Method for controlling at least one access point of at least one wireless network

Abstract

Method for controlling at least one access point to at least one wireless network. The invention relates to a method for controlling a first access point to a communication network via a first radio communication channel, comprising activating and / or deactivating the first radio communication channel based on at least one parameter relating to the quality of service of the first access point for the first radio communication channel for a first communication device of the communication network. Figure for the abstract: Fig. 1

Description

Title of the invention: Method for controlling at least one access point of at least one wireless network 1. Technical field

[0001] The present application relates to the field of radio frequency communications, for example to wireless communications based on technologies allowing the use, by one or more access points of a wireless communication network, of at least two distinct frequency bands for the transmission and / or reception of data.

[0002] It relates in particular to a method of controlling at least one access point of at least one wireless network, as well as a corresponding electronic device, computer program product and medium. 2. State of the art

[0003] In order to offer their user equipment better transmission capabilities, communication devices providing radio frequency access to a wireless communication network often use several frequency bands (also called radio frequency channels) to transmit or receive data with their user equipment. This offers numerous advantages, such as an increase in the effective transmission rate from the perspective of a user's equipment, or the possibility of having different transmission power levels, energy consumption, and maximum data rates depending on the different frequency bands.

[0004] However, using multiple frequency bands can also increase the overall energy consumption of these devices, even when they are not transmitting data. For example, the hardware and / or software components required for multi-frequency transmission or reception may be more numerous or more complex than the components used for transmission and / or reception with a single frequency, sometimes increasing the residual power consumption of the device during periods of inactivity.The communication protocol used may also sometimes require exchanges between access points and equipment using these access points (such as the transmission or reception of signaling frames between these devices) for certain frequency band(s) used by this equipment, and this can also increase the energy consumption of a device operating in multiple frequency bands compared to its consumption in single-frequency operation.

[0005] For example, in the case of a device (such as an interconnection gateway or a repeater) comprising at least one access point using a wireless transmission technology based on the IEEE 802.11 standard and its evolutions, also known as Wi-Fi (for "Wireless Fidelity" in English), the switched-on electronic components of the device consume energy due to the signaling data exchanges provided for in the 802.11 protocol on each frequency band (2.4GHz, and 5GHz in this example).Indeed, the protocol notably provides for the regular transmission, by the access point, of beacon-type signaling frames for the purpose of discovery and / or synchronization with other communication equipment on the network, including other access points (such as another gateway or a Wi-Fi repeater) or equipment using these access points (such as user terminals (phone, computer, tablets, etc.), connected objects or Wi-Fi repeaters).

[0006] The transmission of data or signaling frames in a frequency band involves many components, some of which (such as antennas for example) are specific to the frequency band used.

[0007] Leaving all components of the transmission chain (transmission and / or reception) switched on across all frequency bands results in an expenditure of energy that is unnecessary during periods of inactivity.

[0008] Furthermore, it is not possible to analyze data transiting over the network in real time. On the one hand, analyzing client traffic ("useful" data) in real time is often prohibited. On the other hand, analyzing certain data streams (useful or signaling data) may not be possible. This analysis is particularly impossible for protected streams circulating via virtual private networks (VPNs), which are inaccessible from outside these virtual private networks.

[0009] Some prior art solutions teach, for communication devices such as Wi-Fi gateways, which can operate in both the 2.4GHz and 5GHz frequency bands, to cut off the 5GHz frequency band of these devices during their periods of inactivity (or low activity) in order to limit their energy consumption.

[0010] For example, some of these solutions disclose switching off certain components (such as antennas) whose operation is imperative for transmission and / or reception in the 5GHz band by these devices, when the measured throughput on a frequency band falls below a certain threshold, and switching these components back on when the measured throughput on the active band exceeds another threshold.

[0011] However, determining these throughput thresholds is a difficult compromise because it is necessary to ensure both permanent connectivity of the device with a satisfactory flow rate (depending on the use of the device) and limit unnecessary energy consumption (related for example to too frequent stops and starts of these components).

[0012] Other known solutions propose stopping components involved in transmission and / or reception based on the occupancy rate of a radio channel (also called "Air-Time" in English terminology), that is, the percentage of time remaining available for transmitting or receiving packets on a radio channel. However, these solutions rely on charts, previously created from empirical measurements, the determination of which is difficult because it is necessary to take into account a very large number of types of devices (or components), types of transmission, the distance between the transmitting and receiving devices, and possible external interference. Therefore, it is necessary to define different types of charts to be able to cover all possible situations. Moreover, it is complex to deduce an occupancy rate in a frequency band (such as band 2).4GHz) of an occupancy rate in another frequency band (such as the 5GHz band).

[0013] Finally, determining the occupancy thresholds for activating or deactivating a component necessary for using a frequency band is a delicate matter. This leads to difficulties in developing nomograms, which depend heavily on the operating conditions of the devices. Furthermore, it is very complex to obtain, in advance, nomograms adapted to new types of equipment or new uses. Therefore, solutions based on nomograms are difficult to generalize.

[0014] The purpose of this application is to propose improvements to at least some of the drawbacks of the prior art. 3. Description of the invention

[0015] The present application aims to improve the situation by means of a method for controlling at least one first access point to at least one communication network via at least one first radio communication channel.

[0016] According to the present application, said method includes an activation and / or deactivation of at least one of said first radio communication channel as a function of at least one parameter relating to a quality of service of said at least one first access point for said first radio communication channel for at least one first communication device of said at least one communication network.

[0017] It should be noted that, depending on the embodiment, the first access point may be a single-band or dual-band access point, and that in certain embodiments applying to the control of a plurality of access points, the plurality of access points may include one or more single-band access point(s) and / or one or more dual-band access point(s).

[0018] Such a process offers the advantage of helping to limit the energy consumption of at least one access point to a wireless communication network while helping to provide an acceptable quality of service (from the point of view of equipment connected to that access point (i.e. at least sufficient to meet the needs of that equipment and the expectations of a user of that equipment).

[0019] In this application, quality of service (QoS) of a communication network means the ability of that network to provide efficient (e.g. in terms of transmission and / or reception delays, and / or bandwidth), reliable and uninterrupted data transmission.

[0020] Quality of Service (QoS) can vary depending on network conditions and impact applications using the network, in terms of performance and / or user experience. It is therefore important to be able to assess and control QoS in a network, particularly for the implementation of applications with real-time constraints between sender and receiver, such as Voice over IP or videoconferencing.

[0021] An example of a QoS-related parameter is the parameter known as RTD (Round Trip Delay) described later.

[0022] A first communication equipment as introduced above may be, for example, a Wifi repeater, or a user device (such as a phone, a computer, a tablet, a connected object) including wireless communication means.

[0023] According to at least one embodiment, the method includes, during said activation, respectively said deactivation, a commissioning, respectively a decommissioning of at least one first electronic component essential to the operation of said at least one first access point via said at least one first radio communication channel.

[0024] Depending on the embodiment, this may be a component of the access point, a device including this access point, or a third-party device supervising this access point.

[0025] According to at least one embodiment, said activation, respectively said deactivation, is a function of at least one current value (actual or estimated) of said at least one parameter.

[0026] According to at least one embodiment, said activation, respectively said deactivation, is a function of a result provided by an inference of an artificial intelligence model whose learning was based on real values ​​of said parameter.

[0027] In particular, according to at least one embodiment, said method includes a commissioning or decommissioning of at least one first electronic component essential to the operation of said at least one first access point via at least one first radio communication channel, said commissioning or decommissioning being a function of at least one actual or estimated current value of at least one parameter relating to a quality of service of said at least one first access point for said first radio communication channel, or of a result provided by the inference of an artificial intelligence model whose learning was based on actual values ​​of said parameter, for at least one first communication equipment of said at least one communication network.

[0028] According to at least one embodiment, said communication network is accessible via at least one second radio communication channel.

[0029] According to at least one embodiment, said at least a first radio communication channel and said at least a second radio communication channel operate in disjoint radio frequency bands.

[0030] According to at least one embodiment, said parameter relating to a quality of service belongs to a group comprising:

[0031] - a duration (RTD) between a dispatch to said first equipment, by said first point access, via said first communication channel, of a data packet and a reception, by said first access point, of said data packet;- a proportion of data packets lost or altered during a transmission, via said at least one first radio communication channel, between said first access point and said first communication equipment.

[0032] - a difference between the RTDs of at least two transmitted data packets consecutively, via said at least one first radio communication channel, through said first access point to said first communication equipment,

[0033] - a fluctuation, for several consecutive packets, of a difference between: • a time interval between the respective times of transmission, via said first communication channel, of two consecutive data packets by said first access point, and • a time interval between the respective moments of reception, via said first communication channel, by the access point, of said two consecutive packets transmitted;

[0034] - an average value of the RTDs over a first time interval;

[0035] - a parameter derived from at least two of the above parameters.

