WO2011055058A1

WO2011055058A1 - Communication method, transmitter/receiver node, and related computer program

Info

Publication number: WO2011055058A1
Application number: PCT/FR2010/052271
Authority: WO
Inventors: Cedric Gepner; Aymeric Puech; Antoine Venet
Original assignee: Sas Smartgrains
Priority date: 2009-11-06
Filing date: 2010-10-22
Publication date: 2011-05-12
Also published as: EP2497322A1; FR2952500B1; FR2952500A1

Abstract

The invention relates to a method for communicating in a radio communication network (12) comprising a given node (1) and nodes (1), referred to as adjacent nodes, located within the radio range of the given node, said nodes having a means for radio frequency transmission and reception of messages over a shared radio channel (5), said method including the step wherein, during a communication cycle comprising time intervals such that a separate time interval is assigned to each node, a setting is made such that the messages to be transmitted by the adjacent nodes to said given node during said cycle are transmitted by said adjacent nodes during the time interval assigned to the given node.

Description

COMMUNICATION METHOD, TRANSMITTER / RECEIVER NODE AND ASSOCIATED COMPUTER PROGRAM.

The present invention relates to communication techniques used in telecommunications networks, in particular, but not exclusively, in ad hoc networks.

Ad-hoc networks are communication networks with no fixed infrastructure, in which a number of wireless stations are equipped with means of transmission and / or radio reception and adequate protocols to form the nodes of the ad hoc network. These stations forming the ad hoc network may be in the form of computers or laptops, handheld computers, mobile phones, vehicles, household appliances, etc.

The transmission-reception means can also be associated with simple objects such as sensors or actuators. An ad hoc network of sensors makes it possible to collect information, for example to monitor or control installations.

The invention applies for example to a distributed and automated system, for example for the detection of objects or persons for remote activation of equipment. This system is low time latency, consisting of a dense and extensive network of sensors and possibly activators communicating wireless and autonomous energy.

For example, the system is adapted to obtain, in near real-time, a panorama of the available parking spaces in a parking lot, and the consecutive updating of available counting displays, arranged at the entrance of the levels and aisles. of circulation of said parking lot. This also makes it possible to have detailed statistics of the number of visitors to the park, to guide vehicles entering the park to their parking space by means of displays, and thus to speed up the search for available places, so the vehicle turnover rate and the overall average occupancy rate of the fleet; to identify cars exceeding an authorized parking time limit; etc. The commonly adopted solution consists of deploying detection modules equipped with sensors at the same time as wiring to, on the one hand, supply the sensors ("electrical"wiring); and on the other hand, to recover the occupancy information provided by the sensors ("computer" wiring). In the context of application of parking facilities described above, a well-known solution of the state of the art is to dispose of vehicle detectors (generally, ultrasonic sensors, so-called rangefinders, or magnetic sensors) to level of each parking space (or both places if the sensor allows simultaneous detection, called two-seater). However, the scope of the work to be done for the installation and wiring of the sensors (partial closure of the car park, zone after zone, for several weeks) makes the deployment of such solutions, called "wired", slow and expensive.

Another known solution, making it possible to overcome electrical and computer wiring, is to use networks of wireless nodes consisting of autonomous elements that, once disseminated in an environment, are able to take measurements, convey them from node to node over large areas and, depending on the context, to control equipment to which they may have been connected.

The constraints on these networks include:

consumption: typically of the order of a hundred microwatts (order of magnitude to a factor, division or multiplication, of ten), allowing the node to reach, once powered by batteries, an energy autonomy of several years ;

compactness: a volume that is insignificant compared to that of the battery;

- the ability to communicate: able to self-configure to form a wireless communicating network, having the following characteristics:

■ latency: the average time elapsing between the receipt of useful information (eg detection of a vehicle) and its operation (example: update of a seat counting display) - the latency must be precisely defined for each application framework;

■ bandwidth: amount of information transmitted from node to node, per unit of time; ■ reliability: ability to transmit information without losses;

■ robustness: the ability to reconfigure a network in the event that some nodes become faulty;

■ tropism: ability to convey information in defined directions (ascending, descending, bidirectional, iso-tropism);

■ convergence: ability to optimize network characteristics, especially latency, by favoring a propagation direction.

The constraints can be contradictory. For example, in implementations with very low energy consumption, the nodes remain awake only during a very small proportion of time, designated as the "service cycle" and typically of the order of a few percent (for example a service cycle of two percent means that the node is awake an average of twenty milliseconds every second). The rest of the time, the node is idle and consumes a negligible current. Nodes can only communicate during their awakening, the establishment of a service cycle is a source of latency. Reducing consumption requires reducing service cycles, so space them out, which increases latency. Conversely, if the nodes are more often awake, the information will be communicated more quickly (latency will be reduced) at the cost of more consumption.

