless link 302 be broken, the slave node can still communicate with the master node 100 via the slave nodes 110 and 120. Indeed, if any one of the wireless links 161, 162, 163, 301, 302 are broken, each node remains connected to at least one neighbour node. The overall robustness of the network is improved by adopting the network topology of Figure 3, rather than that of Figure 2.

In another scenario, Figure 4 presents a situation where the slave node 140 can be connected only to slave node 130 through the wireless link 402.

Furthermore, the connection between the master node 100 and slave node 110 is established by the cable 401, rather than by a wireless link as in the previous Figures. There is thus no radio channel assignment for the communication modules 101 and 111.

In this scenario, a new slave node 160, equipped with two communication modules 151 and 152, is ready to join the network 40. The slave node 160 may therefore either connect to the slave node 150 by assigning radio channel 2 to its communication module 151, or connect to the slave node 120 by assigning radio channel 1 to its communication module 151.

Here, the preferred option is for the slave node 160 to assign radio channel Ito its communication module 151, and thereby establish a wireless link with slave node 120. This is because slave node 160 would be connected to the master node 100 through only two wireless links (the link between the master node 100 and slave node 110 being borne by the cable 401 and thus not subject to wireless shadowing, noise, interterence, etc.), rather than the three as would be the case if slave node 160 established a wireless connection to slave node 150.

The objective of the invention is to be able to evaluate the assignment of the radio channels within the network as it exists at the introduction of a new slave node, and to make a decision as to the assignment of radio channels to the new slave node which takes network connectivity criteria into account.

The application of the invention is not limited to the above examples.

The invention can be applied to nodes with a number of communication modules N (where N»=1) to which radio channels are assigned from a pool of K available radio channels (where K»=1), or equally to cases where a wireless connection with several different other nodes is possible through the same radio communication module.

For this purpose, the beacon frame transmitted by each node contains information arranged in the format described in Figure 5.

The beacon frame 50 starts with the Preamb'e 501 section, which is inserted by the communication module and enables the synchronization of the physical layer of the receiver.

The payload part of the beacon frame 50 is divided into five sections.

The section PayLoadl 502 contains information such as the identifier of the transmitter node, the TDMA sequence format and duration, the identification of the current time slot within the TDMA sequence, and the list of radio channels in use.

These data may be exploited by a receiver node to achieve network synchronization and to identify the time slot it should use for transmission.

The section neighbors 503 is used by the transmitter node to indicate the number of neighbour nodes visible to said transmitter node. For instance, if 4 bits are reserved in the beacon frame 50 for the neighbors 503 section, a maximum of 15 neighbour nodes may be reported.

The section distance 504 is used by the transmitter to indicate the number of intermediate links existing between the master node and itself The section wires 505 is used by the transmitter to indicate the number of wired links existing between the master node and itself.

As with neighbors 503, the maximum number of intermediate links and wired links that may be reported in the beacon frame 50 is a function of the length of the sections distance 504 and wires 505. Thus, if 4 bits are reserved for each of the distance 504 and wires 505 sections, the beacon frame 50 may report a maximum of 15 intermediate links, of which up to 15 are wired links.

The section Pay] cc d2 506 can contain other useful information for the management of the network. Finally, the section ECS 507 is an error-detecting code (e.g. a check digit) for enabling a receiving node to verify that the beacon frame 50 has been received without errors.

Typically, the sections 502 and 506 are created by the master node and repeated by the slave nodes, except that the portions of the Fayioadl 501 section containing the identifier of the transmitting node and the time slot used are updated by each node as it transmits the beacon frame 50. The sections 503, 504, and 505 represent the local network connectivity status for, and are thus always provided by, the node transmitting the beacon frame, regardless of its master/slave status. The values may vary over time as the network environment changes, e.g.as nodes are introduced or withdrawn from the network or with the presence of obstacles or interference conditions.

The usage of these particular sections 501, 502, 503, 504, 505, & 506 within the beacon frame 50 is now described within the context of the examples of Figures 1, 2, and 3. During the identifying step, the slave node 140 detects and receives the beacon frame transmitted by each of the visible neighbour nodes.

When the radio communication module 141 of slave node 140 scans on radio channel 2, it will receive a beacon frame from the master node 100 and from the slave node 110. Upon reception of this beacon frame, the slave node 140 records the values representative of the local network connectivity status, here in the format (neighbors, distance, wires).

