A communication method and a communication device

Abstract

An embodiment of the present application provides a communication method and a communication device. The method can be applicable to related fields such as artificial intelligence (AI) and machine learning. The method includes: determining a first channel state; predicting, through a model, a block error rate of sending a data packet in the first channel state based on different transmission parameters, where a first data set for training the model is a data set obtained by expanding values of transmission parameters corresponding to a second data set, the second data set includes first sample data, the first sample data includes information for indicating a second channel state, a first transmission parameter corresponding to the second channel state, and feedback information, the feedback information is used to indicate whether a data packet sent based on the first transmission parameter is successfully received; and sending a first data packet based on a second transmission parameter corresponding to the lowest block error rate predicted by the model.

Description

Technical Field

[0001] This application relates to the field of communication technologies, and in particular, to a communication method and a communication device. Background Art

[0002] In a wireless network communication system, when a terminal device or a network device sends a data packet, it can select relatively appropriate transmission parameters for sending the data packet to improve the transmission quality. For example, the transmission parameters include a modulation and coding scheme (MCS) and the number of spatial division multiplexing streams. How to select relatively accurate transmission parameters has become a technical problem to be solved urgently at present. Summary of the Invention

[0003] Embodiments of this application provide a communication method and a communication device, which are used to improve the accuracy of the selected transmission parameters to reduce the block error rate of data packets.

[0004] In a first aspect, a communication method is provided. This method can be implemented by a sending device, and the sending device can be a terminal device or a network device. The method includes: determining a first channel state; predicting, through a model, the block error rate of sending a data packet in the first channel state based on different transmission parameters, where a first data set for training the model is a data set obtained by expanding the values of the transmission parameters corresponding to a second data set, the second data set includes first sample data, the first sample data includes information for indicating a second channel state, a first transmission parameter corresponding to the second channel state, and feedback information, and the feedback information is used to indicate whether a data packet sent based on the first transmission parameter is successfully received; sending a first data packet based on a second transmission parameter corresponding to the lowest block error rate predicted by the model.

[0005] In the embodiments of this application, the sending device can train a model, and the model can predict the block error rate of sending a data packet based on different transmission parameters according to the channel state. According to the block error rate predicted by the model, relatively accurate transmission parameters can be selected. In addition, the sample data included in the second data set is, for example, historical data collected, and the first data set used by the sending device to train the model is a data set obtained by expanding the values of the transmission parameters corresponding to the second data set. Therefore, the first data set can include more sample data than the second data set, and the accuracy of the model trained by the sending device based on the first data set is higher, that is, the accuracy of the prediction result of the model is higher. It can be seen that the transmission parameters selected by the sending device based on the block error rate predicted by the model are more accurate. By sending a data packet through the transmission parameters, a lower block error rate can be achieved.

[0006] In a possible design, the method further includes: dividing the second data set into a first sub-data set and a second sub-data set based on the feedback information included in the sample data in the second data set, where the feedback information included in the sample data in the first sub-data set is used to indicate successful packet reception, and the feedback information included in the sample data in the second sub-data set is used to indicate failed packet reception; performing a first operation on the sample data in the first sub-data set based on the second sub-data set to obtain a first augmented data set corresponding to the first sub-data set, and performing the first operation on the sample data in the second sub-data set based on the first sub-data set to obtain a second augmented data set corresponding to the second sub-data set; determining that the first data set includes the second data set, the first augmented data set, and the second augmented data set.

[0007] That is to say, after dividing the second data set into a first sub-data set and a second sub-data set based on the feedback information in the sample data, the first sub-data set is augmented based on the sample data in the second sub-data set, and the second sub-data set is augmented based on the sample data in the first sub-data set. It is determined that the second data set and the two augmented data sets are the first data set, so that the sample data included in the first data set is richer than the sample data included in the second data set, and the transmission parameters corresponding to the sample data are more balanced, improving the accuracy of the model's predicted block error rate.

[0008] In a possible design, performing a first operation on the sample data in the first sub-data set based on the second sub-data set includes: obtaining the information indicating the third channel state included in the second sample data, where the second sample data is the sample data in the first sub-data set; determining third sample data from the second sub-data set based on the information indicating the third channel state, where the first similarity between the information indicating the channel state included in the third sample data and the information indicating the third channel state is greater than a first value, and the third sample data includes at least one sample data; increasing the second sample data based on the number of the third sample data; and respectively replacing the transmission parameters and feedback information included in the increased second sample data with the transmission parameters and feedback information included in the third sample data.