The expected performance of a system as described above, for example for parking lots, include: a life of the nodes, on battery, of the order of five years; Latency less than five seconds a congestion of the nodes of the order of five centimeters (length), by five centimeters (width), by one centimeter (height). The known state of the art, for these data congestion and latency, and with a battery of the order of five watt hours, reaches an autonomy of the order of one year only.

The nodes generally comprise a microcontroller in charge of controlling the different peripherals and of sequencing the logic operations, a transmitter and / or radio frequency receiver, a detection module (sensor or combination of one or more sensors and signal processing circuits) . The access mechanisms of the microcontroller to the radiofrequency transceiver and, by extension, the organization of radiofrequency exchanges between nodes within radio range of each other under the name of layer or mechanism "MAC" (Medium Access Control) are designated. , while the problems of routing information, from node to node, from a transmitter to a recipient, constitute the problem of "routing".

The MAC mechanism determines at what point a node has the ability to transmit a radio frequency message to a node within radio range (ie at what point in time the destination node will be listening, and the undisturbed wave environment so to allow the propagation of the message), while the routing determines to which next node (ie in which direction of the network of nodes) the radiofrequency message must be communicated in order to take the optimal path towards the final recipient of the message. In the broad sense, these two issues are grouped together in the concept of "wireless sensor network protocol", or Wireless Sensor Networks (WSN) protocol.

State-of-the-art solutions, such as the original ZigBee specification based on the 802.15.4 standard implemented by ΙΊΕΕΕ, provide for tree networks in which the routing of information can only be carried out by powered nodes. The adjustments of this standard for lattice applications, with very low energy consumption, are unsuitable because they sacrifice the latency.

The MAC mechanism can be synchronized temporally according to cycles: the state of the art then speaks of TSMP type protocols (Time Synchronized Mesh Protocol). The MAC mechanisms used in these very low-power WSN protocols, so as to respect very low duty cycles, organize a simultaneous awakening of all nodes that can then exchange messages throughout their waking period. After the exchanges, the nodes return to a standby mode to save energy. One of the difficulties in designing MAC mechanisms lies in the risk of collision of several messages transmitted over the air on the same radio channel at the same time: this risk imposes nodes to emit in turn. Turn-around is determined on the basis of the "first come, first served" principle, the nodes listening to the environment prior to any transmission, and transmitting their message only if the environment is not disturbed by other exchanges of information. messages. If so, they wait for a random time before retrying a broadcast. Alternatively the role can be fixed by a master node, which allocates fixed speech times to each of its neighboring nodes. In all cases, many nodes are awake even though the wave environment is disturbed by neighboring nodes exchanging messages, and this phenomenon is generating "over-listening", that is to say active listening a radio channel by a receiving node, consuming energy, but unsuccessful, that is to say without receiving a signal for this receiving node during listening.

Some attempts to adapt WSN protocols have been made for the specific context of the industrial problem. The "automated system enabling the optimized management in real time of a parking lot" (patent FR 2 918 491 filed July 3, 2007) describes a system based on a network of autonomous nodes partially improving the problems of latency and energy, by organizing node-to-node communications along a looped "circuit" around an administration console: latency can then be defined as the time required for a message to be transmitted from node to node throughout circuit (a "cycle"); by optimizing the waking moments of the nodes, by synchronization, the latency can be reduced in admissible proportions. On the other hand, this technical process does not make it possible to optimize the energy consumption, insofar as each node is solicited, at each cycle, to receive the circulating message and retransmit it to the next node, and this even in cases where no information useful needs to be conveyed. Again, this gap is described in the scientific literature as "over-listening" and "over-transmitting". By "over-transmission" is meant to designate a transmission on the radio channel by an emitter node, energy-consuming, but unsuccessful, no node for which it was intended having listened to the broadcast. In fact, the known hardware implementations of said automated system are provided with bulky batteries and do not respect the congestion constraints imposed by the industrial problem.

On the other hand, the state of the art concerning the detection modules implemented to solve the industrial problem is based on local measurements, which can lead to prediction errors concerning the absence or the presence of the objects. or people. The prediction errors induced are of two types: passage in the free stable state while the object or the person has not moved away (in this case, the prediction is designated as a false negative); symmetrically, passage into the stable state occupied when no object or person has approached (in this case, we speak of false positive).

In the vast majority of cases, the errors are related to: disruptive external events of measured physical quantities (ambiguity 1); or external events that disrupt the operation of the detection module (ambiguity 2); or at a low level of changes in physical quantities induced by the object or person to be detected, with regard to the detection thresholds set (ambiguity 3); or lack of locality changes in physical quantities induced by the object or person to detect, inducing changes of state on several nodes at once (ambiguity 4).