[0036] According to at least one embodiment, said activation is implemented when said variation of the current value of said parameter or said result indicates a decrease in the quality of service of said first access point.

[0037] According to at least one embodiment, said deactivation is implemented when said variation of the current value of said parameter or said result indicates an increase in the quality of service of said first access point.

[0038] According to at least one embodiment, said decommissioning of said component is an event in a group of events comprising: - At least a partial shutdown of said component; - At least a partial shutdown of said component.

[0039] According to at least one embodiment, said commissioning of said component is an event in a group of events comprising: - Ignition of said component; - A startup of said component; - A wake-up call for said component - A combination of the above events.

[0040] According to at least one embodiment, said activation, respectively said deactivation, is a function at least of the current value of said parameter and said current value of said parameter is a value measured and / or calculated from data measured on said network.

[0041] According to at least one embodiment, said activation, respectively said deactivation, is a function at least of the current value of said parameter and said current value of said parameter is an estimated value.

[0042] According to at least one embodiment, said current value is estimated by inference from an artificial intelligence model.

[0043] According to at least one embodiment, said component is put into service for a period at least equal to a first operating period.

[0044] Such an embodiment can ensure a minimum activation time of the first communication channel.

[0045] According to at least one embodiment, said component is taken out of service for a period at least equal to one second operating period.

[0046] Such an embodiment can ensure a minimum deactivation time of the first communication channel.

[0047] According to at least one embodiment, said second operating time is less than said first operating time.

[0048] For example, the first duration may be on the order of several milliseconds and the second duration may be on the order of several seconds.

[0049] According to at least one embodiment, a decrease in the value of said parameter represents an increase in the quality of service of said access point, and said activation is implemented when said current value of said parameter is less than a first value for at least a third operating period.

[0050] According to at least one embodiment, a decrease in the value of said parameter represents an increase in the quality of service of said access point and said deactivation is implemented when said current value of said parameter is greater than a second value (identical or different from the first value introduced above, according to the embodiments) for at least a fourth operating time.

[0051] Such embodiments can help to avoid "bagotages" (repeated activations and deactivations close in time).

[0052] According to at least one embodiment, a current value of said parameter is calculated, for said at least a first radio communication channel of said access point, for at least a second communication equipment of said communication network and said activation, respectively said deactivation, takes into account the current value of said parameter for said at least a second communication equipment of said communication network.

[0053] The second piece of equipment can, for example, be a repeater (or router), or even a user device.

[0054] According to at least one embodiment, said activation, or said deactivation, takes into account the communication capabilities of at least one piece of equipment present on said communication network.

[0055] This equipment present on the network can be, for example, the first and / or second equipment already introduced, or a third-party piece of equipment.

[0056] The features, presented individually in this application in connection with certain embodiments of the control process, can be combined with each other according to other embodiments of this process.

[0057] According to another aspect, the present application also relates to an electronic device adapted to implement the process of the present application in any of its embodiments.

[0058] For example, the present application relates to an electronic device comprising at least one configured processor, for controlling at least one first access point to at least one communication network via at least one first radio communication channel, for:

[0059] an activation and / or deactivation of at least one of said first radio communication channel as a function of at least one parameter relating to a quality of service of said at least one first access point for said first radio communication channel for at least one first communication equipment of said at least one communication network.

[0060] This device, capable of implementing the method of the present application in all its embodiments, may, for example, be a device comprising at least one access point (for example, a Wi-Fi access point or a base station of a cellular network). It may also be a device controlling (remotely, for example) at least one access point of at least one third-party device.

[0061] The present application also relates to a computer program comprising instructions for implementing the various embodiments of the above process, when the program is executed by a processor and a recording medium readable by an electronic device and on which the computer program is recorded.

[0062] For example, the present application thus relates to a computer program comprising instructions for the implementation, when the program is executed by a processor of an electronic device, of a method for controlling at least one first access point to at least one communication network via at least one first radio communication channel, said method comprising an activation and / or a deactivation of at least one of said first radio communication channel as a function of at least one parameter relating to a quality of service of said at least one first access point for said first radio communication channel for at least one first communication device of said at least one communication network.

[0063] For example, the present application also relates to a processor-readable recording medium of an electronic device on which is recorded a computer program comprising instructions for the implementation, when the program is executed by the processor, of a method for controlling at least one first access point to at least one communication network via at least one first radio communication channel, said method comprising an activation and / or deactivation of at least one of said first radio communication channel as a function of at least one parameter relating to a quality of service of said at least one first access point for said first radio communication channel for at least one first communication device of said at least one communication network.

[0064] The programs mentioned above may use any programming language, and be in the form of source code, object code, or intermediate code between source code and object code, such as in a partially compiled form, or in any other desirable form.

[0065] The recording (or information) media referred to in this application may be any entity or device capable of storing the program. For example, a medium may include a storage means, such as a ROM, for example a CD ROM or a microelectronic circuit ROM, or a magnetic recording medium.

[0066] Such a means of storage can, for example, be a hard drive, a flash memory, etc.

[0067] On the other hand, an information carrier can be a transmissible medium such as an electrical or optical signal, which can be transmitted via an electrical or optical cable, by radio, or by other means. A program according to the invention can, in particular, be downloaded via an Internet-type network or via Bluetooth ©.

[0068] Alternatively, an information (or recording) medium may be an integrated circuit in which a program is incorporated, the circuit being adapted to execute or to be used in the execution of any of the embodiments of the method which is the subject of this patent application.

[0069] Generally speaking, in the present application, obtaining an element means, for example, receiving that element from a communication network, acquiring that element (via, for example, user interface elements or sensors), creating that element by various processing means such as copying, encoding, decoding, transformation, etc., and / or accessing that element from a local or remote storage medium accessible to at least one device implementing, at least partially, that obtaining. 4. Brief description of the drawings

[0070] Other features and advantages of the invention will become more apparent upon reading the following description of particular embodiments, given by way of simple illustrative and non-limiting examples, and the accompanying drawings, among which:

[0071] Figure 1 presents a simplified view of a system, cited by way of example, in which at least some embodiments of the method of the present application can be implemented.

[0072] Figure 2 presents a simplified view of a device adapted to implement at least certain embodiments of the process of the present application.

[0073] Figure 3 presents an overview of the control procedure for this application, in some of its embodiments.

[0074] Figure 4 presents two confusion matrices illustrating the results of an experiment. 5. Description of the implementation methods

[0075] The present application aims to control the activation of at least one access point to a wireless communication network, on at least one radio frequency communication channel, in order to limit the energy consumption of that access point, while striving to provide an acceptable quality of service to the equipment using This is at least one access point to the communication network. It may be a multi-band access point, whose activity we want to limit (at least temporarily) to a restricted number of frequency band(s), with at least one frequency band being made (temporarily) inactive, or a single-band access point, which we want to stop (at least temporarily), with other access point(s) remaining active (within the same device or on separate devices located close to each other, for example).

[0076] The number of frequency bands (or radio frequency communication channels) controlled and the number of frequency bands that can be deactivated at the same time may vary according to the embodiments and in particular according to the constraints of the user equipment of this or these access point(s), in terms of energy savings, simplicity, quality of service and / or network reliability.

[0077] Thus, unlike some prior art solutions, the present application proposes to rely, not on the throughput or radio occupancy of radio frequency channels, as perceived by the access points, but on an assessment of the quality of service, as felt by the equipment using the access points, for the control of these access points.

[0078] According to this application, the activation (or deactivation) of a radio frequency channel of an access point is controlled by switching on (or off) at least one component whose operation is necessary for the access point to function in at least one radio frequency communication channel. Switching on / off may, for example, involve a component used specifically for the operation of the access point in the at least one radio frequency communication channel in question.

[0079] For example, in the case of a Wi-Fi access point, such components may include, among other things, a specialized integrated circuit, also called a Wi-Fi chipset, an amplifier and / or filtering module, and / or one or more antennas. Some of these components (such as an antenna) may be specific to a frequency band for a multi-band access point. Similar components may be required for the operation of a cellular network access point. These components may be activated or deactivated when the method of this application is used to control that cellular network access point.

[0080] When a communication channel is deactivated (for example, by switching off the corresponding antennas on the access point), network traffic can automatically focus on the other available communication channel(s).