In this case, the values for the master node 100 are (2,0,0), as it is a master node and is wirelessly connected to two neighbour, slave nodes. The values for the slave node 110 are (2,1,0), as the slave node 110 is wirelessly connected to two neighbour nodes, one being the master node 100.

The radio communication module 141 of slave node 140 then scans on radio channel 1, receiving beacon frames from the slave nodes 120 and 130. The slave node 140 records the connectivity information thereby received: (1,2,0) for slave node 120, as it is wirelessly connected to a single neighbour node with two intermediate links between it and the master node 100, and (1,1,0) for the slave node 130.

In the present example, only one radio communication module of the uninitialized node is able to connect with other nodes in the network. However, in other implementations of this method there may be multiple radio communication modules which are capable of establishing a connection with neighbour nodes. It will be readily recognized that the term (channel assignment' may in some embodiments encompass the assignment of channels to multiple radio communication modules. In such cases, each permutation of channel assignments to the radio communication modules of the node constitutes a possible channel assignment' for the purposes of applying this method.

At this point, the discovery phase is complete, and the slave node 140 will analyse the received network connectivity information, in a step for determining the possible channel assignments for the radio communication module(s) of the uninitialized slave node 140. Each of the possible channel assignments will establish at least one direct wireless link between the slave node 140 and at least one neighbour node.

In the present example, there is a first option where radio channel 2 is assigned to the communication module 141, yielding connectivity information values of (2,1,0) for the slave node 140, (3,0,0) for the master node 100, and (3,1,0) for the slave node 110; the values for the slave nodes 120 and 130 would remain unchanged at (1,2,0) and (1,1,0), respectively.

In a second option, radio channel 1 is assigned to the communication module 141, yielding connectivity information values of (2,2,0) for the slave node 140, (2,1,0) for the slave node 130, and (2,2,0) for the slave node 120; the values for the master node 100 and the slave node 110 would remain unchanged at (2,0,0) and 2,1,0), respectively.

The decision as to which radio channel to allocate to the communication module 141 will depend on an optimisation value calculated for each possible channel assignment. In a broad sense, this means using the optimisation value to select the channel assignment which yields the best possible network connectivity.

More specifically, the decision is made by first selecting a topological aspect of the network to optimize, and determining a target optimisation value that corresponds to an idealized situation where the network achieves the maximum possible in connectivity. The possible channel assignment having a target optimisation value closest to this is then implemented.

Preferably, the optimisation value quantifies a topological aspect of the mesh network, by which it describes how the nodes in the mesh network are connected to each other.

The topological aspect in question is ideally one which has a direct effect upon the performance of the network; in this way, it is possible to assess how any particular possible channel assignment will affect the performance of the mesh network, and to compare possible channel assignments against each other.

In most embodiments, it is envisioned that one, two, or three of the topological aspects described herein can be employed.

It may be sought to minimize the network latency between the uninitialized slave node and the master node. Since network latency increases as the number of links between a slave node and the master node increases, the uninitialized slave node should be connected to the master node via as few links as possible.

In such a case, the optimisation value for each possible channel assignment is equal to the distance value that would result, and the target optimisation value is I (indicating a direct connection between the uninitialized node and the master node).

Thus, returning to the example, if it is sought to minimize the distance between the slave node 140 and the master node 100, the slave node 140 will adopt the first option as depicted in Figure 2. In this option, the distance value for the slave node 140 is I (versus distance = 2 for the second option).

If, however, it is sought to maximize the robustness of the network to broken or lost connections, it is ideal to have as many connections between the nodes as possible. This way, if a connection between nodes is severed, there remains at least one other path by which each of the nodes in question can communicate with the master node.

In such a case, the optimisation value for each possible channel assignment would be equal to the number of nodes in the mesh network having a neighbors value less than or equal to 1. Since nodes connected to only one other node are undesirable, the target optimisation value is zero -in other words, as many of the nodes as possible are connected to two or more other nodes.

Applying this results in the second option (depicted in Figure 3) being selected, as with this option the value of neighbors for all nodes is 2 (Cf. Figure 2, where as noted above two nodes have a neighbors value of 1), and thus the optimisation value is zero. If one of the wireless links 161, 162, 163, 301, 302 is broken, the neighbors value will decrease to 1 for at least two nodes, but all nodes will remain connected to the master node 100 either directly or indirectly and thus remain synchronized.