In this typology of ambiguities potentially leading to detection errors, the state of the art provides methods to reduce the ambiguity 2: in some application frameworks, it is possible to take into account physical quantities responsible for deviations of the detection module. The measurement of these physical quantities makes it possible to correct the values returned by the detection module before being processed by the expert system. This method is only applicable in cases where the deviations are predictable. In the illustrative application framework, AMR sensors are highly sensitive to thermal variations; the thermal impact on the amplified voltage measured across the Wheatstone bridge is nonetheless quantifiable, typically of the order of one percent per degree Celsius; however, the AMR sensors are very stable and are not influenced by other physical quantities likely to vary in the deployment context (wind, visibility, sound, vibrations, humidity, mechanical pressure). For example, if the nodes are equipped with temperature sensors, typically integrated in the microcontroller, the measurement of the temperature variations allows the recalibration of the input data, ultimately reducing the ambiguity 2. On the other hand, the known state of the This technique does not provide a reliable and generalizable method for reducing ambiguities of types 1, 3 and 4.

The invention also provides such a method, based on the operation of the detection module within a wireless sensor network.

According to a first aspect, the invention proposes a method of communication in a radio network comprising a given node and the set of nodes, called neighboring nodes, being in radio range of the given node, said nodes being provided with means of communication. radio transmission and reception of messages on a shared radio channel.

The method comprises the step in which, during a communication cycle having disjoint time intervals such that each node is assigned one of said time slots:

it is imposed that the messages to be transmitted during said cycle by the neighboring nodes to said given node are sent by said neighboring nodes during the time slot assigned to the given node.

Such provisions have the effect of reducing the over-listening in the network, and consequently reducing the energy consumption of the nodes, which has the effect of prolonging their lifetime in dense and extensive networks of nodes.

In one embodiment, the given node is required to actively listen for the shared radio channel for at least a portion of the time slot assigned to said given node.

In one embodiment, it is imposed that the messages to be sent during said cycle by the given node to a neighboring node are transmitted during the time slot assigned to said neighbor node. In one embodiment, it is imposed that the messages to be sent during said cycle by the other nodes to said neighboring node are also transmitted during the time slot assigned to said neighboring node.

These provisions also have the effect of reducing the over-listening in the network, and consequently reducing the energy consumption of the nodes, which has the effect of prolonging their lifetime in dense and extensive networks of nodes.

In one embodiment, the communication cycle comprises a first period comprising said time slots, and a second period during which the nodes remain in standby, the two periods being consecutive. This characteristic has the effect of reducing the energy consumption of the nodes.

In one embodiment, during a time slot assigned to a neighbor node to which the given node has no message to send, said given node is in standby. This characteristic has the effect of reducing the energy consumption of the nodes.

In one embodiment, a hierarchy is defined between the nodes and wherein the given node selects from among its neighbor nodes the node at the maximum hierarchical level, higher than the level of the given node, and the given node synchronizes its internal clock on the clock. internal of the selected node. This makes it possible to avoid a drift by node group of the synchronization between the nodes.

In one embodiment, the given node synchronizes with a signal from the selected node during the time slot assigned to said selected node.

In one embodiment, the given node transmits a signal during the cycle indicating the assignment of the time slots to the neighboring nodes and to the given node. This arrangement allows the nodes to know the assignment of the intervals in their radio neighborhood and the assignment in the vicinity of each of their neighboring node.

In one embodiment, the time slot assigned to the given node is chosen distinct from the time slots assigned to the neighboring nodes and is furthermore chosen separately from the time slots assigned to the radio range nodes of said neighboring nodes.

In one embodiment, the time slot assigned to the given node is chosen:

according to relative distances between nodes and a node corresponding to the final destination of the message, the distances being evaluated according to the minimum number of intermediate nodes on the path of the radiofrequency message between the given node and the node corresponding to the final destination of the message; message; and or

in addition, as a function of relative distances between nodes and a node corresponding to the final destination of the majority of the messages, the distance between two nodes being measured as the quality of the radiofrequency link between said nodes.

In one embodiment, the given node transmits during the cycle a signal indicating in which future cycles it will not remain idle during the time slot assigned to it. This arrangement makes it possible to further reduce consumption without penalizing the efficiency of the network.

The invention further provides a method that can be implemented alone or in combination with the communication method described above. According to this method, the nodes are furthermore equipped with sensors adapted to make measurements of variations in physical quantities and a node is adapted to perform a processing according to the measurements made by the sensor associated with the node.