[0009] That is to say, when the channel states are similar, the feedback information received after sending packets based on the same transmission parameters is also similar. Therefore, the reliability of the sample data obtained after augmenting the current sample data based on the sample data with similar channel states is high. And using the transmission parameters corresponding to the similar channel states as the transmission parameters corresponding to the current channel state makes the transmission parameters corresponding to the current channel state more perfect, and the transmission parameters corresponding to the sample data are more balanced, improving the accuracy of the model's predicted block error rate.

[0010] In a possible design, determining third sample data from the second subset of data based on the information indicating the third channel state includes: determining a first feature vector of the information indicating the third channel state, and determining a second feature vector of the information indicating the channel state included in the sample data in the second subset of data; determining at least one sample data corresponding to at least one second feature vector whose Euclidean distance from the first feature vector is less than a second value, where the Euclidean distance between two feature vectors being less than the second value is used to indicate that the first similarity between the two feature vectors is greater than the first value.

[0011] That is to say, by using the Euclidean distance metric method to find sample data with similar channel states, the measurement method is simple, which can improve the efficiency of finding sample data.

[0012] In a possible design, the transmission parameter includes a modulation and coding scheme (MCS). Determining third sample data from the second subset of data based on the information indicating the third channel state includes: determining a third subset of data from the second subset of data, where the third subset of data includes multiple first-type sample data, and each first-type sample data includes M sample data, and the M sample data indicate the same channel state, and the values of MCS included in the M sample data include all candidate values of MCS; determining the MCS turning point corresponding to the channel state indicated by each first-type sample data, where the MCS turning point corresponding to the channel state is used to indicate the value of MCS when the feedback information corresponding to the channel state changes from a third value to a fourth value, the third value is used to indicate that the data packet is received successfully, and the fourth value is used to indicate that the data packet fails; obtaining a first matrix according to the difference between the turning points corresponding to every two channel states in the third subset of data; determining fourth sample data among the M sample data corresponding to the channel state indicated by each first-type sample data, where the value of MCS included in the fourth sample data is the MCS turning point corresponding to the channel state indicated by each first-type sample data; determining the feature distance between every two fourth sample data based on a pre-constructed first metric matrix to obtain a second matrix; training the first metric matrix based on the first matrix and the second matrix to obtain a second metric matrix; and determining the third sample data through the second metric matrix.

[0013] That is to say, by using the monotonicity characteristic of the MCS parameter, the metric matrix trained based on the distance between the MCS turning points corresponding to each channel state and the distance between the sample data corresponding to the turning points, the selected similar channel states are closer to the current channel state, so that the expanded sample data may be more real, which helps to improve the accuracy of the model.

[0014] In a possible design, training the first metric matrix based on the first matrix and the second matrix to obtain a second metric matrix includes: determining a second similarity between the first matrix and the second matrix; if the second similarity is less than a fifth value, adjusting the first metric matrix until the second similarity is greater than or equal to the fifth value, and then outputting the second metric matrix.

[0015] That is to say, the smaller the distance between turning points, the smaller the feature distance between the sample data corresponding to the turning points. Therefore, training the pre-constructed metric matrix based on the similarity between the distance between turning points and the feature distance between the sample data corresponding to the turning points can obtain a metric matrix with high accuracy.

[0016] In a possible design, adjusting the first metric matrix includes: determining the gradient of the first metric matrix; determining an adjustment amplitude of the first metric matrix based on the gradient; and adjusting the first metric matrix based on the adjustment amplitude.

[0017] That is to say, determining the adjustment amplitude of the metric matrix each time based on the gradient of the metric matrix can accelerate the convergence speed of model training and improve the efficiency of model training.

[0018] In a possible design, the method further includes: determining at least one sample data in the first data set that includes the same channel state; determining that the at least one sample data does not include a sixth value of MCS; determining feedback information corresponding to an MCS whose value has a difference of 1 from the sixth value; and determining feedback information corresponding to the MCS that the at least one sample data does not include based on the feedback information of the MCS whose value has a difference of 1 from the sixth value.

[0019] That is to say, after expanding the second data set based on channel similarity, the MCS parameters corresponding to each channel state can also be supplemented based on the monotonicity characteristic of the MCS parameters, so that the MCS parameters corresponding to the expanded first data set are more perfect and the sample data is more balanced, which helps to improve the accuracy of model training.

[0020] In a possible design, the method further includes: determining the distribution probability of the value of a first parameter corresponding to each channel state indicated by the sample data in the first data set, where the first parameter includes MCS; performing equalization processing on the loss function of the pre-constructed model based on the distribution probability to obtain a target loss function; and training the model based on the target loss function.