[0081] For example, in the case of a dual-band Wi-Fi access point, when one Wi-Fi frequency band becomes unavailable, the access point and connected equipment at this access point can automatically adapt by concentrating (and switching if necessary) their traffic on the remaining available frequency band.

[0082] Conversely, when a communication channel is activated (for example, by putting the corresponding antennas on the access point into service), network traffic is distributed over the available communication channels, depending on the technologies of the access point and the equipment using the access point.

[0083] For example, in the case of a dual-band Wi-Fi access point, when both frequency bands become available, the connected stations automatically adapt by switching (or not) their traffic to the newly available frequency band, depending on their respective generation of Wi-Fi technology.

[0084] By "automatic" failover, we mean a switchover of network traffic of a communication equipment (such as a Wi-Fi client station) from one radio communication channel to another without involving voluntary intervention by the user of the equipment in this regard, once certain prerequisites (similar to initial configuration) have possibly been carried out.

[0085] The present application is now described in more detail in relation to [Fig.1].

[0086] Figure 1 represents a telecommunications system 100 in which certain embodiments of the invention can be implemented. The system 100 comprises electronic equipment using a wireless communication network 180, such as a local area network (LAN) and / or a wide area network (WAN). For example, it may be a corporate or home LAN and / or an internet-type WAN, or a cellular network, a GSM (Global System for Mobile Communications) network, a UMTS (Universal Mobile Telecommunications System) network, a Wi-Fi network, etc.

[0087] The wireless network may, in particular, allow communication via at least two distinct communication channels, corresponding to different frequency bands (for example, disjoint frequency bands). This may be, for example, a network using a standard wireless transmission technology, such as an IEEE 802.11 standard (such as the IEEE 802.1 lax standard, also known as Wi-Fi 6), allowing communication in at least two frequency bands. In the case of the IEEE 802.1 lax standard, for example, two frequency bands can be used: a frequency band around 5 GHz and another frequency band around 2.4 GHz (referred to as the 5 GHz band and the 2.4 GHz band, respectively). However, the invention also applies to devices based on other versions of the IEEE 802.11 standards, particularly those allowing the use of at least two distinct frequency bands.The invention also applies to any wireless technology other than Wi-Fi that allows the use of au. less than two distinct radio communication channels, such as the radio technologies of 4G, 5G or 6GHz cellular networks. The wireless network can therefore be, for example, a cellular network (using 4G, 5G, 6G technology in particular), using frequency bands such as 800, 900 MHz, 1.8 GHz, 2.1 GHz, and / or 2.6 GHz.

[0088] The electronic devices of the wireless network 180 can be diverse depending on the embodiment. For example, the wireless network can include several electronic devices, such as user equipment (such as a terminal 110 (such as a laptop, a smartphone, a tablet), a connected object 120, 130 (a television, a refrigerator, a thermostat, or a watch, for example), a decoder (also called a Set-top Box (STB) according to English terminology)) or other communication equipment such as a repeater (or router) 140, and / or a monitoring device 150, and / or a storage device 160. The network can also include devices dedicated to collecting data used by the process of this application.The system 100 may also include network management and / or interconnection elements such as a gateway 170 (or Gateway according to English terminology) interconnecting the communication network 180 with another communication network 190 (for example for access to a remote storage element 192 or a remote application server 194).

[0089] Certain electronic devices 140, 150, 170 of system 100 may be equipped with at least one access point 144, 172, 174, 176 to network 180. Depending on the embodiment, these may be multi-band access point(s), operating in at least two frequency bands, or single-band access point(s), operating in a single frequency band (Fri, Fr2, or Fr3 in the illustrated example). Of course, system 100 may, in certain embodiments, include both at least one single-band access point and at least one multi-band access point.

[0090] Of course, at least some of these devices may be optional. This is particularly the case for the collection and monitoring devices. Indeed, data collection and / or monitoring can, for example, be carried out by the devices equipped with access point(s) themselves, such as the gateway 170 and / or the repeater 140 (or the monitoring device 150 if it exists), or by equipment using some of these access points (such as the equipment 110, 120, 130, 140 already introduced).

[0091] Some 110, 120, 130, 140, 150 of the electronic equipment of system 100 may be equipped with means of communication 112, 114, 122, 124, 132, 142, 152, 154, 156 with at least some of the access points 144, 172, 174, 176 to the network 180. For example, these may be equipment 110, 120, 150 equipped with means of communication operating in several frequency bands or equipment 130, 140 equipped with means of communication operating in a single frequency band.

[0092] Fig. 2 illustrates a simplified structure of an electronic device 200, such as for example certain 140, 150, 170 of the electronic equipment of system 100, adapted to implement the principles of the present application. Thus, in the illustrated example, device 200 can be, for example, an interconnection gateway 170 or a repeater 140. It can also be a monitoring device 150, remotely controlling the activation of at least one radio frequency channel of at least one access point of one of the electronic equipment 140, 170 of the system 100. This monitoring device can be a device dedicated to controlling the activation of radio frequency channels of access point(s) of the system 100 (as illustrated) or be itself an access point (for example, one of the access points whose radio frequency channel is to be controlled, or another access point).It can also be a user device at an access point to be controlled (such as a terminal, connected object, etc.).

[0093] The device 200 includes, in particular, at least one memory M 210. The device 200 may include, in particular, a buffer memory, volatile memory, for example of the RAM (Random Access Memory) type, and / or non-volatile memory (for example of the ROM (Read Only Memory) type). The device 200 may also include a processing unit UT 220, equipped, for example, with at least one processor P 222, and driven by a computer program PG 212 stored in memory M 210. At initialization, the code instructions of the computer program PG are, for example, loaded into RAM before being executed by the processor P.

[0094] Said at least one processor P 222 of the processing unit UT 220 can in particular implement, individually or collectively, any one of the embodiments of the method of the present application (described in particular in relation to [Fig.3]), according to the instructions of the computer program PG.

[0095] The device 200 comprises, or can be coupled to, at least one input / output module 230, 240. Thus, the device comprises, or is coupled to, at least one communication module 230, enabling, in the illustrated example, the device 200 to communicate with other equipment in the system 100, via wired and / or wireless communication interfaces. For example, the device may include several single-band access points, each using a different frequency band 232, 234, 236 (Fri, Fr2, Fr3 in the example in [Fig. 2]), or a single access point using different frequency bands 232, 234, 236 (Fri, Fr2, Fr3 in the example in [Fig. 2]), or at least one single-band access point and / or at least one dual-band access point.

[0096] The device 200 may also include, or be coupled to, at least one other input / output module 240, such as a user interface module for the device 200 (also referred to more simply in this application as a "user interface").

[0097] By user interface of the device, we mean, for example, an interface integrated into the device 200, or a part of a third-party device coupled to that device by wired or wireless means (for example, by the wireless means described above). For example, it could be a secondary display of the device or a set of speakers connected wirelessly to the device

[0098] A user interface can in particular be an "output" user interface adapted for rendering (or controlling a rendering) of an output element of a computer application used by the device 200, for example an application running at least partially on the device 200 or an "online" application running at least partially remotely, for example on the server 194 of the system 100. Examples of output user interfaces of the device include one or more screens, in particular at least one graphics screen (touchscreen for example), one or more speakers, a connected headset.

[0099] By rendering, we mean here a display (or "output" according to English terminology) on at least one user interface, in any form whatsoever, for example including textual, audio and / or video components, or a combination of such components.

[0100] Furthermore, a user interface can be an "input" user interface adapted for acquiring a command from a user of the device 200. This may include an action to be performed in connection with a returned item, and / or a command to be sent to a computer application used by the device 200, for example, an application running at least partially on the device 200 or an "online" application running at least partially remotely, for example on the server 194 of the system 100, or information (such as a configuration parameter) to be saved in a file. Examples of input user interfaces for the device 200 include a sensor, an audio and / or video acquisition means (microphone, camera (webcam), for example), a keyboard, and a mouse.

[0101] Said at least one microprocessor of the device 200 may in particular be adapted for controlling at least one first access point to at least one communication network via at least one first radio communication channel, the control comprising:

[0102] an activation and / or deactivation of at least one of said first radio communication channel as a function of at least one parameter relating to a quality of service of said at least one first access point for said first radio communication channel for at least one first communication device of said at least one communication network.

[0103] Some of the above input-output modules are optional and may therefore be absent from device 200 in some embodiments.

[0104] The term "module" or the term "component" or "element" of the device refers here to a hardware element, in particular a wired one, or a software element, or a combination of at least one hardware element and at least one software element. The method according to the invention can therefore be implemented in various ways, in particular in wired and / or software form.

