Method for controlling the configuration of a communication system of a platform and associated systems

Abstract

Method for controlling the configuration of a platform communication system and associated systems. The present invention relates to a method for controlling a platform communication system, each communication system of the platform being configurable according to several configurations. The control method comprises the following steps: - acquisition of a plurality of environmental data, - application of a configuration function to the plurality of environmental data and parameters relating to each communication system of the platform to obtain configuration parameters defining a configuration for a communication system of the platform, and - control of a communication system of the platform so that the controlled communication system is configured according to the configuration obtained at the end of the application step. Figure for the abstract: Figure 2

Description

Title of the invention: Method for controlling the configuration of a communication system of a platform and associated systems

[0001] The present invention relates to a method for controlling a communication system of a platform. It also relates to an associated control system and a platform comprising such a control system.

[0002] In the field of communication between antennas of two platforms, it is desirable to be able to configure the link parameters as precisely as possible to ensure that the signal best suited to the communication situation is sent, particularly in terms of transmission power or antenna selection.

[0003] The link parameters involved in a communication are numerous and lead to the use of multiple different models to determine which values ​​of link parameters optimize the communication.

[0004] By way of example, one model will be used to obtain the radiation pattern of an antenna, another model to estimate the transmission power, yet another model to calculate the propagation, or a model of the performance of a wave function as a function of the signal-to-noise ratio and a target quality of service.

[0005] In addition, the models used still have imperfections related to the fact that it is difficult to take into account the variations of the environment in which the platform is used.

[0006] There is therefore a need for a method of controlling a communication system of a platform allowing communication between two platforms to be achieved which has better performance.

[0007] To this end, the description relates to a method for controlling a communication system of a platform, each communication system of the platform being configurable according to several configurations, the control method comprising the following steps:

[0008] - acquisition of a plurality of environmental data,

[0009] - application of a configuration function on the plurality of data environmental parameters and settings specific to each communication system on the platform are used to obtain configuration parameters that define a configuration for a communication system on the platform.

[0010] the configuration function having been obtained by implementing a learning process so that the communication carried out with the configuration obtained as output of the configuration function respects at least one predefined constraint, and

[0011] - control of a communication system of the platform so that the system controlled communication is configured according to the configuration obtained at the end of the application step.

[0012] According to other advantageous aspects, the control method comprises one or more of the following features, taken individually or in all technically possible combinations:

[0013] - the configuration parameters include a choice of the system of communication to be controlled when the platform includes several communication systems, the frequency of emission of a wave by the communication system to be controlled and the emission power of the communication system to be controlled and the waveform of the communication system to be controlled.

[0014] - each communication system comprises a plurality of elements, the Parameters relating to each communication system of the platform include the arrangement of elements, the losses introduced by the elements, and the performance of the elements.

[0015] - the plurality of elements of a communication system includes an antenna, a amplifier and a coupler.

[0016] - environmental data are meteorological data or data relating to an object evolving within a platform environment

[0017] - the configuration function is obtained by learning carried out on synthetic data.

[0018] - the configuration function is a neural network comprising layers of convolution, a grouping layer and linear layers.

[0019] - a predefined constraint is a constraint limiting the power emitted by each communication system.

[0020] - the process further comprises the steps of:

[0021] - obtaining at least one return data, a return data being a data representative of the quality of communication, and

[0022] - refinement of the configuration function based on the return data obtained, to obtain a refined configuration function.

[0023] - in which the steps of the process are implemented iteratively, the configuration function used at an iteration being the refined configuration function obtained at the previous iteration.

[0024] - the next iteration is implemented only if a criterion is met, the criterion being a comparison between a difference between the return data obtained and the return data obtained in the previous iteration and a predefined threshold.

[0025] - the return data is a data representative of the signal quality of communication transmitted.

[0026] The description also relates to a control system for a communication system of a platform, each communication system of the platform being configurable according to several configurations, the control system being specific to:

[0027] - acquire a plurality of environmental data,

[0028] - apply a configuration function to the plurality of data environmental parameters and settings specific to each communication system on the platform are used to obtain configuration parameters that define a configuration for a communication system on the platform.