In a third option, the topological aspect quantified by the optimisation value is the presence of a wired connection between the uninitialized node and the master node. The optimisation value is thus equal to the wires value, and the target optimisation value is equal to the distance value, or alternately infinity (i.e. maximizing the number of wired links overall without regard to the distance between the uninitialized node and the master node).

In this way, the mesh network may be configured to prioritize the usage of more reliable wired links between the nodes.

Of course, depending on the particularities of the mesh network and the performance characteristics sought to be realised, it may be preferable to determine the target optimisation values in a different way, or to quantify a topological aspect of the network different from those discussed above. It should, therefore, be understood that one skilled in the art will be readily able to adapt this method to the peculiarities of any specific implementation.

Figure 4 illustrates the usage of the wires section of the beacon frames. If the slave node 160 selects radio channel 2 for its radio communication module 151, then its connectivity information values would be (1,3,0): one neighbour (slave node 140) and 3 wireless links between it and the master node 100. If the slave node 160 selects radio channel I for its radio communication module 151, its connectivity information values would be (1,3,1): one neighbour (slave node 120), 3 links between it and the master node, of which one is a wired link and two are wireless links.

The connectivity of the slave nodes 120 and 140 being equivalent in both cases, the remaining criterion is the robustness of the connection between the slave node 160 and the master node 100. The second option -assigning radio channel I to the radio communication module 151 -is selected as it takes advantage of the wired link, and therefore offers a more robust connection.

Of course, the above-mentioned criteria may be combined so as to determine the channel assignment having optimal connectivity. This is particularly useful when there exists a plurality of possible channel assignments which yield an equally-good connectivity according to one criterion; by iterating the algorithm with additional, different criteria, the user determines the possible channel assignment having the best possible connectivity. This may be performed, for instance, by applying a first test according to a first criterion, determining a sub-set of possible channel assignments which yield a substantially equal level of connectivity according to that first criterion, and then evaluating that sub-set according to a second criterion (preferably different from the first), and so on until a satisfactory level of optimisation is reached.

For instance, it may be preferable to assess the possible channel assignments by first determining which possible channel assignments yield a minimal number of nodes connected to only one other node. If there is more than one possible channel assignment fulfilling that condition, then the sub-set of such channel assignments is assessed to determine which yields a minimal distance value, and so on. The precise number of criteria and the order in which they are applied to the possible channel assignments may thus vary, and one skilled in the art will be readily able to adapt the algorithm to the particulars of the application in question.

The following Figures describe an algorithm that may be applied by all types of nodes and in all situations.

Figure 6 is a flow chart of the algorithm 60 executed by a slave node upon joining the network.

In step 601, the slave node starts the operation by detecting its physical connectivity: the presence and number of wired communication links and wireless communication modules present in the slave node. This can be performed by e.g. reading status registers available in the communication modules.

The antenna discovery phase starts in step 602, by selecting a first radio communication module and then, in step 603, selecting a first radio channel. In step 604, the slave node listens to the medium to receive a beacon frame from nodes already integrated in the network. If directional antennas are used, the slave node may also scan the medium trying several different antenna settings.

The preamble in the beacon frame enables the detection of the frame by the physical layer.

After receiving the frame payload, the slave node can detect the beacon frame thanks to the frame check sequence.

Once a beacon frame has been identified, the slave node gets the identifier and connectivity information (neighbors, distance, and wires fields) from the transmitting node.

After a predefined time has elapsed, the slave node moves to step 605 to check if all of the radio channels have been tested. If there remain radio channels which have not yet been tested, the node returns to step 603 to select another radio channel and reiterate the information gathering of step 604.

If all radio channels have been tested, the slave node checks in step 606 whether all radio communication modules have been tested. If this is not the case, the node returns to step 602 to select another radio communication module.

Otherwise, the slave node moves to step 607 to analyse the received connectivity information, including the connectivity information received from its neighbour nodes connected by wires.

At this point, the slave node executes the algorithm described in the section above, which will now be discussed.

In step 607, the slave node determines the set of possible channel assignments for each of the radio communication modules incorporated therein.