[0105] Depending on the embodiment, the method can be implemented centrally or distributedly. As noted above, in some embodiments, the method of this application can be implemented (centrally) in a device comprising one or more access points operating, as a whole, on different (for example, disjoint) frequency bands, to control the activation of at least some of these frequency bands. In other embodiments, the method can be implemented (distributedly) in a monitoring device controlling the activation of several access points operating, as a whole, on different frequency bands, at least one of these access points being located on a third-party device (other than the monitoring device).

[0106] Figure 3 illustrates certain embodiments of method 300 of this application, wherein the control of at least one communication channel of at least one access point is a function of at least one current value (actual or estimated) of at least one parameter relating to a quality of service (see below). Method 300 can, for example, be implemented by the electronic device 200 illustrated in Figure 2 (this being, for example, a gateway, a monitoring device, or a repeater).

[0107] As illustrated in [Fig. 3], the method 300 may include obtaining 320 the current value of at least one parameter relating to the quality of service (QoS) of the access point, with respect to at least one communication device (for example, a user device such as a mobile terminal) connected to that access point via at least one communication channel. This obtaining may be performed several times, for example, periodically. A parameter representative of the quality of service of an access point, with respect to a communication device using a communication channel of that access point, may, for example, be: - a time delay (RTD) between the sending, by the access point, via the communication channel considered, of a data packet on the communication network, and the reception, by the access point, of this same packet (sent by the communication equipment);

[0108] This duration is known as "Round-Trip Delay" (RTD) in English terminology. The lower the RTD, the faster and smoother the communication. - a proportion (or percentage) of data packets lost or corrupted during a transmission between the access point and the communication equipment;

[0109] This percentage is known as the "Packet Loss Rate" (PLR), according to English terminology. The lower the PLR, the more reliable the transmission.

[0110] - a difference between the RTDs of at least two packets transmitted consecutively by the access point to the communication equipment;

[0111] - a fluctuation, for several packets, of the difference between the time interval between the respective times of transmission of two consecutive packets by the access point and the time interval between the respective times of reception of these two consecutive packets (transmitted by the user equipment) by the access point;

[0112] This temporal fluctuation (or jitter (also sometimes called "jigue", or "jitter" according to English terminology)) is a parameter that can prove to be very important for real-time services such as voice and video communications, or for certain video games.

[0113] - an average value of the RTDs over a first time interval;

[0114] - a parameter derived from at least two of the above parameters (for example a a combination of at least two of these parameters).

[0115] The parameters described above depend in particular on the characteristics of the network (such as available bandwidth, congestion, noise and / or disturbances in the network), the positioning and characteristics of the devices sending and receiving the packets (mutual distance within the network or from a geographical point of view for example, type and software versions of the processors equipping these devices, etc.).

[0116] In the detailed embodiment, for the sake of simplicity, the RTD is used as a parameter representing the quality of service. Other parameters (such as those introduced above, for example) can of course be used (alone or in combination with other parameters such as the RTD, for example) in other embodiments.

[0117] According to embodiments, the current value of the QoS representative parameter of the access point, as perceived by a communication device (such as a user device) can be obtained differently: in particular, it can be measured dynamically, and / or calculated from data obtained (for example collected) 322 dynamically, or estimated 324 from data obtained 322 dynamically.

[0118] In embodiments where the representative parameter is the RTD (or depends on the RTD), it can, for example, be measured dynamically, for example by measuring the time between the sending of a specific command (e.g., a "ping" command) to the equipment in question by the access point via a given communication channel (a "probe" type operation). This embodiment offers advantages in terms of reliability (since an actual measurement is performed).

[0119] The representative parameter can also be obtained by estimation from the data collected by the device. For example, an estimate of the RTD can be implemented after an attempt at a dynamic measurement of the RTD, for example, when the equipment in question does not respond to the command issued to perform this measurement. In certain embodiments, the representative parameter can be estimated systematically (an actual measurement can optionally be performed periodically to verify the reliability of the estimate).

[0120] Estimating the representative parameter can make it possible to avoid increasing the energy consumption of the access point or overloading the communication channel considered due to sending a command specific to this measurement and the expected response to this command.

[0121] In some embodiments, the representative parameter can be estimated by providing at least some of the collected data to an artificial intelligence model previously trained to predict this representative parameter.

[0122] In such embodiments, prior to or during the implementation of the method of this application, the artificial intelligence model may be trained to adapt the model used to the relevant environment. Such training may, in particular, include providing the model with input data collected under numerous operating conditions (for example, in the presence of varied network traffic), and the value of the representative parameter used, measured under those same operating conditions.

[0123] Examples of data used as input to such a model (for its training and / or inference) include, in embodiments where the representative parameter is the RTD (or depends on the RTD): - Bandwidth used; - An occupancy rate of the radio channel in question; - A packet loss rate (or alternatively, a number of packets lost); - A packet retransmission rate (or alternatively, a number of packet retransmissions); - An attenuation rate of a signal emitted by the access point; - A signal power level emitted by a communication device (such as a user device) and received by the antenna of the access point (RSSI for Received Signal Strength Indication / indicator in English); - Information relating to the sending and / or receiving of data; - A combination of the above data.

[0124] Such data include, for example: - general statistical information about the system; - statistical information specific to each radio frequency channel (in terms of occupancy and quality of these channels, quantity of upstream or downstream traffic on these channels, state of the channel); - statistical information specific to each device connected to a channel (in terms of connection quality and / or quantity of upstream or downstream traffic for each device).

[0125] This data can for example be received in response to commands from the device, these commands being able to use, depending on the embodiment, a format specific to the operating system of the access point in question (such as WL commands specific to the Broadcom operating system), or a format compatible with different operating systems (such as the Prpl © operating system).

[0126] For example, such data may be collected in response to WL commands such as: "wl -i BAND chamin_stats", "wl -i BAND sta_info MAC_ADDRESS", "wl -i BAND counters", "wl -i BAND bss", and / or "wl -i BAND assoclist".

[0127] According to another example, such data can be collected in response to query commands of a data model defined in the Prpl © operating system, such as: “WiFi.Radio.BAND.getRadioStats()”, and / or “WiFi.Radio.BAND.getRadioAirStats()”, “WiFi.AccessPoint.BAND.AssociatedDevice.INDEX.”.

[0128] In some embodiments, the collected data can be formatted ("feature engineering") before being provided as input to the model, during inference.

[0129] The model may, for example, implement a classifier-type machine learning algorithm. This could include models such as the XGBoost model, the LGBM (Light Gradient-Boosting Machine) model, or a decision tree model. It could also be a model based on a multilayer neural network, such as a multilayer perceptron (MLP). The model can, for example, be used in regression mode (to predict the value of the representative parameter which, when provided as input data to the model, results in data, at the output of the model, corresponding to the collected data).

[0130] Using numerous different input data points helps to obtain a model that provides a reliable estimate of the representative parameter, and therefore a reliable representation of the quality of service offered by the access point (via the radio frequency channel considered) to the equipment under consideration. Using fewer input data points can increase the simplicity of the model and thus limit processing time and the use of processing resources. Also, in some embodiments, a selection of the model's input data can be made to obtain a "simplified" ("sparse" in English terminology) model that considers only the most important data points (in terms of their impact on the model's results).

[0131] Once the current value of the parameter representing the quality of service (the RTD in the detailed embodiment) has been obtained 320 for at least one communication device, the method may include a verification 330 that this current value corresponds to an acceptable quality of service. For example, the method may include a comparison of the current value of the parameter to a first constant value, used as a threshold. For example, in the detailed embodiment where the quality of service is evaluated by the current value of the RTD, it may be necessary to verify that the current value of the RTD is indeed less than a first constant value (hereafter denoted RTDmax). This constant value RTDmax may be, for example, a value on the order of one, ten, or a few tens of milliseconds, or a hundred milliseconds such as 1, 10, 15, 20, 25, 30, 35, 100 ms (for example, 25 ms).It can vary depending on the implementation, for example depending on the type(s) of communication equipment and the type of network (Wi-Fi, cellular, etc.) involved.

[0132] Since the parameter (such as the RTD) whose value is being checked represents a quality of service, its comparison to a constant value (RTDmax in our example), representing an expected quality of service, allows the operation of the access point to be controlled by ensuring a certain quality of service (corresponding to an RTD with a value of RTDmax). This is an advantage over prior art solutions based, for example, on a radio channel occupancy rate, which cannot guarantee a user of communication equipment connecting to the access point an expected quality of service.

[0133] In some embodiments, the representative parameter (such as the RTD) can be obtained (by measurement or evaluation) for each connected communication device and each communication channel used by that device to connect to the access point. In other embodiments, the parameter representative (such as RTD) can be obtained for a subset of communication equipment connected to the network.