According to this method, the measurements made by at least one sensor of a node adjacent to the given node are sent to the given node, and the given node performs a processing based on at least one measurement performed by the sensor associated with the node. given and said measurement sent by the neighbor node.

Such a method makes it possible to use the measurements transmitted by one node in the processing carried out by another node. This arrangement makes it possible to pool at the nodes themselves the information gathered by different nodes, while using the distributed processing capabilities of the nodes.

According to a second aspect, the invention proposes a radiofrequency transmitter / receiver node of messages for a radio communication network, said node being adapted to implement the steps, which are incumbent on the given node, of a method according to the first aspect of the invention. 'invention.

According to a third aspect, the invention proposes a computer program to be installed in a radiofrequency transmitter / receiver node of messages for a radio communication network comprising nodes each equipped with radio transmission and reception means for messages, said program comprising instructions for implementing the steps that are incumbent on the given node, a method according to the first aspect of the invention during execution of the program by processing means of said node.

Other features and advantages of the invention will become apparent on reading the description which follows. This is purely illustrative and should be read in conjunction with the attached drawings in which:

- Figure 1 is a detection system of a parking lot in an embodiment of the invention;

FIG. 2 represents a node of the detection system of FIG.

1;

FIG. 3 illustrates a communication cycle in one embodiment of the invention.

In Figure 1 is shown a detection system 100 of a car park in one embodiment of the invention.

The detection system comprises a computer 10 serving as an administration console, and a wireless communication network 12 comprising a plurality of transceiver stations 2 each intended to constitute a node of the network. The nodes of the network are of various types among which are for example:

nodes 1 performing vehicle detection, designated as the sensor nodes; - nodes 2 provided with indicator lights, designated as the beacon nodes, allowing, in the illustrated case, to indicate to the motorist user the precise location of the available places (green light) or occupied (red light) in the park parking ;

- Nodes 3 backed by equipment (eg displays of available space counting), designated as activating nodes;

and finally, a single node 4 connected to the administration console 10 of the system, and designated as the gateway 4.

This distinction between types is functional because all nodes, of all types, share the same basic hardware architecture shown in Figure 2.

Thus each node comprises a processing module 6, a transmission / reception module 7 which provides the physical layer and link layer processing (layers 1 and 2 of the OSI model) with a view to exchanging signals with a neighboring node via a shared radio channel 5. The transmission / reception module 7 is connected to an antenna 7 '. Control means 9, for example a very low power microcontroller, drive the various modules, and are particularly adapted to control the power on and off (standby) of the transmitting / receiving module 7. The node comprises in addition a quartz oscillating at a frequency of 32768 hertz for example.

In addition, in the case of a sensor node 1, the processing module 6 is provided with a person detection module or object with very low consumption, connected and controlled by the microcontroller 9, with analog or digital output .

The sensor nodes are fed alternately by a battery or an accumulator recharged by a solar panel placed in the upper part of the housing.

The beacon nodes 2 are provided with light diodes connected to the microcontroller 9. The power supply is alternately provided by a battery recharged by a solar panel placed in the upper part of the housing, or by connection to another external power source; in this last case we speak of external power supply (example: grooving of the roadway allowing the passage of an electric cable connected to the nodes).

The activator nodes 3 are provided with a serial type of wired connection, communicating for example according to the RS-232 or RS-485 standards, and connected to the equipment that the pilot node. The power supply is external.

The gateway 4 is identical to the activators 3.

In one embodiment, the detection module selected for the sensor nodes 1 is a magnetic sensor type AMR (anisotropic magneto-resistive sensor) associated with signal amplification circuits. The principle of operation is as follows: the presence of ferromagnetic materials in vehicles (especially in the chassis) modifies the lines of the Earth's magnetic field. The AMR sensor consists of an assembly of Wheatstone bridge resistors. One or more of the bridge resistors consist of a specific material, called AMR, whose resistance varies depending on the intensity of the ambient magnetic field (component in the axis of the material). The measurement of the amplified voltage across the Wheatstone bridge is a good approximation of the field strength in the mounting direction of the sensor. This voltage varies in the presence of a vehicle, due to changes in the Earth's magnetic field induced by the vehicle.

The detection of objects or persons is performed by the sensor nodes 1, according to a very precise measurement cycle clocking.

The typical duration of a measurement cycle is one thousand and twenty-four milliseconds, but the sensing module is only activated for the duration of the measurement - in the case illustrated, approximately three milliseconds. This duration can be lengthened dynamically in situations where the deployment environment allows an increase in latency - for example, in the case of a parking lot, when said park is far from being saturated with vehicles.