Following this, in step 608 the slave node selects a topological aspect from which the optimisation value will be determined for each member of the set of possible channel assignments. In step 609, the target optimisation value is determined; the target optimisation value will depend on the topological aspect quantified by the optimisation value, and thus may be an integer (e.g. 2), a binary state (Le. yes/no), a limit value (e.g. infinity), or some other value representing an idealized mesh network.

Once the target optimisation value is determined in step 609, in step 610 the slave node calculates an optimisation value for each of the possible channel assignments determined in step 607. In step 611, the algorithm assesses whether only one possible channel assignments has an optimisation value closest to the target value.

Depending on the configuration of the network, it may be the case that several possible channel assignments have an optimisation value closest to the target optimisation value. In such a circumstance, this subset of possible channel assignments is retained in step 612. The algorithm then performs a second iteration of steps 608, 609, 610, and 611, starting from the subset retained in step 612 and employing a different topological aspect of the mesh network in steps 608, 609, and 610. This may be repeated as many times as necessary, until only one possible channel assignment remains (Le. the one which yields the best possible connectivity). Of course, the selection of the topological aspects done in step 608, and the order in which they are applied to the possible radio channel assignments over the several iterations of this portion of the algorithm, will depend on the particulars of the implementation in question, and may be determined by one skilled in the art according thereto.

In this way, the possible channel assignment having an optimisation value closest to the target value is selected in step 613.

Thanks to the timing information included in the beacon frame, the slave node is able to identify the position of the contention time slot so as to transmit a join request message in step 614. Then in step 615 it waits for the answer from the master node. If no answer is received after several attempts and after a predefined time, the slave node re-starts the discovery phase to account for changes to the environment that may have made the radio channel allocation inappropriate or undesirable in light of the changed situation. If access to the network is granted by the master node, then the slave node identifies its allocated time slot and starts transmitting a beacon frame regularly which includes its connectivity information, as shown here in step 616.

Figure 7 is a flow chart for the algorithm 70 executed by a slave node already incorporated into the network. In the initialization step 701, the slave node starts the process for updating the connectivity information when a new node joins the network. In step 702, the slave node waits for the detection of a join request message, either directly received during the contention time slot of the TDMA sequence, or received form a relay node already incorporated into the network.

Where a join request message is received, the slave node forwards the received join request message, in step 703, so that all nodes in the network receive the message. In step 704, the slave node waits for the granting message from the master node. When no answer is received after a predefined time, the slave node goes back to step 702 to detect another join request message.

Otherwise, in step 705 the slave node checks the reception of the beacon frame from the newly-incorporated node (accounting also for various antenna settings, in the case where directional antennas are used).

If the slave node does not receive this beacon frame, it means the newly introduced node is not visible; there is thus no need to update the connectivity information, and the slave node goes back to step 702. If the new beacon frame is received, the slave node gets the connectivity information from the newly-introduced node in step 706. Then in step 707, the slave node updates its local connectivity information according to the new situation, Le. to take into account the additional neighbour node, any change in the distance to the master node and/or number of wired links between it and the master node. Lastly, in step 708, the slave node updates the content of its own beacon frame.

Figure 8 is a flow chart of the algorithm 80 executed by the master node. In the initialization step 801, the master node starts the process for managing the association of the new node. In step 802, the master node waits for the detection of a join request message, received either directly during the contention time slot of the TDMA sequence, or received from a relay node already inserted in the network. In the case where a join request message is received, the master node can decide in step 803 to grant access to the requesting node according to the status of the network. If this is the case, the master node sends, in step 804, an access-granting message that is also relayed by the slave nodes already incorporated into the network. In step 805, the master node checks the reception of beacon frame from the newly inserted node.

If the master node does not receive this beacon frame, it means that the newly-introduced node is not visible, that there is no need to update connectivity information, and the master node thus returns to step 802. If the new beacon frame is received, the master node gets the connectivity information from the newly inserted node in step 806. Then in step 807, the master node updates its local connectivity information according to the new situation (Le. there is one more neighbour node). Lastly, in step 808, the master node updates the content of its beacon frame.

Algorithms described in Figures 6, 7, and 8 are executed when a new node wishes to join the network. However, it should be understood that the radio channel allocation may be changed at any time upon the request of any node, but always under the control of the master node. If the application requesting data transmission over the network is not running correctly, it may be due to changes in the ambient conditions that alter the connectivity between the nodes. It is thus possible to re-allocate radio channels in all nodes according to the new connectivity information.