[0134] For example, in some embodiments, the representative parameter (such as the RTD) can be obtained iteratively (e.g., periodically (with a period on the order of a few seconds, for example)) for each device connected to the access point. Obtaining the value of the representative parameter of a QoS can be performed, depending on the embodiment, in parallel or successively for different devices.

[0135] The value (constant, for example) of the representative parameter used as a threshold (and representative of a desired QoS) can be set by configuration (pre-configured or dynamic) of the device. It may, in particular, depend on the equipment against which the quality of service of the access point is evaluated, the technology implemented by the access point, and the type of service (real-time (voice, for example) or not) to which the communication between the equipment and the access point is linked. For example, in some embodiments, different "threshold" values ​​can be used depending on the services to which communications with equipment are linked. Thus, a first RTDmaxl value (for example, 10 ms) can be used for communications linked to real-time services, while a second RTDmax2 value (for example, 25 ms), higher than RTDmaxl, is used for other services.

[0136] In embodiments implementing communications corresponding to different types of service associated with different threshold values ​​at the same time, the strictest threshold value in terms of quality of service can be used for all communications. Thus, according to the example above, the RTD Maxl threshold value can be used for all communications of all equipment when at least one of these communications is related to a "real-time" service.

[0137] Depending on the embodiment, the threshold values ​​used may be identical for all equipment with which the access point communicates or depend on the equipment considered and / or the services (real-time or not constraints in particular) offered for this equipment.

[0138] When the current value of the representative parameter corresponds to a quality of service of a communication channel of the access point better than expected for the equipment considered (for example in the detailed embodiment when the current value of RTD is less than the constant value RTDmax), it may be possible to decrease the quality of service offered while still maintaining an acceptable quality of service.

[0139] Also, when for at least one piece of equipment, the current value of the representative parameter corresponds to a quality of service of the access point better than expected (i.e. in the example an RTD less than the "threshold" value RTDmax), the method 300 may include a limitation 350 of the number of communication channels activated (i.e. active or available) in order to achieve energy savings, by deactivating at least one communication channel, which is currently activated, with network traffic then being shared on the channels that remain activated.

[0140] The limitation may include taking out of service (for example, putting into sleep mode and / or stopping) at least one component used specifically for operation of the access point in at least one radio communication channel (to be disabled).

[0141] In some embodiments, this limitation 350 can be performed when the current value of the representative parameter corresponds to a quality of service of the access point better than expected for all equipment connected to the network (i.e. in our example if for all equipment the RTD is less than the threshold value RTDMax).

[0142] In some embodiments, the 350 limitation can be performed as soon as the current value of the representative parameter corresponds to a quality of service of the access point better than expected for a piece of equipment (i.e. in our example as soon as the RTD is less than the RTDMax threshold value for one of the pieces of equipment).

[0143] In some embodiments, the 350 limitation can be performed as soon as the current value of the representative parameter corresponds to a better than expected quality of service of the access point for n devices (with n an integer > 1) (i.e. in our example if for n devices the RTD is less than the threshold value RTDMax), or when the percentage of devices, for which the current value of the representative parameter corresponds to a better than expected quality of service of the access point, reaches a certain value (for example 30%, or 70%).

[0144] Waiting until the QoS of the access point is better than expected for all equipment connected to the network (i.e. in our example, waiting until the RTD is less than the RTDMax threshold value for all equipment) before performing a limitation can help to provide advantages in terms of network reliability (since we wait to ensure that the quality of service is much higher than expected for all equipment before voluntarily "deteriorating" this quality of service).

[0145] Limiting the number of channels as soon as the QoS of the access point is better than expected for a subset of equipment can help to save more energy, since the number of channels can for example be reduced earlier and more frequently.

[0146] Note that the limitation of the number of activated channels can be carried out conditionally (340, 342). The method may, for example, include a check 340 of the current activation state of the channels.

[0147] For example, a minimum number of channels that must remain active (greater than or equal to 1) may be required in certain embodiments. An embodiment in which it is required to always have at least one active channel (even if there is no communication on all channels) will allow the connection of a new device or the initiation of communication from a device. In such embodiments, the limitation may only be enforced if the current number of active channels is strictly greater than the minimum number of active channels required.

[0148] In some embodiments, where different activation priorities can be assigned to channels, the check may include selecting a channel to deactivate (or activate as explained later) based on these priorities.

[0149] In some embodiments, only certain channels can be deactivated (others remaining active at all times).

[0150] For example, in some embodiments where the access point is a Wi-Fi access point operating on multiple frequency bands, the 2.4 GHz band may not be able to be disabled, or alternatively, the 5 GHz band may be disabled as a priority. Indeed, the 2.4 GHz Wi-Fi band is historically the first frequency band offered in the 802.11x protocol. Therefore, the vast majority of Wi-Fi equipment (e.g., user terminals) is theoretically capable of falling back to the 2.4 GHz band (in other words, the 2.4 GHz band is the base frequency for Wi-Fi).

[0151] Moreover, since the range of Wi-Fi in the 2.4GHz band is much greater than the range of Wi-Fi in the 5GHz band, equipment previously connected in 5GHz should not have any problem reconnecting in 2.4GHz (from a network coverage point of view).

[0152] Moreover, the limitation 350 of the number of channels being carried out during a period of inactivity or low activity of the equipment, the bandwidth available in 2.4GHz, although lower than the bandwidth available with other communication channels (for example in 5Ghz), is a priori sufficient to allow the maintenance of the quality of service from the point of view of the equipment using this access point.

[0153] The method may include, in certain embodiments, a memorization of the current state of each communication channel, or at least of the communication channels that can be deactivated (or conversely activated), so as for example to more easily determine the channels remaining to be deactivated (or on the contrary to be activated as explained below in case of too low QoS).

[0154] It should be noted that in certain embodiments, the limitation 350 on the number of active channels, in the event of better-than-expected QoS, can be timed 342. For example, the limitation 350 can be applied only if the QoS remains better than expected (for each device, or for a subset of devices) for a certain duration (i.e., in the detailed example, with an RTD remaining above the RTDmax value for this duration). This time-delay duration can be defined by parameterization (static or dynamic). It can optionally vary depending on the devices to which the QoS relates.

[0155] For example, in some embodiments where the access point is a dual-band Wi-Fi access point, with both bands (2.4GHz and 5GHz) commonly activated, the 350 limitation (e.g., a 5GHz band cutoff) may only be implemented if, for all equipment, the RTD (e.g., simulated) remains below the RTDmax value for a timeout period.

[0156] The timeout duration can vary depending on the embodiment. For example, it can range from a few tens of seconds to a few minutes, such as 30 seconds, 1 minute, or 5 minutes. Embodiments where the limitation is time-delayed can prevent very rapid and repeated alternations of channel deactivations and activations (channel chatter). Indeed, such alternations can cause premature aging of the electronic components of the access point and / or equipment using the access point's radio frequency channel, or disrupt the operation of equipment connected to the access point and the associated communications. Thus, communications can, for example, be paused at each alternation. Furthermore, these alternations can potentially generate transmission delays and / or packet loss (thus impairing QoS when repeated).Finally, these repeated alternations are likely to cause repeated peaks in energy consumption (due to the switching on and / or off of components during shutdowns and / or startups), and therefore reduce the energy savings expected from the process described in this application.

[0157] An (optional) time delay before limiting the number of active channels (i.e., before at least one channel is deactivated) has been described above. Similarly, after at least one channel is deactivated, a time delay 352 can (optionally) be activated to prevent the number of active channels from decreasing or increasing too rapidly again due to QoS instability (more precisely, instability of its representative parameter) for at least one of the devices.

[0158] In the previously introduced example of a dual-band Wi-Fi access point, the time delay after disabling a band (5GHz for example) can be on the order of a few seconds to a few tens of seconds, such as a duration of 5, 10, or 15 seconds. Of course, these values ​​may be different in embodiments where the needs of users and / or operators and / or associated uses are different.

[0159] When, for a piece of equipment, the current value of the representative parameter corresponds to a QoS that is lower than expected (the case, in the detailed example, where the current RTD is greater than the constant value RTDmax), the method may include a 360° check of the number of communication channels already activated. If the number of activated channels is less than the number of channels that can be activated simultaneously by the access point, i.e., if at least one communication channel can still be activated, the method may include a 370° activation of at least one communication channel that was previously inactive. This 370° activation may include, in particular, a commissioning (for example, a wake-up and / or a power-on and / or a startup) of all the components of the device and / or the access point necessary for the operation of the access point via this previously inactive channel.