[0021] That is to say, by adopting inverse propensity score (IPS) for the cost function, the contributions of different transmission parameter values to model training can be made similar, resulting in an unbiased model and improving the accuracy of model prediction.

[0022] In a second aspect, a communication method is provided. This method can be implemented by a receiving device, which can be a terminal device or a network device. The method includes: determining a first channel state; predicting, through a model, the block error rate of sending data packets based on different transmission parameters in the first channel state, where the first data set used to train the model is a data set obtained by augmenting the values of the transmission parameters corresponding to a second data set, the second data set includes first sample data, the first sample data includes information for indicating a second channel state, the first transmission parameter corresponding to the second channel state, and feedback information, and the feedback information is used to indicate whether the data packet sent based on the first transmission parameter is successfully received; sending the second transmission parameter corresponding to the lowest block error rate predicted by the model to the sending device for the sending device to send a first data packet to the receiving device based on the second transmission parameter. It can be understood that the sending device can be a terminal device or a network device.

[0023] In a possible design, the method further includes: dividing the second data set into a first sub-data set and a second sub-data set based on the feedback information included in the sample data of the second data set, where the feedback information included in the sample data of the first sub-data set is used to indicate that the data packet is successfully received, and the feedback information included in the sample data of the second sub-data set is used to indicate that the data packet is received unsuccessfully; performing a first operation on the sample data in the first sub-data set based on the second sub-data set to obtain a first augmented data set corresponding to the first sub-data set, and performing the first operation on the sample data in the second sub-data set based on the first sub-data set to obtain a second augmented data set corresponding to the second sub-data set; determining that the first data set includes the second data set, the first augmented data set, and the second augmented data set.

[0024] In a possible design, performing a first operation on the sample data in the first sub-dataset based on the second sub-dataset includes: obtaining information indicating a third channel state included in second sample data, where the second sample data is sample data in the first sub-dataset; determining third sample data from the second sub-dataset based on the information indicating the third channel state, where the information indicating the channel state included in the third sample data has a first similarity greater than a first value with the information indicating the third channel state, and the third sample data includes at least one sample data; increasing the second sample data based on the quantity of the third sample data; and respectively replacing the transmission parameters and feedback information included in the increased second sample data with the transmission parameters and feedback information included in the third sample data.

[0025] In a possible design, determining third sample data from the second sub-dataset based on the information indicating the third channel state includes: determining a first feature vector of the information indicating the third channel state, and determining a second feature vector of the information indicating the channel state included in the sample data in the second sub-dataset; determining at least one sample data corresponding to at least one second feature vector whose Euclidean distance from the first feature vector is less than a second value as the third sample data, where the Euclidean distance between two feature vectors being less than the second value is used to indicate that the first similarity between the two feature vectors is greater than the first value.

[0026] In a possible design, the transmission parameter includes a modulation and coding scheme (MCS). Determining third sample data from the second subset of data based on the information indicating the third channel state includes: determining a third subset of data from the second subset of data, where the third subset of data includes a plurality of first-type sample data, and each of the first-type sample data includes M sample data, the M sample data indicating the same channel state, and the values of the MCS included in the M sample data include all candidate values of the MCS; determining the MCS turning point corresponding to the channel state indicated by each of the first-type sample data, where the MCS turning point corresponding to the channel state is used to indicate the value of the MCS when the feedback information corresponding to the channel state changes from a third value to a fourth value, the third value being used to indicate successful packet reception, and the fourth value being used to indicate packet failure; obtaining a first matrix based on the difference between the turning points corresponding to every two channel states in the third subset of data; determining fourth sample data among the M sample data corresponding to the channel state indicated by each of the first-type sample data, where the value of the MCS included in the fourth sample data is the MCS turning point corresponding to the channel state indicated by each of the first-type sample data; obtaining a second matrix based on the characteristic distance between every two fourth sample data based on a pre-constructed first metric matrix; training the first metric matrix based on the first matrix and the second matrix to obtain a second metric matrix; and determining the third sample data through the second metric matrix.

[0027] In a possible design, training the first metric matrix based on the first matrix and the second matrix to obtain a second metric matrix includes: determining a second similarity between the first matrix and the second matrix; if the second similarity is less than a fifth value, adjusting the first metric matrix until the second similarity is greater than or equal to the fifth value, and then outputting the second metric matrix.

[0028] In a possible design, adjusting the first metric matrix includes: determining the gradient of the first metric matrix; determining the adjustment amplitude of the first metric matrix based on the gradient; and adjusting the first metric matrix based on the adjustment amplitude.