Figure 9 is a functional block diagram of a node 90 according to an embodiment of the invention. A node comprises a main controller 901 and, in this example, two communication modules (here denoted PHY A and PHY B) 911 and 912.

The main controller 901 is itself composed of -a Random Access Memory ("RAM") 933; -a Read-Only Memory ("ROM") 932; -a microcontroller or Central Processing Unit ("CPU") 931 -a user interface controller 934; -a Medium Access Controller ("MAC") 938; -a Protocol Adaptation Layer ("PAL") 935; and -a Random Access Memory ("RAM") 937.

CPU 931, MAC 938, PAL 935, and user interface controller 934 exchange control information via a communication bus 924, on which is also connected RAM 933 and ROM 932. The CPU 931 controls the overall operation of the node as it is capable of executing, from the RAM 933, instructions pertaining to a computer program once those instructions have been loaded form the ROM 932.

Thanks to the user interface controller 934, the installer can configure a node. For instance, it can indicate which node is the master node of the network.

This interface can be a wired interface such as Ethernet or Universal Serial Bus ("USB"), or a wireless interface such as infrared or Wi-Fi. The user-defined settings are stored in RAM 933.

The PAL 935 performs all necessary transformations of application data which are temporarily stored in application RAM 937. The PAL 935 achieves the interface between applications through the interface 922 and the MAC 938 to transmit and to receive application data on the network.

The MAC 938 is in charge of controlling the transmission and reception of MAC frames conveying control data and application data. In this example, it can rely on the two communication modules 911 and 912. The MAC 938 can be connected to additional communication modules (not shown) and accessible through the interface 921.

The radio communication modules 911, 912 each comprise a modem, a radio module, and antennas. The radio module is responsible for processing a signal output from the modem, which is subsequently broadcast through the antenna. For example, the processing can be done by frequency transposition and power amplification processes. Conversely, the radio module is also responsible for processing a signal received by the antenna before providing it to the modem.

The modem is responsible for modulating and demodulating the digital exchanged with the radio module.

The MAC 938 acts as a synchronization control unit, which controls scheduling of transmissions over the network. This means that the MAC 938 schedules the beginning and the end of a transmission of radio frames over the medium, as well as the beginning and end of the reception of radio frames from the medium. In a preferred embodiment, access to the medium is scheduled according to a TDMA (Time Division Multiple Access) scheme, where each transmission time slot is associated to a unique node (with the notable exception of the contention access time slot). A single MAC frame is transmitted during each transmission slot.

The set of MAC frames transmitted during one TDMA sequence is called a "Superframe." Typically, the duration of the Superframe is 20 ms, and the time slot duration does not exceed 200 ps.

Among the nodes in the network, one is in charge of defining the beginning of each Superirame cycle. For instance, it may be master node 100 of Figure 1, transmitting a first MAC frame at a fixed periodic interval. This MAC frame is generally called a beacon frame, marking the beginning of the Superframe. Other nodes can then determine the beginning of each super-frame cycle according to the reception time of the beacon frame from the master node.

Considering now the reception side, the modem of each of communication modules 911 and 912 collects the radio frames received from the radio module through the antenna, and transmits the radio frames to the MAC 938.

The MAC 938 is therefore able to detect if a radio frame is missing. This detection is made when the MAC 938 is expecting a radio frame during a scheduled receiving time slot but no data is received from the communication module. This situation occurs when the modem has failed to recognize the radio frame preamble: the synchronization was unsuccessful because the signal-to-noise ratio (SNR") or received signal strength indication (RSSI") was too low (La the signal was indiscernible from noise).

Also, the MAC 938 is able to detect transmission errors within a radio frame. As the radio frame payload can be divided into several packets, Cyclic Redundancy Check ("CRC") data computed by the MAC 938 can be appended to the end of each packet. For a given packet received in a destination node, if the CRC computation result is different from that appended to the end of the packet, the MAC 938 can decide to drop the packet as it is very likely to contain erroneous data.

Of course, other variants upon the principles presented above are possible, and the person skilled in the art will readily be able to adapt the above disclosure to the demands of any particular implementation. The above disclosure, therefore, should be regarded as strictly illustrative and non-limiting, and as encompassing any variations arising at least partly therefrom.