At each measurement cycle, the detection module is powered, a measurement is taken, and in the processing module 6 of a sensor node 1, an on-board processing algorithm associating an expert system with signal processing methods, deals with the measure in order to decide the occurrence of a change of state or not (example: arrival or departure of a vehicle).

The method used by the expert system is for example that of a state machine, such as those used in AFD (Deterministic Finite Automata): two so-called stable states, called "busy" and "free", respectively correspond to the detections of presence and absence of the object or the person on the site. Two other states named "free? And "busy? Are states of uncertainty, through which the expert system transits between two stable states, respectively between busy and free, and between free and busy.

Transition conditions are thus defined between the following pairs of states: (free, busy?), (Busy ?, busy), (busy, free?), (Free ?, free). Transitions to a state of uncertainty, free? or busy ?, are conditioned to exceed a threshold by the local energy level of the signal, defined as the square-to-average distance of the series of measurements returned by the detection module. The transitions to a stable state are conditioned by the stabilization of the signal, smoothed by a low-pass filter, around values compatible with the stable state in question (in the case illustrated: stabilization around measured magnetic field values compatible with the presence or absence of a vehicle).

Once the stable state, free or busy, in which the expert system is located, the processing module 6 of the sensor node 1 decides whether useful information must be transmitted to other nodes - in particular the tag nodes 2, the activator nodes 3 and / or gateway 4.

In the embodiment considered, the useful information determined by the sensor node 1 assigned to a parking space is a change of occupation of this parking space. Thus, when a change of occupation is determined, this information must be the subject of a communicated report:

at the beacon node 2 placed on the roadway, in front of the parking space assigned to the sensor node 1; at the node (s) activator (s) (3) loaded (s) to update the (s) display (s) available seating arranged in the area of the sensor node (1);

finally, at the gateway 4 connected to the administration console 10 of the system 100, for the update of a parking management interface (plan indicating the occupation panorama of the places, tracking cars parked beyond a time limit, detailed statistics, etc.).

The communications implemented on the shared radio channel 5 concerned comprise, for example, the dialogue between the gateway node 4 and a specific activator node, for controlling a device from the console 10 (example: closing a barrier giving on a parking aisle, updating of guidance instructions, etc.); the report made by a beacon node 2 or an activator node 3 to the gateway node 4 (example: need for maintenance, current energy level, etc.); etc.

The technical process of inter-node communication is as follows: each node is associated with a unique identifier (unique address of the node), dynamically built during its first activation.

The system is configured such that each node having useful information to transmit knows the unique address of the destination nodes of this information and indicates this address in a header of the message containing the information.

The routing method used is based on known mechanisms of the state of the art, and inspired by research work conducted in the French laboratories of INRIA (National Institute of Research in Computer Science and Automation), summarized in the notes of Conference of the International Conference on Mobile Ad Hoc and Sensor Networks (MSNO8) in Wuhan, China, in 2008. This mechanism is described in particular in the article: HECTOR, an efficient Energy Tree-based Optimized Routing Protocol for wireless networks Nathalie Mitton, Tahiry Razafindralambo, David Simplot-Ry and Ivan Stojmenovic. A double addressing system is superimposed on the unique addresses: one is associated with a routing algorithm guaranteeing, regardless of the configuration of the network 12, the existence of a path, from node to node, from the node X initial sender of the message, to the final destination node Y of the message. The other is associated with a competing routing algorithm, providing an optimal path but no guarantee of delivery. The exploitation of these two addresses allows a node X, wishing to transmit information to a node Y which is not directly in radio range, to select the best next neighbor Z, within radio range of X, for routing some information. Once this is transmitted to Z, the routing problem is postponed for a routing from Z to Y. This mechanism is iterated until the information is transmitted to a node neighbor of Y, which finally communicates to Y .

A method of communication in an embodiment of the invention is such that each node is assigned a time slot (in English "time slot") in which it alone, among the nodes in its radio neighborhood ( ie within radio range of the node) is entitled to receive messages.

Consider, for example, any node N1 in the network 12 and the set of nodes in radio range of the node N1 named neighboring nodes of the node N1.

The communication is organized by successive communication cycles (in English "timeframes") coinciding with the measurement cycles described above and whose duration is for example one thousand-twenty-four milliseconds.

With reference to FIG. 3, each communication cycle C is segmented into two periods: a sleep period T1 and a wake period T2.

The period T1 is a period where the nodes are in standby mode (ie the radio transmission / reception means are idle and do not consume energy), controlled by their microcontroller and which gives rise to no event.