[0029] In a possible design, the method further includes: determining at least one sample data in the first subset of data that includes the same channel state; determining that the at least one sample data does not include a sixth value of the MCS; determining the feedback information corresponding to the MCS whose value has a difference of 1 from the sixth value; and determining the feedback information corresponding to the MCS that the at least one sample data does not include based on the feedback information corresponding to the MCS whose value has a difference of 1 from the sixth value.

[0030] In a possible design, the method further includes: determining the distribution probability of the value of each channel state indicated by the sample data in the first data set, where the first parameter includes MCS; equalizing the loss function of a pre-constructed model based on the distribution probability to obtain a target loss function; and training the model based on the target loss function.

[0031] In a third aspect, a communication device is provided. The communication device may be the sending device described in the first aspect or the second aspect above. The communication device has the functions of the above sending device. The communication device is, for example, a sending device, or a larger device including the sending device, or a functional module in the sending device, such as a baseband device or a chip system, etc. In an optional implementation, the communication device includes a baseband device and a radio frequency device. In another optional implementation, the communication device includes a processing unit (sometimes also referred to as a processing module) and a transceiver unit (sometimes also referred to as a transceiver module). Among them, the transceiver unit can implement the function of sending and receiving signals, and the processing unit can implement other functions except for sending and receiving signals.

[0032] In an optional implementation, the processing unit is configured to determine a first channel state; the processing unit is further configured to predict, through a model, the block error rate of sending a data packet in the first channel state based on different transmission parameters, where the first data set for training the model is a data set obtained by expanding the values of the transmission parameters corresponding to a second data set, the second data set includes first sample data, the first sample data includes information for indicating a second channel state, the first transmission parameter corresponding to the second channel state, and feedback information, and the feedback information is used to indicate whether the data packet sent based on the first transmission parameter is successfully received; the transceiver unit is configured to send a first data packet based on the second transmission parameter corresponding to the lowest block error rate predicted by the model.

[0033] In an optional implementation, the communication device further includes a storage unit (sometimes also referred to as a storage module), and the processing unit is configured to be coupled with the storage unit and execute the programs or instructions in the storage unit to enable the communication device to execute the functions of the sending device described in the first aspect or the second aspect above.

[0034] Fourthly, a communication device is provided. The communication device may be the receiving device described in the first aspect or the second aspect above. The communication device has the functions of the above receiving device. The communication device is, for example, a receiving device, or a larger device including the receiving device, or a functional module in the receiving device, such as a baseband device or a chip system, etc. In an alternative implementation, the communication device includes a baseband device and a radio frequency device. In another alternative implementation, the communication device includes a processing unit (sometimes also referred to as a processing module) and a transceiver unit (sometimes also referred to as a transceiver module). Among them, the transceiver unit can implement the function of transmitting and receiving signals, and the processing unit can implement other functions except for transmitting and receiving signals.

[0035] In an alternative embodiment, the processing unit is configured to determine a first channel state; the processing unit is further configured to predict, through a model, a block error rate of transmitting a data packet in the first channel state based on different transmission parameters, where a first data set for training the model is a data set obtained by augmenting the values of the transmission parameters corresponding to a second data set, the second data set includes first sample data, the first sample data includes information for indicating a second channel state, a first transmission parameter corresponding to the second channel state, and feedback information, and the feedback information is used to indicate whether a data packet transmitted based on the first transmission parameter is received successfully; the transceiver unit is configured to send the second transmission parameter corresponding to the lowest block error rate predicted by the model to a sending device, for the sending device to send a first data packet to the receiving device based on the second transmission parameter.

[0036] In an alternative embodiment, the communication device further includes a storage unit (sometimes also referred to as a storage module), and the processing unit is used to be coupled with the storage unit and execute programs or instructions in the storage unit to enable the communication device to execute the functions of the receiving device described in the first aspect or the second aspect above.

[0037] Fifthly, a communication device is provided. The communication device may be the sending device in the above method design, or a chip provided in the sending device. The communication device includes a communication interface and a processor. Optionally, a memory is further included. Among them, the memory is used to store a computer program, and the processor is coupled with the memory and the communication interface. When the processor reads the computer program or instruction, the communication device is enabled to execute the method provided in the first aspect above.

[0038] In a sixth aspect, a communication device is provided. The communication device can be the receiving device in the above method design or a chip provided in the receiving device. The communication device includes a communication interface and a processor. Optionally, a memory is further included. The memory is used to store a computer program. The processor is coupled to the memory and the communication interface. When the processor reads the computer program or instruction, the communication device executes the method provided in the second aspect above.