[0160] For example, when the access point is a Wi-Fi access point operating on the 2.4GHz and 5GHz frequency bands, with only the 2.4GHz band being commonly active, the method may include putting into service at least one component necessary for operation of the access point in the 5GHz band (but not necessary for operation in the 2.4GHz frequency band), such as an amplifier connected to an antenna operating in 5GHz, and previously out of service.

[0161] For example, the method may include accessing an activation state (stored, for example) of each communication channel, or of communication channels that may not be active (i.e., not yet activated or deactivated), in order to determine at least one channel to be activated (i.e., for example, verifying that at least one "inactive" (i.e., activatable) channel remains, according to the stored activation states of the channels, and if several channels are inactive, choosing one channel to activate from among its inactive channels). The choice of the channel to be activated may, in particular, take into account the maximum RTD threshold value associated with that channel, when several channels are activatable, so as to activate a channel for which the current value of the representative parameter will correspond to an acceptable QoS.

[0162] Consider, for example, a Wi-Fi access point operating with three frequency bands (2.4 GHz, 5 GHz, and 6 GHz), with only the 2.4 GHz band active. If the RTD (Return Transmission Difference) becomes too high for one of the devices connected to the access point, it is necessary to activate a channel in another frequency band. The choice of the frequency band to activate can be based on the RTD evaluated in the 2.4 GHz band (assuming that an RTD can be transposed from one frequency band to another). For example, if the RTD evaluated for the 6 GHz band 2.4 GHz (and higher than the RTDmax of the 2.4 GHz frequency band) is also higher than the RTDmax of the 5 GHz band but lower than the RTDmax of the 6 GHz band, the channel to be activated can be that of the 6GHz frequency band.

[0163] The choice of channel to activate may also take into account the technology used by the connected equipment and the equipment's ability to connect to one channel or another. In certain embodiments, where several communication channels can be activated, the number of activated channels can be increased gradually (for example, in increments of one channel in terms of the number of active channels), so as not to change the quality of service offered by the access point too abruptly. Such an embodiment can help prevent excessive fluctuations in the network's quality of service from the perspective of a user of equipment using the access point in question.

[0164] If all channels are already activated, no channel-specific component necessary for its operation is a priori out of service. Therefore, in such a situation, the method may not include any action to activate such components. Alternatively, in the case of a dual-band access point, for example, no identification or verification of the currently activated channel(s) may be performed, as activation is carried out systematically (and without effect, for example, on components that are already active).

[0165] As already explained in connection with the deactivation of at least one communication channel, timers can be optionally armed in certain embodiments before (not illustrated) and / or after (element 372 [Fig.3]) an activation of a radio channel, so as to avoid fluctuating too frequently the activation state of the channels (and in particular the number of active channels).

[0166] A very short or non-existent delay before channel activation, following detection of a deterioration in the access point's quality of service (such as an RTD value exceeding the RTDmax value), will allow for better responsiveness to this deterioration and therefore better network reliability. A longer delay, on the other hand, will offer advantages in terms of energy savings and ease of processing, since a very transient disturbance in a radio frequency channel will be ignored.

[0167] A delay after channel activation can help to improve network reliability in case of temporary QoS instability.

[0168] With reference to the examples previously cited in connection with a dual-band Wi-Fi access point, the duration of the time delay after activation of a frequency band (5GHz for example) can be on the order of a few minutes (for example 1, 5, or 10 minutes)

[0169] In some embodiments (such as in the detailed embodiment), the time delays before deactivation, respectively after activation, may be greater than the time delays before activation, respectively after deactivation, so as to prioritize network reliability (in terms of quality of service) over expected energy savings.

[0170] In other embodiments (for example, when the services offered via these access points are not critical), the timeout periods before deactivation may be of the same order of magnitude, or even shorter, than the timeout periods before activation, in order to prioritize energy consumption control. The use of timeouts, as well as their respective durations, may in certain cases be defined (for example, by parameter 310), according to the needs and / or objectives of the users of the equipment likely to connect to the access point (particularly in terms of criticality and / or real-time constraints of the services running on these devices and / or energy savings).

[0171] The process can be implemented iteratively.

[0172] It is noted that the values ​​used as thresholds of the parameter representing a quality of service (for example RTDmax) as well as the time delays mentioned may vary according to the embodiments and in particular according to the equipment connected to the access point via the radio frequency channel considered or according to the services offered by the access point (and in particular the criticality of these services) via the radio frequency channel considered.

[0173] In some embodiments, the method may include obtaining descriptive information on the communication capabilities of at least one network device. For example, this information may be obtained by read access to a local or remote storage area (for example, a storage element as illustrated by elements 160, 192 of [Fig. 1]), and / or to a database provided by a manufacturer of one of the network devices, and / or by an operator providing a device 200 implementing the method of this application, and / or an application intended to run on this device 200 to implement the method of this application.

[0174] This can be achieved during an initialization 310 (including parameterization for example) of the application implementing the process of this application), and / or dynamically, after its start-up, upon detection of the presence of a new piece of equipment on the network.

[0175] Descriptive information on the communication capabilities of equipment present in the network can be used in particular when limiting the number of active communication channels or activating a communication channel.

[0176] For example, the limitation may include a selection of a communication channel to be deactivated taking into account this descriptive information, so as to deactivate a channel only if all the equipment in the network has the ability to communicate via another communication channel that is already active (i.e. activated).

[0177] Such an embodiment may, for example, allow, in the case of a dual-band Wi-Fi access point, the 2.4 GHz band to be deactivated and not the 5 GHz frequency band) when all the network equipment can operate on the 5 GHz frequency band but some equipment (for example a Wi-Fi repeater) only operates in the 5 GHz frequency band.

[0178] The value of the parameter representing a QoS is obtained (calculated, and / or measured, and / or estimated) for a communication channel from the data collected in the system 100 (via the access point(s) for example).

[0179] The data collection frequency may vary depending on the embodiment. In particular, immediately after the activation of a radio frequency channel, obtaining the value of the parameter representing a QoS for that channel may rely on data collected at least partially before the channel activation and relating to at least one other communication channel already active. Similarly, immediately after the deactivation of a first radio frequency channel, obtaining the value of the parameter representing a QoS for a second channel may rely on data collected at least partially for the first channel before its deactivation. In this case, obtaining the current value of the parameter representing a QoS for a channel may involve transposing the value of the data previously collected for another channel.

[0180] For example, in the above-cited example of a dual-band Wi-Fi access point, for which QoS is represented by RTD, after disabling the 5GHz frequency band, it may be necessary to evaluate the RTD for equipment using the 2.4GHz radio frequency channel of the access point by translating data relating to the 5GHz frequency band into data relating to the 2.4GHz frequency band.

[0181] This transposition may be optional in certain embodiments, at least for some of the data collected, or for certain activations / deactivations (example to be integrated).

[0182] Thus, in the already cited example of a dual-band Wi-Fi access point, we can, for example, assume that the RSSI measured on the 5GHz band will be close to that which would have been measured on the 2.4GHz band, and / or consider the transmission error rates and retransmissions measured in 5GHz as identical to those which would have been measured on the 2.4GHz band. Figure 3 illustrates some embodiments of method 300 of this application, where the control of at least one communication channel of at least one access point is a function of at least one current value of the parameter relating to a quality of service, the artificial intelligence model described in connection with Figure 3 being used in regression mode to predict an RTD from the collected data.

[0183] Alternatively, the control method may take into account a result provided by the inference of an artificial intelligence model whose learning was based on real values ​​of said parameter relating to a quality of service. For example, it may be the class assigned by a classifier-type artificial intelligence model, allowing a classification between a first and a second class of the collected data (data similar to those described in connection with [Fig.3] for example), this classification being learned, during the model training, based on a set of real values ​​of the RTD, the first class (“deactivation”, being associated with a value of the RTD lower than the threshold value RTDmax previously introduced, the second class (activation”) being associated with a value of the RTD higher than the threshold value RTDmax previously introduced).

[0184] In such an embodiment, the QoS test 330 of the process can be performed, not by comparing the current value of the RTD with the threshold value RTD max, but by testing the current class obtained at the output of the classifier.

[0185] The artificial intelligence model can be similar to that used according to [Fig.3], but be used, no longer in regression mode (to predict the current value of the representative parameter), but in classification mode in order to predict, from the collected data, the associated class (therefore, due to the learning carried out, in order to predict a class corresponding to a value of the representative parameter less than a threshold value or a class corresponding to a value of the representative parameter greater than a threshold value).

[0186] The learning of such a model during an experiment is detailed further by way of example.