[0029] the configuration function having been obtained by implementing a learning process so that the communication carried out with the configuration obtained as output of the configuration function respects at least one predefined constraint, and

[0030] - to control a communication system of the platform so that the system of controlled communication is configured according to the configuration obtained after applying the configuration function.

[0031] The description also relates to a platform comprising a control system as previously described.

[0032] In this description, the expression "specific to" means interchangeably "suitable for", "adapted to" or "configured for".

[0033] The invention will become clearer upon reading the following description, given solely by way of non-limiting example, and made with reference to the drawings in which: - [Fig. 1] [Fig. 1] is a schematic representation of two platforms, - [Fig.2] [Fig.2] is a flowchart of an example of the implementation of a method for controlling communication between two antennas of the platforms of [Fig. 1], and - [Fig.3] [Fig.3] is an example of a configuration function used in the control process of [Fig.2].

[0034] In [Fig.1], two platforms 10 are represented, namely a first platform 12 and a second platform 14.

[0035] Each platform 10 is typically intended for any type of land, sea and / or air environment.

[0036] Platform 10 is typically a ship.

[0037] Alternatively, the platform 10 is an aircraft such as an airplane, a helicopter, or a drone.

[0038] According to yet another variant, platform 10 is a land vehicle.

[0039] According to the example described, each platform 10 comprises a plurality of communication systems 16 comprising, as its name indicates, at least two communication systems 18.

[0040] However, in specific embodiments, it is possible to consider cases where one or each of the platforms 10 comprises a single communication system 16.

[0041] For the present case, it is assumed that the plurality of communication systems 16 includes at least one communication system forming part of the first platform 12 capable of transmitting a communication signal to the second platform 14 and one communication system forming part of the second platform 12 capable of receiving the communication signal thus transmitted.

[0042] For the remainder, without limitation, it is assumed that a transmitting communication system 20 of the first platform 12 communicates with a receiving communication system 22 of the second platform 14.

[0043] The transmitting communication system 20 and the receiving communication system 22 are suitable for communicating via a transmission channel 24.

[0044] The transmission channel 24 is formed by all the elements of the environment involved in the propagation of a signal emitted by the transmitting communication system 20 towards the receiving communication system 22.

[0045] Thus, the transmission channel 24 depends in particular on the masking elements of the environment but also on the interactions between the signal emitted and the atmosphere or the surface of the sea in the case of a platform 10 which is a ship.

[0046] The transmission channel 24 thus has a response function to an emitted signal which therefore evolves over time depending on the environment.

[0047] The performance of communication between the two platforms 12 and 14 depends heavily on the transmission channel 24.

[0048] To optimize these performances as much as possible with regard to predefined constraints, each platform 12 and 14 is equipped with a control system 30 allowing each communication system 18 to be configured.

[0049] Before detailing the operation of the control system, it should be specified that, in what is presented below, the description will focus on the control of the communication systems 18 of the first platform 12, it being understood that a similar control could be implemented for the communication systems 18 of the first platform 14.

[0050] Each communication system 18 of the first platform 10 is configurable according to several configurations.

[0051] A configuration here corresponds to a setting of each element of the communication system 18.

[0052] Indeed, each communication system 18 comprises a plurality of elements which can each operate according to a specific operating point.

[0053] For example, the communication system 18 includes an antenna, an amplifier, a coupler or an antenna control device.

[0054] Each of these elements can be adjusted to achieve a desired operation for the communication system 18.

[0055] The way in which each communication system 18 communicates is thus adjustable.

[0056] Each of these degrees of freedom corresponds to a configuration parameter.

[0057] The configuration parameters thus define a configuration for a communication system 18 of the first platform 10.

[0058] By way of example, a configuration parameter is the choice of the communication system 18 which should transmit when the first platform 12 has several communication systems 18.

[0059] Another example of a parameter is the emission frequency of a wave emitted by the communication system 18.