The period T2 is a period during which at least some of the nodes awaken under the control of their microcontroller for at least a certain time. The period T2 is segmented into time intervals T. For example the duration of each time interval T is sixteen milliseconds, the number of time intervals T is sixteen in the period T2, which leads to a respective duration of the waking period T2 equal to two hundred and fifty-six milliseconds, and to a duration equal to one hundred and sixty-eight milliseconds for the idle period T1.

In one embodiment of the invention, each radio node N1 is assigned a specific time interval, different from the time slots respectively assigned to neighboring nodes of the node N1.

During a communication cycle C, the neighbor nodes are adapted to transmit, under the control of their microcontroller 9, messages to said node N1 on the shared radio channel 5 exclusively during the time slot assigned to the node N1.

And the node N1 is controlled by its microcontroller 9, to place itself in active listening of the radio channel 5 at least during part of the time interval which is assigned to the node N1, in order to receive the messages sent to it .

In one embodiment, perfect node synchronization is required within a given radio neighborhood. The synchronization mechanism is distributed, and based on the broadcast on the shared radio channel of beacon signals, containing in particular the addresses of the sending node, the time slot occupation vector in the radio neighborhood of the node (see below). , and network maintenance information.

Each node determines, on the basis of its own addresses and those of its neighboring nodes, a single node "master of time" (in English "clockmaster"). The choice of the "time master" is important to avoid synchronization mechanisms by clique of nodes, synchronized with each other, but whose synchronism is isolated and drifts with respect to the rest of the network.

In one embodiment, a hierarchical structure (tree) of the nodes dominated by the gateway node 4 is defined. Each node chooses as "time master" among its neighboring nodes the node having the highest level in the hierarchy.

During a communication cycle C, the node N1 awakens in particular twice during the waking period T2 under the control of its microcontroller. The first time, during the time slot assigned to its "time master", the node N1 listens to the beacon signal emitted by its "time master", in order to resynchronize its own internal clock on that of the "master of time" sound ". Thus node N1 is programmed to wake up at the beginning of the time interval of the "time master", receive the beacon signal, and then return to standby mode immediately thereafter.

The second awakening is programmed at the beginning of the time interval dedicated to the node N1 itself. The node N1 then sends its own beacon signal, indicating that the time slot that has been assigned to it starts and that it will be placed, immediately after the sending of its beacon signal, by listening to neighboring nodes desirous of it. communicate a message. The node N1 can itself be "master of time" of other neighboring nodes, and its beacon signal will then be used for the synchronization of these neighbors. After sending its beacon signal, node N1 goes into receive mode: if a neighbor node communicates a message, node N1 listens to this message, stores it in its volatile memory, sends an acknowledgment to confirm receipt, and is placed again in receive mode listening for any other messages. If no neighbor communicates message within a period of time, typically much less than the duration of the time slot that has been assigned to him, then the node N1 turns off its antenna and goes back to standby mode.

Node N1 will also wake up during the time slots allocated to neighboring nodes to which it must transmit a message during communication cycle C. On the other hand, node N1 will remain in standby under the control of its microcontroller 9 at the time slots assigned to neighboring nodes (outside of its "time master" node) to which node N1 does not have a message to be transmitted.

The time slot allocation mechanism is based on the sharing of time slot occupancy vectors in the radio neighborhood of each node, mentioned above. This vector (16 bits of data corresponding to the 16 time intervals T of the communication cycle C, and each worth 0 when the time interval is free or 1 when the time slot is already allocated to a node) is sent by every node within its beacon signal: it indicates which time slots are already dedicated to the node and its neighbors.

At the initialization of node N1, it is placed in passive reception mode (it emits no message and listens to its radio neighborhood). At the end of a cycle (approximately one second), all the beacon signals from its neighboring nodes have reached it - in practice, the listening is repeated over several cycles so as to minimize the risk of tag losses related to disturbances of the radiofrequency environment). The node N1 therefore knows the time slots dedicated to its neighbors Xi. In addition, the beacon signals emitted by each neighbor Xi contain the time slot occupation vector in the radio neighborhood of Xi. This information is useful at node N1, because one of the neighbors Xj of the node Xi may not be close to N1, which is not aware of the existence of Xj and yet N1 and Xj must not select the same interval of time within the communication cycle, because then Xi could not simultaneously communicate information to N1 and Xj, and the risk of message collisions would ensue. This would be second-order interference: emissions from neighbors of N1 neighbors could interfere with N1 emissions.

Using the vectors of occupation of the time intervals of its neighbors (logical OR operation on these vectors) the node N1 thus recovers all the time slots occupied by its neighbors and by the neighbors of its neighbors. He can choose among the intervals of time left free by both his neighbors and neighbors of his neighbors, his dedicated time interval, at the beginning of which he then starts to emit his beacon signal. It then executes the communication cycles as described above.