[0187] An embodiment in which the model is used in regression mode to predict the representative parameter may, at least in some embodiments, offer certain advantages in terms of flexibility. Indeed, it may be possible to easily vary the value of the parameter used as a threshold to activate or deactivate a frequency band. There is no need to load a new model into the device 200; only a new parameterization is required. Such an embodiment also allows for different threshold values ​​of the parameter depending on the communication channels (as described above).

[0188] An embodiment in which the model is used in classification mode can contribute to achieving performance in certain environments, in terms of energy savings, better than those obtained when using the model in regression mode (as the inventors experienced).

[0189] The present method can help to limit the energy consumption of at least one access point to a communication network.

[0190] For example, certain embodiments in which the access points are dual-band (2.4 GHz and 5 GHz) access points of a Wi-Fi router in a home Wi-Fi network may result, for a typical household, in an expected outage of the 5 GHz interface for several additional hours per day (sometimes more than 12 hours) compared to existing solutions based on throughput measurement. Such an outage can represent a few Wh of savings.

[0191] On a national scale, often equipped with millions of routers, the gain on the amount of energy consumed by these routers can prove to be significant (on the order of several million Wh, for example).

[0192] Examples using separate (non-contiguous) frequency bands have been detailed above. The present application is not limited, of course, to non-contiguous frequency bands. Thus, in some embodiments, at least two of the frequency bands may be contiguous (their combination forming a non-discontinuous frequency band). 6. Experimentation

[0193] Here we present the results of an experiment aimed at evaluating the interest of the process of the present application.

[0194] For this evaluation, the system comprises three user devices connected to a dual-band Wi-Fi access point (Livebox), capable of operating in the 2.4 GHz and 5 GHz bands. The method of this application is implemented by a monitoring device directly connected via Ethernet to the access point. Data retrieval scripts run on the monitoring device to collect data useful for training the artificial intelligence model used, via WL commands issued every 5 seconds. The representative QoS parameter of the access point for each of the connected user devices is chosen for this evaluation as the RTD (Return Transmission Difference). The RTD is measured, in the example presented, using the performance measurement tool "SmokePing" ©

[0195] The training data collection and RTD measurements are performed with a wide variety of traffic combinations, to and from the user equipment, in order to obtain measurements under varying operating conditions, in terms of connection quality as well as received or transmitted power. For example, measurements were taken by varying (from 1 to 3) the number of user devices connected to the access point. The position of the user devices relative to the access point was also varied during these measurements. In addition, several measurement campaigns were carried out in environments with varying levels of interference.

[0196] The monitoring device periodically collects (every 5 seconds) data from the access point, by executing "WL" commands on the access point via a Telnet connection (commands "wl -i BAND chamin_stats", "wl -i BAND sta_info MAC_ADDRESS" and "wl -i BAND counters").

[0197] The data collected (to feed the artificial intelligence model, in particular) relates to the status of the access point's Wi-Fi interfaces and the connection quality of each user device connected to the access point. For example, the data used (directly or indirectly) for the model are, for the command wl -i BAND sta_info MAC_ADDRESS, the parameters:

[0198] idle: percentage of time in idle mode;

[0199] tx total pkts: total number of packets sent;

[0200] tx total bytes: total number of bytes sent;

[0201] tx failures: transmission failures;

[0202] rx data pkts: total packets received;

[0203] rx data bytes: total number of bytes received;

[0204] Smooth_RSSI: attenuation of wifi signal.

[0205] According to another example, the data used (directly or indirectly) for the model are, for the command wl -i BAND counters, the parameters: txframe: total of frames sent by the interface.

[0206] txbyte: total number of bytes sent through the interface.

[0207] rxframe: total frames received on the interface.

[0208] rxbyte: total number of bytes received on the interface.

[0209] (Of course, other commands and / or other parameters of these commands can be used for data collection according to the embodiments).

[0210] During data collection, the monitoring device generates independent traffic streams to and from the equipment connected to the access point. These traffic streams are defined (in terms of throughput, for example) in a configuration file (of type "y ami") and have a duration of 10 seconds. Thus, every 10 seconds, a new stream is used for each piece of equipment. Random files are transferred, at the command of the monitoring device, to and from the equipment to generate this traffic. The use of random streams makes it possible to simulate certain real-world situations.

[0211] In the detailed example, the traffic was generated following a normal distribution aggregation. The protocol used for the transfer commands could, for example, be the protocol known as SFTP (for "Secure File Transfer Protocol") according to the (English terminology) or the protocol known as SCP (for "Secure Copy Protocol" according to English terminology), depending on the equipment involved

[0212] Under the evaluation conditions, the data rate can vary from 0 Mbps to 150 Mbps, with the cumulative traffic for all connected devices not exceeding 250 Mbps when all devices generate traffic simultaneously. It is observed that most of the traffic is below 12 Mbps, which corresponds to 75% of the maximum throughput.

[0213] To measure RTD, the Smokeping © tool is run on the monitoring device. The tool sends a series of test packets across the network and deduces, from the time intervals between the sending of these packets to their recipients and their return receipt by the sender, an average RTD value over a time interval. Thus, every 2 seconds, series of 10 packets are sent (with 20 ms between each packet). These measurements also allow for the measurement of jitter.

[0214] The collected data are associated with the measured values ​​of the RTD and the PLR ​​over corresponding time periods (e.g., every 5 seconds). Other data obtained from these collected data are also associated with them (such as statistical data like moving averages of certain indicators, and / or values ​​with a time lag).

[0215] Numerous artificial intelligence models were tested, in regression or classification, during the evaluation. The quality indicator chosen (to evaluate these models) was, in the case where the model was used according to a regression algorithm, the root mean squared error (RMSE). In the case where the model was used according to a classification algorithm, the indicator chosen was the indicator known as the Fl score, which is expressed as: [Math.l] p । _ _____LE____ TP+ 4 (FP + FA)

[0216] Where:

[0217] TP represents the number of “true positives” (“True Positive” in English);

[0218] TN represents the number of “true negatives” (“True Negative” in English);

[0219] FP represents the number of “false positives” (“False Positive” in English);

[0220] FN represents the number of “false negatives” (“False Negative” in English);

[0221] The Fl score can indeed be particularly effective in evaluating the performance of a model, especially in the case of unbalanced classes.

[0222] Other indicators have been used to qualify prediction models, for example, indicators known as Receiver operating characteristics (ROC)- Area under the curve (AUC), Accuracy, Precision, Recall, Error rate, FO score. 5, score F2), where

[0223] Accuracy: represents the total proportion of correct predictions (true positives + true negatives) relative to the total number of samples.

[0224] Error: represents the total proportion of incorrect predictions (false positives + false negatives) relative to the total number of samples.

[0225] Precision (Precision): represents the proportion of true positives among the cases predicted as positive.

[0226] Recall (Recall in French): represents the proportion of true positives among the actually positive cases.

[0227] F2 Score (or Score F2): represents the harmonic mean which gives more weight to recall than to accuracy.

[0228] The Fl score already introduced represents the harmonic mean between accuracy and recall.

[0229] F0.5 Score: (or Score F0.5) represents the harmonic mean that gives more than weight to precision as to recall.

[0230] The evaluation of the artificial intelligence models was based on a comparison of the indicators selected for these algorithms using the "Kfold Cross validation" methods. For each of the tested algorithms, several optimizations of the artificial intelligence models were implemented. These optimizations focused on model weights, scaling certain types of input data (features), and determining the most relevant input data types (using AI model optimization algorithms such as Recursive Feature Elimination (RFE), Recursive Feature Addition (RFA), or methods like Select Kbest, Lasso, and / or Boruta). Internal parameter optimization of the tested models was also performed using OPTUNA© tools and / or the BayesSearchCV© method.

[0231] In the end, the LGBM and XGBoost models achieved the best results during the evaluation once optimized using the Bayesian Search © tool (with Fl scores of more than 81%) and the RFE algorithm.

[0232] A comparison of the results obtained in the evaluation environment is presented below, on the one hand by a method of controlling the activation (and deactivation) of the 5GHz radio frequency channel based on the occupied bandwidth of the communication channels (of the prior art), and on the other hand by the XGBoost_C model used as a classifier.

[0233] The evaluation dataset used corresponds to the collected data, to which are associated the bandwidths calculated for this data, the measured RTDs corresponding values, as well as a class (activation, deactivation) obtained by comparing the measured RTD to the threshold value RTDmax (here 25 ms), are used. This evaluation data is divided into two sets: a first set (training data) is used to train the artificial intelligence model used by the method described in this application, and a second set (test data) is used to verify the model's accuracy at the end of the training period. The first set is used to train the artificial intelligence model to predict a class based on the corresponding collected data.