[0060] Yet another example, a configuration parameter is the transmission power of the communication system 18 chosen to transmit.

[0061] According to yet another example, the waveform emitted by the communication system 18 is a configuration parameter.

[0062] Another example of a configuration parameter is the type of modulation used by the communication system 18.

[0063] Yet another example of such a configuration parameter is the coding efficiency for a digital link.

[0064] Depending on the embodiment, one or more of these examples form the set of configuration parameters.

[0065] The control system 30 is suitable for controlling the configuration of the communication systems 18 by determining the appropriate value of one or more (preferably all) of the configuration parameters of a communication system 18 of the first platform 12.

[0066] The control system 30 is an electronic circuit designed to manipulate and / or transform data represented by electronic or physical quantities in computer registers and / or memories into other similar data corresponding to physical data in register memories or other types of display devices, transmission devices or storage devices.

[0067] As specific examples, the calculator 50 is implemented in the form of a programmable logic component, such as an FPGA (Field Programmable Gate Array). Array), or even an integrated circuit, such as an ASIC (Application Specifies Integrated Circuit)

[0068] Alternatively, when the method is implemented in the form of one or more software programs, that is to say, in the form of a computer program, also called a computer program product, it is further capable of being stored on a computer-readable medium, not shown. The computer-readable medium is, for example, a medium capable of storing electronic instructions and being connected to a bus of a computer system. By way of example, the readable medium is an optical disc, a magneto-optical disc, ROM, RAM, any type of non-volatile memory (for example, FLASH or NVRAM), or a magnetic card. A computer program comprising software instructions is then stored on the readable medium.

[0069] The role of the control system 30 is now described with regard to [Fig.2] which illustrates an example of the operation of the control system 30.

[0070] More specifically, [Fig.2] illustrates a flowchart of an example of implementation of a method for controlling the communication system 22 (and thus the value of the configuration parameters).

[0071] In the illustrated example, the control process corresponds to a phase of use carried out on board the first platform 12 by the control system 30.

[0072] Its implementation assumes that a learning phase has been carried out previously.

[0073] The learning phase is generally carried out offline.

[0074] The learning phase aims to determine a configuration function of the communication systems 18 of the first platform 10.

[0075] The configuration function is a function which provides configuration parameters from environmental data and parameters relating to each communication system of the first platform.

[0076] This is schematically represented in [Fig.3], in which the input data is denoted DE and the output data is denoted DSI and DS2.

[0077] By way of example, the first output data is the choice of the communication system and the second output data DS2 is the output power of the antenna of the chosen communication system.

[0078] By definition, environmental data is data relating to the environment of the observed scene, in particular meteorological data or data relating to an object evolving in the scene.

[0079] The plurality of environmental data includes, for example, a volume of precipitation, humidity outside the platform and atmospheric pressure.

[0080] However, other environmental data can be considered, such as platform altitude, refraction profile, outside temperature, including air temperature or water temperature, outside humidity, wind speed, precipitation volume, sea state or outside pressure.

[0081] The environmental data are advantageously a combination of one of the preceding examples.

[0082] The parameters relating to a communication system 18 of the first platform 12 are parameters characterizing each element of the communication system 18 which are independent of environmental data.

[0083] For example, the number of communication systems 18 or the type of antenna do not depend on the environment and structurally define the communication system 18.

[0084] More generally, the parameters relating to a communication system 18 include the arrangement of the elements, the losses introduced by the elements and the performance of the elements.

[0085] By performance in this context, it is understood that intrinsic performance, typically the effective gain of an amplifier.

[0086] The configuration function outputs a configuration so that the communication carried out with this configuration respects at least one predefined constraint.

[0087] An example of a predefined constraint is a quality of communication or a limitation of the power emitted by each communication system (discretion).

[0088] Preferably, several predefined constraints are used.

[0089] The configuration function here is a machine learning model (more commonly referred to by the corresponding English term "machine learning"), that is to say a module obtained by implementing a learning process.

[0090] The learning is supervised learning applied to training data.

[0091] This training data is most often synthetic data.