In one embodiment, the choice of a time interval, among those remaining free, can be operated according to a heuristic making it possible to optimize the waking timing of the nodes in the different radio neighborhoods. This heuristic consists in having the given node select, as soon as possible, a time interval that precedes the time slot allocated to the neighboring node receiving a majority of the messages transmitted by the given node. This neighbor node is determined by the routing mechanism by considering neighboring nodes to which the given node has transmitted a message during a fixed number of cycles, and selecting the neighboring node recipient of the maximum messages. The heuristic is therefore located at the border of MAC and routing mechanisms. This provides the system with potential convergence, that is, the ability to optimize message propagation directions to further reduce latency. In particular, in the case of the parking lot, it is expected that the change of occupancy information of a place is raised as quickly as possible, from node to node, to the gateway node 4 - for the purpose of management of the park in real time. For this, the direction "towards the gateway node 4" can be favored by organizing the allocation of the time slots to the nodes so that the nodes that awaken in turn are located in the direction of the journey to be traveled . Each piece of information, collected at a time interval numbered x, can thus realize up to (16-xl) (sixteen minus one minus one) jumps in the direction of the gateway 4 and during the same cycle. On the next cycle, this number is increased to (16-1) (fifteen). The transition from one cycle to another is paid at the cost of additional latency corresponding to the duration of the waking period (typically, seven hundred and sixty-eight milliseconds).

Edge effects are likely to occur in very dense environments, where the numbers of nodes in the neighborhoods are large, typically greater than 16 nodes. In these particular cases, an adaptive mechanism locally changes the number of time slots in each communication cycle from sixteen to thirty-two.

Conversely, in the less dense topologies, or when the communications are very strongly disturbed, the network neighborhoods may be sparsely populated: a range modulation mechanism is then set up, to increase the radio range of each node; at the expense of low overconsumption, the neighborhood of the nodes is thus enlarged.

Finally, in order to further reduce the overall consumption of the nodes, an additional mechanism is set up, consisting in including in the beacon signal of a node a vector (typically, ten bits each worth 0 or 1) indicating in which cycles of future communication the node will be awake. The node may not participate in all cycles and remain in standby longer. This mechanism, fed by a custom heuristic built around business needs, is useful in contexts where latency requirements vary. In the illustrated application case, higher latency can be tolerated when the car park is far from saturation in vehicles.

Such a communication method effectively reduces over-listening and over-transmission of messages and gives rise to very good performance in cases where the network traffic and bandwidth are low (few transmissions), but the needs in terms of latency very high. Such a method of communication between nodes optimizes the constraints relating to latency and energy consumption In this sense it responds well to the constraints encountered in the systems used in parking lots for example.

The communication method according to the invention has been described above in an application for a parking lot. However, this communication method can be implemented for any type of node network.

In one embodiment of the invention, a method of collaboratively detecting objects or people, allows at a sensor node to take advantage of measurements made by another sensor node and improve the reliability of the detection. .

This method can be implemented alone or in combination with the communication method described above.

At the end of a cycle, for example, the neighboring sensor nodes of a sensor node N1 (defined as the radiofrequency range nodes of N1) communicate to N1 the measurements provided by their own detection module. This communication can be systematic (in particular through the maintenance frames of the network synchronization); event (sending local measurements to neighboring nodes at the occurrence of a predefined local event: for example, when the signal energy exceeds a threshold, or when the signal respects a predefined pattern); or on request, a node interrogating its neighbors when necessary, that is to say for the removal of ambiguity, when the local information retrieved by node N1 is not sufficient not to the expert system to make a reliable decision of presence or absence of the object or the person.

In this way, the embedded processing algorithm of N1 bases its decision (whether or not a change of state) on the basis of enriched information: the measurement returned by the detection module of N1, as well as the measurements returned by the detection modules of neighboring nodes of N1. The exploitation of this enriched information is involved in the detection method according to several methods described below by way of example.

In the transition from states of uncertainty to stable states, the expert system of the N1 sensor node compares the variation of the physical quantities measured locally with the variations measured simultaneously at the level of the neighboring nodes: this comparison makes it possible to fill the locality deficit of the modifications of quantities physical effects induced by the arrival (respectively, departure) of the object or person (reduction of ambiguity 4).

The use of this first ambiguity removal mechanism makes it possible to very significantly reduce the erroneous transitions from the states of uncertainty to the stable states. In this way, the thresholds set in the transition functions, for the signal energy, and for the permissible levels of signals in the stable states, can be reduced until it is possible to detect the objects or persons inducing very small modifications of the signals. measured physical quantities (reduction of ambiguity 3).