Claims (11)

1. CLAIMS1. A method of assigning communication channels in a mesh network (10, 40) comprising multiple nodes (100, 110, 120, 130, 140, 150, 160) and at least one wireless link (161, 162, 163, 201, 202, 301, 302, 402) between two of said nodes (100, 110, 120, 130, 140, 150), comprising the steps of: -identifying, for an uninitialized node (140, 160) not connected to the mesh network (10, 40), at least one neighbour node to which at least one wireless communication channel has been assigned; -determining at least one possible channel assignment for the uninitialized node (140, 160), said at least one possible channel assignment permitting the establishment of at least one direct wireless link (201, 202, 301, 302) between said uninitialized node (140, 160) and said at least one neighbour node; -calculating, for each possible channel assignment, an optimisation value; and -implementing a possible channel assignment having an optimisation value closest to a predetermined target optimisation value.
2. 2. The method of Claim 1, wherein the optimisation value quantifies a topological aspect of the mesh network (10, 40).
3. 3. The method of Claim 2, wherein the topological aspect quantified by the optimisation value corresponds either to the number of nodes (120, 130) in the mesh network (10) connected to only one other node (100, 110), the number of links (161, 162, 163, 201, 202, 301, 302, 402) between the uninitialized node (140, 160) and a master node (100) of said mesh network (10, 40), or the number of wired links (401) between the uninitialized node (160) and said master node (100).
4. 4. The method of any of the preceding claims, wherein when there exists a subset of possible channel assignments each having an optimisation value closest to the target optimisation value, a sub-optimisation value is calculated for each possible channel assignment in said subset, the channel assignment having a sub-optimisation value closest to a pre-determined target sub-optimisation value being thereby implemented.
5. 5. The method of Claim 4, wherein the sub-optimisation value quantifies a topological aspect of the mesh network (10, 40) different from that quantified by the optimisation value.
6. 6. The method of Claim 5, wherein the topological aspect of the mesh network (10, 40) quantified by the sub-optimisation value corresponds either to the number of nodes (120, 130) in the mesh network (10, 40) connected to only one other node (100, 110), the number of links between the uninitialized node (140, 160) and a master node (100) of said mesh network (10, 40), or the number of wired links (401) between the uninitialized node (160) and said master node (100).
7. 7. A method of assigning wireless communication channels in a mesh network (10, 40) comprising multiple nodes (100, 110, 120, 130, 140, 150, 160) and at least one wireless link (161, 162, 163, 201, 202, 301, 302, 402) between two of said nodes (100, 110, 120, 130, 140, 150, 160), comprising the steps of: -identifying, for an uninitialized node (140, 160) not connected to the mesh network (10, 40), at least one neighbour node to which at least one wireless communication channel has been assigned; -determining a plurality of possible channel assignments for the uninitialized node (140, 160), each of said possible channel assignments permitting the establishment of at least one direct wireless link (201, 202, 301, 302) between said uninitialized node (140, 160) and said at least one neighbour node; -for each possible channel assignment, calculating an optimisation value corresponding to the number of nodes (100, 110, 120, 130, 140, 150, 160) in the mesh network (10, 40) connected to only one other node (100, 110, 120, 130, 140, 150, 160); -determining a subset of possible channel assignments each having an optimisation value closest to zero; -for each possible channel assignment in said subset, calculating a sub-optimisation value corresponding to the number of links between the uninitialized node (140, 160) and a master node (100) of said mesh network (10, 40); and -implementing the possible channel assignment having a sub-optimisation value closest to 1.
8. 8. A mesh network (10, 40) configured according to the method of any one of Claims ito 7.
9. 9. A device (900) for assigning communication channels in a mesh network comprising multiple nodes and at least one wireless link between two of said nodes, characterized in that it comprises: -a means for identifying, for an uninitialized node not connected to the mesh network, at least one neighbour node to which at least one wireless communication channel has been assigned; -a means for determining at least one possible channel assignment for the uninitialized node, said at least one possible channel assignment permitting the establishment of at least one direct wireless link between said uninitialized node and said at least one neighbour node; -a means for calculating, for each possible channel assignment, an optimisation value; and -a means for implementing a possible channel assignment having an optimisation value closest to a predetermined target value.
10. 10. The device of Claim 9, wherein said device (900) is a wireless node for a mesh network.
11. 11. A computer program comprising instructions for implementing a method according to any one of Claims 1 to 7.