[0092] Synthetic data is, for example, generated from one or more of the following pieces of information:

[0093] - a link quality metric, typically the packet loss rate for a digital connection,

[0094] - the location of the two platforms 12 and 14,

[0095] - meteorological and sea state data (or environmental data),

[0096] - the attitude of platform 12,

[0097] - a satellite image of the area in which the two platforms 12 and 14 are operating, And

[0098] - a link budget calculated using a propagation model.

[0099] In the example described, the configuration function is a neural network as schematically illustrated by [Fig.3].

[0100] Before explaining the specific neural network used here, it is useful to recall some general information about neural networks.

[0101] A neural network comprises an ordered succession of layers of neurons, each of which takes its inputs from the outputs of the previous layer.

[0102] More precisely, each layer comprises neurons taking their inputs from the outputs of the neurons of the previous layer, or from the input variables for the first layer.

[0103] Alternatively, more complex neural network structures can be envisaged with a layer that can be linked to a layer further away than the immediately preceding layer.

[0104] Each neuron is also associated with an operation, that is to say a type of processing, to be carried out by said neuron within the corresponding processing layer.

[0105] Each layer is connected to the other layers by a plurality of synapses. A synaptic weight is associated with each synapse, and each synapse forms a link between two neurons. It is often a real number, which takes both positive and negative values. In some cases, the synaptic weight is a complex number.

[0106] Each neuron is capable of performing a weighted sum of the value(s) received from the neurons of the preceding layer, each value being multiplied by the respective synaptic weight of each synapse, or connection, between said neuron and the neurons of the preceding layer, then applying an activation function, typically a non-linear function, to said weighted sum, and delivering at the output of said neuron, in particular to the neurons of the next layer connected to it, the value resulting from the application of the activation function. The activation function introduces non-linearity into the processing performed by each neuron. The sigmoid function, the hyperbolic tangent function, and the Heaviside function are examples of activation functions.

[0107] As an optional complement, each neuron is also capable of applying, in addition, a multiplicative factor, also called bias, to the output of the activation function, and the value delivered at the output of said neuron is then the product of the bias value and the value from the activation function.

[0108] In the example of [Fig.3], the activation function comprises a succession of layers, namely 5 successive convolution layers denoted CCI to CC5, a pooling layer CP and two linear layers CL1 and CL2.

[0109] In a convolutional layer, each neuron in the same layer exhibits exactly the same connection pattern as its neighboring neurons, but at different input positions. The connection pattern is called the convolutional kernel or, more commonly, the "kernel" in reference to the corresponding English term.

[0110] The CP pooling layer or grouping layer allows the information obtained from the output of the succession of convolution layers to be compressed.

[0111] The CP pooling layer can operate in average or maximum mode.

[0112] The CP pooling layer allows output of a fixed-size vector (called space latent), suitable for processing by a series of linear layers in order to proceed with classification.

[0113] The linear layers CL1 and CL2 serve to transform the information from the previous layers of the latent space into configuration parameters.

[0114] It should be noted here that other neural network architectures can be considered.

[0115] In particular, the number of convolution layers and / or linear layers may vary.

[0116] The neural network under consideration may also include attention-based layers or state-space model layers in addition to the previous layers or in replacement of one or more of these layers.

[0117] The use phase aims, on the one hand, to exploit the configuration function obtained during the learning phase and, on the other hand, to improve the learned configuration function by taking into account the cases encountered during the use of the learned model.

[0118] The usage phase thus allows for continuous learning of the configuration function to converge towards the configuration function best suited to controlling communication between the two platforms 12 and 14.

[0119] In the example described, the use phase includes an acquisition step E40, an application step E42, a control step E44, a obtaining step E46 and a refinement step E48.

[0120] During the acquisition step E40, the control system 30 acquires a plurality of environmental data.

[0121] This environmental data generally comes from sensors on the first platform 12 with which the control system 30 is in communication, but can also come from other sources such as, for example, images of the area where the first platform is located, 12 satellite images or an environmental map.