Finally, the collaborative sharing of measurements between nodes makes it possible to characterize a majority of the disturbing external events of the measured physical quantities, insofar as these are temporary and non-local. This characterization must be done on a case-by-case basis, not only according to the scope of the invention, but also according to the deployment context. As an illustration of this last mechanism of collaborative detection: in the case of a given parking lot, a typology of disturbing events of the ambient magnetic field is drawn up (example: passage of a subway train near the park of parking); the disturbance profile, defined as the amplitude function of the magnetic disturbances as a function of time, is established locally for each node; a global disturbance profile, for a neighborhood of nodes (in particular: for the node N1 and the neighbors of N1), is established; this profile serves to adjust the transition functions, the recognition of a disturbance profile by the N1 node preempting an undesired state change (reduction of the ambiguity 1).

In one embodiment, the methods described above are implemented in particular following the execution in the processing modules of the computer program instruction nodes.

Claims

Communication method in a radiocommunications network (12) comprising a given node (1, 2, 3, 4) and all the nodes (1,

2,

3,

4), said neighboring nodes, located within radio range of the given node, said nodes being equipped with means of radio frequency transmission and reception of messages on a shared radio channel (5),

said method comprising the step according to which, during a communication cycle (C) comprising disjoint time intervals (T) such that each node is assigned one of said time intervals:

- it is required that the messages to be sent during said cycle by neighboring nodes to said given node are sent by said neighboring nodes during the time interval allocated to the given node.

Method according to claim 1, according to which the given node (1, 2) is required to be actively listening to the shared radio channel

(5) for less part of the time interval (T) assigned to said given node.

Method according to claim 1 or 2, according to which it is required that the messages to be transmitted during said cycle (C) by the given node to a neighboring node are transmitted during the time interval (T) assigned to said neighboring node.

Method according to claim 3, according to which it is required that the messages to be transmitted during said cycle (C) by the other nodes intended for said neighboring node are also transmitted during the time interval (T) assigned to said neighboring node.

Method according to any one of the preceding claims, according to which the communication cycle (C) comprises a first period (T2) comprising said time intervals and a second period (T1), during which the nodes remain on standby, the two periods being consecutive.

6. Method according to any one of the preceding claims, according to which, during a time interval (T) assigned to a neighboring node to which the given node has no message to send, said given node is in day before.

7. Method according to any one of the preceding claims, according to which a hierarchy is defined between the nodes (1, 2, 3, 4) and according to which the given node selects among its neighboring nodes the node at the maximum hierarchical level, higher than level of the given node, and the given node synchronizes its internal clock to the internal clock of the selected node.

8. Method according to the preceding claim, according to which the given node synchronizes using a signal transmitted by the selected node during the time interval (T) assigned to said selected node.

9. Method according to any one of the preceding claims, according to which the given node transmits during the cycle (C) a signal indicating the allocation of time intervals (T) to neighboring nodes and to the given node.

10. Method according to any one of the preceding claims, according to which the time interval (T) assigned to the given node is chosen distinct from the time intervals assigned to neighboring nodes and is further chosen distinct from the time intervals assigned to the nodes within radio range of said neighboring nodes.

1 1 . Method according to the preceding claim, according to which the time interval (T) assigned to the given node is chosen in addition:

- depending on relative distances between nodes and a node corresponding to the final destination of the message, the distances being assessed according to the minimum number of intermediate nodes on the path of the radio frequency message between the given node and the node corresponding to the final destination of the message ; and or

- depending on relative distances between nodes and a node corresponding to the final destination of the majority of messages, the distance between two nodes being measured as the quality of the radio frequency link between said nodes.

12. Method according to any one of the preceding claims, according to which the given node transmits during the cycle (C) a signal indicating during which future cycles it will not remain on standby during the time interval (T) which it is affected.

13. Method according to any one of the preceding claims, according to which the nodes (1) are further provided with sensors adapted to carry out measurements of variations of physical quantities and a node being adapted to carry out processing according to the measurements carried out by the sensor associated with the node,

according to which the measurements carried out by at least one sensor of a node neighboring the given node are sent to the given node, and according to which the given node carries out processing based on at least one measurement carried out by the sensor associated with the node given and said measurement transmitted by the neighboring node.

14. Transmitter/receiver node (1, 2, 3, 4) of radio frequency messages for a radiocommunications network, said node being adapted to implement the steps, which are the responsibility of the given node, of a method according to any one of the preceding claims.

15. Computer program to be installed in a node (1, 2, 3, 4) of a radiocommunications network (12) comprising nodes each provided with means of radio frequency transmission and reception of messages, said program comprising instructions to implement the steps which are the responsibility of the given node, of a method according to any one of claims 1 to 13 during an execution of the program by processing means of said node.