[0234] The second batch is used on the one hand to test the relevance of the artificial intelligence model and on the other hand to compare the prior art method and the model, after its training.

[0235] According to the method based on occupied bandwidth, deactivation is implemented when the bandwidth is less than 1 Mbits and activation is implemented when the bandwidth is greater than 2 Mbits. This results in a series of activation / deactivations from the second batch of evaluation data.

[0236] Furthermore, by providing data from the first batch (respectively from the second batch) as input to the learned artificial intelligence model, we obtain a series of activations / deactivations.

[0237] For the bandwidth-based method, we then compare, on the one hand, the activation / deactivation series obtained as a function of bandwidth and, on the other hand, the activation / deactivation series obtained from the measured RTDs (considered as the true results).

[0238] For the artificial intelligence model, we compare the series of activations / deactivations obtained respectively at the output of the model for the measured RTDs and for the predicted RTDs.

[0239] The table "Table 1" below presents these results, using different performance indicators (including the Fl score and the ROC AUC indicator mentioned above).

[0240] [Tables 1] Indicator Type Indicator Value (Bandwidth-Based Method) Indicator Value (X Model GBoost_C) mdl_parameters AU AU Number of Samples 15674 15674 TP 9951 8909 TN 326 4316 FP 5213 1223 FN 184 1226 TP % 0.6349 0.5684 TN % 0.0208 0.2754 FP % 0.3326 0.0780 FN % 0.0117 0.0782 FPR 0.9411 0.2208 FNR 0.0182 0.1210 accuracy 0.6557 0.8438 Error 0.3443 0.1562 Precision 0.6562 0.8793 Recall 0.9818 0.8790 Fl score 0.7867 0.8792 F2 score 0.8932 0.8791 F0.5 score 0.7028 0.8792 ROC-AUC 0.5204 0.9142

[0241] Where:

[0242] TP__% (Percentage of True Positives): proportion of true positives relative to the total number of positive cases.

[0243] TN__% (Percentage of True Negatives): proportion of true negatives relative to the total number of negative cases.

[0244] FP__% (Percentage of False Positives): proportion of false positives relative to the total negative cases.

[0245] FN__% (False Negative Percentage): proportion of false negatives relative to the total number of positive cases.

[0246] FPR (False Positive Rate): proportion of false positives among negative cases.

[0247] FNR (False Negative Rate): proportion of false negatives among positive cases.

[0248] The results are illustrated graphically in binary classification confusion matrices (0 = deactivation (“negative”), 1 = activation (positive)) shown in [Fig.4] (confusion matrix 410 relating to the bandwidth-based method and confusion matrix 420 relating to the artificial intelligence model)

[0249] On these matrices, the number of activations (act) and deactivations (des) measured (or calculated) is found on the ordinate (tr) and the number of activations (act) and deactivations (des) predicted.

[0250] TN (True Negatives) (412, 422) here corresponds to the number of cases where the model correctly predicted the "deactivation" class.

[0251] FP (False Positives) (414, 424) here corresponds to the number of cases where the model predicted the "activation" class when it was the "deactivation" class.

[0252] FN (False Negatives) (416, 426) here corresponds to the number of cases where the model predicted the "deactivation" class when it was the "activation" class.

[0253] TP (True Positives) (418, 428) here corresponds to the number of cases where the model correctly predicted the "activation" class.

[0254] In general, a good model will have a high number of TP (418,428) and TN (412,422), and a low number of FP (414, 424) and FN (416, 426) (like the confusion matrix 420 of the model used by the process of this application).

[0255] It can be seen that, according to its 410 confusion matrix, the bandwidth-based model exhibits limited accuracy. The 410 confusion matrix highlights anomalous behavior, with a tendency to almost continuously activate the 5 GHz frequency band. Consequently, it can be concluded that this model is inefficient.

[0256] Moreover, the AUC indicator is close to 0.5, which indicates that the algorithm does not offer significantly better performance than a random choice.

[0257] On the contrary, for the XGBoost_C model, the important indicators (accuracy, recall, Fl score) are close to 88% and the AUC is greater than 91%.

[0258] It can therefore be seen that the application process, when using an XGBoost_C model, is clearly more efficient than the prior art method in the evaluation environment.

[0259] Furthermore, it is also specified that, according to the ROC and Accuracy curves corresponding to the model used by the process of this application, there is an AUC of 0.9142: This indicates that the model has an excellent ability to distinguish between positive and negative classes. (An AUC greater than 0.9 is generally considered very good).

[0260] In summary, the results show that the "XGBoost_C" model is efficient and capable of properly classifying positive and negative instances, which is particularly important in applications where classification errors can lead to significant consequences.

[0261] Table 2 below shows the percentages of 5GHz channel activation time in the case of the bandwidth-based method and in the case of the artificial intelligence model.

[0262] [Tables2] Percentage of 5GHz channel activations during evaluation. Traffic type. Bandwidth-based method. Model XGBoost_C) Good traffic conditions 95.69% 2.96% High traffic 99.63% 14.05% Very high traffic 100% 67.30%

[0263] These percentages highlight the significant energy savings that the process which is the subject of this application can provide compared to certain prior art solutions.

[0264] Thus, under favorable traffic conditions, the method of this application activates the 5 GHz channel approximately 3% of the time, whereas the bandwidth-based method activates it almost 96% of the time (i.e., 32 times more). The bandwidth-based method activates the 5 GHz Wi-Fi channel 32.3 times more often under good traffic conditions (low traffic), 7.1 times more often under high traffic, and 1.5 times more often under very high traffic, respectively, than the method of this application.

Claims

Demands

1. Method of controlling at least one first access point to at least one communication network via at least one first radio communication channel, said method comprising activating and / or deactivating at least one of said first radio communication channel as a function of at least one parameter relating to a quality of service of said at least one first access point for said first radio communication channel for at least one first communication device of said at least one communication network.

2. A control method according to claim 1 wherein the method comprises, during said activation, respectively of said deactivation, a commissioning, respectively a decommissioning, of at least a first electronic component essential to the operation of said at least a first access point via said at least a first radio communication channel.

3. A control method according to claim 1 or 2 wherein said activation, respectively said deactivation, is a function of at least one actual or estimated current value of said at least one parameter.

4. A control method according to claim 1 or 2, wherein said activation, respectively said deactivation, is a function of a result provided by an inference of an artificial intelligence model whose learning was based on real values ​​of said parameter.

5. A control method according to any one of claims 1 to 4 wherein said communication network is accessible via at least one second radio communication channel.

6. A control method according to claim 5 wherein said at least one first radio communication channel and said at least one second radio communication channel operate in disjoint radio frequency bands.

7. A control method according to any one of claims 1 to 6, wherein said parameter relating to a quality of service belongs to a group comprising: - a time (RTD) between the sending of a data packet by said first equipment, via said first communication channel, and its reception by said first access point

8.

9.

10.

11.

12. access, of said data packet;- a proportion of data packets lost or altered during a transmission, via said at least one first radio communication channel, between said first access point and said first communication equipment. - a difference between the RTDs of at least two data packets transmitted consecutively by said first access point, via said first communication channel, to said first communication equipment, - a fluctuation, for several consecutive packets, of a difference between: • a time interval between the respective times of transmission, via said first communication channel, of two consecutive data packets by said first access point and • a time interval between the respective moments of reception, via said first communication channel, by the access point, of said two consecutive packets transmitted; - an average value of the RTDs over a first time interval; - a parameter derived from at least two of the above parameters. Control method according to at least one of claims 1 to 7, wherein said activation, or said deactivation respectively, is a function at least of the current value of said parameter and wherein said current value of said parameter is a value measured and / or calculated from data measured on said network. Control method according to at least one of claims 1 to 7 wherein said activation, respectively said deactivation, is a function at least of the current value of said parameter and wherein said current value of said parameter is an estimated value. Control method according to claim 9 wherein said current value is estimated by inference from an artificial intelligence model. A control method according to at least one of claims 2 to 10 wherein said component is put into service for a period at least equal to a first operating period. A control method according to at least one of claims 2 to 11 wherein said component is taken out of service for a period at least equal to one second operating period.

13. A control method according to any one of claims 1 to 12 wherein a current value of said parameter is calculated, for said first radio communication channel of said access point, for at least one second communication device of said communication network and wherein said activation, respectively said deactivation, takes into account the current value of said parameter for said at least one second communication device of said communication network.

14. A control method according to any one of claims 1 to 13 wherein said activation, respectively said deactivation, takes into account the communication capabilities of at least one piece of equipment present on said communication network.

Images