[0122] During the application step E42, the control system 30 obtains desired configuration parameters from the determined configuration function and the plurality of acquired environmental data.

[0123] More specifically, the control system 30 applies the configuration function to the plurality of environmental data and parameters relating to each communication system of the first platform 10 to obtain the configuration parameters.

[0124] During the control step E44, the control system 30 controls a communication system 18 (the one chosen in the example according to one of the configuration parameters) of the first platform 10 so that the controlled communication system 18 is configured according to the configuration obtained at the end of the application step E42.

[0125] For example, the control system 30 is designed to send a control law to each element of the communication system 18 to be controlled so that the element operates according to the configuration parameters specific to it or in which it intervenes.

[0126] To be more concrete, either we have configuration parameters giving the values ​​of each of the settings of each element of the communication system 18 to be controlled or configuration parameters giving the characteristics of the wave to be emitted.

[0127] In the first case, a control signal to adopt the requested setting can be emitted, whereas in the second case, there is an intermediate step of converting the configuration parameter into setting values ​​of the elements.

[0128] By way of illustration, wave power can be obtained by adjusting the power supplied to the antenna or by changing the amplifier gain.

[0129] Such a conversion (configuration parameter - setting values) can be carried out at the level of the control system 30 or another control unit intended for this purpose.

[0130] During the E46 retrieval step, the control system 30 is configured to obtain return data.

[0131] The return data is a data representative of the quality of the transmitted communication signal.

[0132] During the refinement step E48, the control system 30 refines the configuration function to obtain a refined configuration function.

[0133] This refinement is carried out based on the return data obtained at the acquisition stage by retraining the network on the basis of the new data thus obtained.

[0134] Such a step is more often referred to by the corresponding English term "fine tuning".

[0135] The refinement relates to the hyperparameters of a part of the layers of the configuration function.

[0136] In particular, the last layers of the configuration function are concerned.

[0137] In the example described, the refinement thus relates to the series of linear layers dedicated to classification and / or the last convolutional layer preceding the pooling layer to effectively achieve improved configuration function performance.

[0138] According to the example described, the usage phase is implemented iteratively. The configuration function used in an iteration is the refined configuration function obtained in the previous iteration.

[0139] Advantageously, the next iteration is implemented only if a criterion is met.

[0140] For example, the criterion is a comparison between a difference between the acquired return data and the acquired return data in the previous iteration and a predefined threshold.

[0141] Another criterion is the number of communications carried out by the control system 30.

[0142] The control method thus makes it possible to adapt the power of the transmitted communication signal and the number of amplifiers according to a current state of the transmission channel 24.

[0143] The process relies in part on the quality of a configuration function that adapts to the use of the platform and improves over time.

[0144] In this sense, the control process can be seen as a process for managing communications between several communication systems by continuous learning on real data.

[0145] The configuration function obtained is more suitable than the propagation functions obtained by solving equations because no assumption is made about the transmission channel.

[0146] Furthermore, the configuration function is highly dependent on the environment in which the platform is used, an environment that cannot be easily simulated. Using data acquired during platform use overcomes this difficulty in obtaining representative data for training the configuration function.

[0147] Obtaining this configuration function also has the advantage of involving few resources.

[0148] To further increase the reliability of the configuration function, an active learning technique can be implemented.

[0149] Such a technique is more often referred to by the corresponding English term "active learning".

[0150] According to a particular example, a quality questionnaire is provided to the operator, in particular by requesting an evaluation of the analog voice that occurred during communication between the antennas.

[0151] This makes it possible to improve the configuration function either by directly using this evaluation as feedback data in the usage phase or subsequently by consulting the communication history between platforms 12 and 14.

[0152] As an alternative or in addition, it is useful to implement a control phase.

[0153] The control phase seeks to control the performance of the function of configuration.

[0154] Indeed, it is necessary to ensure that the configuration function allows for satisfactory performance and has an operation compatible with the usage constraints of platform 10.

[0155] For this purpose, it is possible to obtain an evaluation from the users of the configuration function (in the form of a score for example).

[0156] Such an achievement can be implemented periodically.

[0157] Depending on the value of the evaluation, it is determined whether the configuration function improves or not.

[0158] In the event that there is a degradation of the evaluation corresponding to a drift that is too large and cannot be regulated by an adjustment, the control phase leads to replacing the controlled configuration function with a previous configuration function presenting a better evaluation.

[0159] In such an embodiment, this implies that each configuration function obtained at an iteration of the usage phase is memorized and retained throughout the use of the configuration function.

[0160] Such a control phase thus makes it possible to better guarantee good control.

[0161] With such an optimized configuration function, it may be advantageous not to implement the learning phase for another use case.

[0162] Typically, for a different platform, instead of learning a configuration function on synthetic data specific to the platform, the configuration function resulting from the implementation of the control process on the other platform would be used.

[0163] In a more elaborate embodiment, the configuration function used would result from an aggregation of several optimized propagation functions. A federated learning mechanism can be used.

Claims

Demands

1. A method for controlling a communication system (18) of a platform (10), each communication system of the platform (10) being configurable according to several configurations, the control method comprising the following steps: - acquisition of a plurality of environmental data, - application of a configuration function on the plurality of environmental data and parameters relating to each communication system of the platform (10) to obtain configuration parameters defining a configuration for a communication system (18) of the platform (10), the configuration function having been obtained by implementing a learning process so that the communication carried out with the configuration obtained as output of the configuration function respects at least one predefined constraint,and - control of a communication system (18) of the platform (10) so that the controlled communication system (18) is configured according to the configuration obtained at the end of the application step.

2. A method according to claim 1, wherein the configuration parameters include a choice of the communication system (18) to be controlled when the platform (10) has several communication systems (18), the frequency of emission of a wave by the communication system (18) to be controlled and the emission power of the communication system (18) to be controlled and the waveform of the communication system (18) to be controlled.

3. A method according to claim 1 or 2, wherein, each communication system (18) comprises a plurality of elements, the parameters relating to each communication system (18) of the platform (10) include the arrangement of the elements, the losses introduced by the elements and the performance of the elements.

4. A method according to claim 3, wherein the plurality of elements of a communication system (18) comprises an antenna, an amplifier and a coupler.

5. A method according to any one of claims 1 to 4, wherein the environmental data are meteorological data or data relating to an object evolving in an environment of the platform (10)

6. A method according to any one of claims 1 to 5, wherein the configuration function is obtained by learning performed on synthetic data.

7. A method according to any one of claims 1 to 6, wherein the configuration function is a neural network comprising convolution layers, a grouping layer and linear layers.

8. A method according to any one of claims 1 to 7, wherein a predefined constraint is a constraint limiting the power emitted by each communication system.

9. A method according to any one of claims 1 to 8, wherein the method further comprises the steps of: - obtaining at least one return data, a return data being a data representative of the quality of the communication, and - refining the configuration function as a function of the return data obtained, to obtain a refined configuration function.

10. A method according to claim 9, wherein the steps of the method are carried out iteratively, the configuration function used at an iteration being the refined configuration function obtained at the previous iteration.

11. A method according to claim 10, wherein the next iteration is implemented only if a criterion is met, the criterion being a comparison between a difference between the return data obtained and the return data obtained in the previous iteration and a predefined threshold.

12. A method according to any one of claims 9 to 11, wherein the return data is data representative of the quality of the transmitted communication signal.

13. Control system (30) of a communication system (18) of a platform (10), each communication system of the platform (10) being configurable according to several configurations, the control system (30) being capable of: - acquiring a plurality of environmental data, - applying a configuration function on the plurality of environmental data and parameters relating to each communication system of the platform (10) to obtain configuration parameters defining a configuration for a communication system (18) of the platform (10), the configuration function having been obtained by implementing a learning process so that the communication carried out with the configuration obtained as output of the configuration function respects at least one predefined constraint, and - control a communication system (18) of the platform (10) so that the controlled communication system (18) is configured according to the configuration obtained after application of the configuration function.

14. Platform comprising a control system according to claim 13.