[0039] In a seventh aspect, a computer-readable storage medium is provided. The computer-readable storage medium is used to store a computer program. When the computer program runs on a computer, the computer is caused to execute the method provided in the first aspect or the second aspect above.

[0040] In an eighth aspect, a chip system is provided, including a processor and an interface. The processor is used to call and run an instruction from the interface, so that the chip system implements the method described in the first aspect above.

[0041] In a ninth aspect, a chip system is provided, including a processor and an interface. The processor is used to call and run an instruction from the interface, so that the chip system implements the method described in the second aspect above.

[0042] In a tenth aspect, a computer program product is provided, including a computer program. When the computer program runs on a computer, the computer is caused to execute the method described in the first aspect or the second aspect above.

[0043] For the beneficial effects of the second to tenth aspects above, refer to the beneficial effects of the first aspect, and no repeated description will be given. Description of the Drawings

[0044] Figure 1 It is a schematic diagram of an application scenario of an embodiment of the present application;

[0045] Figure 2 It is a flowchart of the first communication method provided by an embodiment of the present application;

[0046] Figure 3 It is a flowchart of a method for obtaining a model provided by an embodiment of the present application;

[0047] Figure 4 It is a schematic diagram of a process for collecting a second data set provided by an embodiment of the present application;

[0048] Figure 5 It is a schematic diagram of expanding a data set provided by an embodiment of the present application;

[0049] Figure 6 It is a flowchart of a method for obtaining a metric matrix provided by an embodiment of the present application;

[0050] Figure 7 A flowchart of a method for training a model provided by an embodiment of the present application;

[0051] Figure 8 A flowchart of a second communication method provided by an embodiment of the present application;

[0052] Figure 9 A schematic structural diagram of a transmitting device provided by an embodiment of the present application;

[0053] Figure 10 A schematic structural diagram of a receiving device provided by an embodiment of the present application;

[0054] Figures 11A - 11B Two schematic structural diagrams of a communication device provided by an embodiment of the present application. Detailed implementation manners

[0055] In order to make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the embodiments of the present application will be further described in detail below with reference to the accompanying drawings.

[0056] Hereinafter, some terms in the embodiments of the present application will be explained to facilitate the understanding of those skilled in the art.

[0057] 1) A terminal device, including a device that provides voice and / or data connectivity to a user. For example, it may include a handheld device with wireless connection capabilities, or a processing device connected to a wireless modem. The terminal device can communicate with a core network via a radio access network (RAN) and exchange voice and / or data with the RAN. The terminal device may include a user equipment (UE), a wireless terminal device, a mobile terminal device, a subscriber unit, a subscriber station, a mobile station, a mobile, a remote station, an access point (AP), a remote terminal device, an access terminal device, a user terminal device, a user agent, or a user device, etc. For example, it may include a mobile phone (or a "cellular" phone), a computer with a mobile terminal device, a portable, pocket-sized, handheld, computer-integrated, or vehicle-mounted mobile device, a smart wearable device, etc. For example, a personal communication service (PCS) phone, a cordless phone, a session initiation protocol (SIP) phone, a wireless local loop (WLL) station, a personal digital assistant (PDA), etc. It also includes restricted devices, such as devices with lower power consumption, or devices with limited storage capacity, or devices with limited computing power, etc. For example, it includes information sensing devices such as barcodes, radio frequency identification (RFID), sensors, global positioning system (GPS), laser scanners, etc.

[0058] By way of example and not limitation, in the embodiments of the present application, the terminal device may also be a wearable device. A wearable device, also known as a wearable intelligent device, is a general term for devices developed by applying wearable technology to the intelligent design of daily wear, such as glasses, gloves, watches, clothing, and shoes. A wearable device is a portable device that is directly worn on the body or integrated into the user's clothes or accessories. A wearable device is not just a hardware device, but also realizes powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable intelligent devices include those with complete functions and large sizes that can realize complete or partial functions without relying on a smart phone, such as smart watches or smart glasses, etc., and those that only focus on a certain type of application function and need to cooperate with other devices such as smart phones, such as various smart bracelets for vital sign monitoring, smart helmets, and smart jewelry.

[0059] 2) A network device, for example, including a base station (for example, an access point), may refer to a device in an access network that communicates with a wireless terminal device through one or more cells over an air interface. The network device can be used to mutually convert received air frames and Internet Protocol (IP) packets, and act as a router between the terminal device and the rest of the access network, where the rest of the access network may include an IP network. The network device can also coordinate the management of the attributes of the air interface. For example, the network device may include an evolved base station (NodeB or eNB or e-NodeB, evolutional Node B) in a Long Term Evolution (LTE) system or an evolved LTE system (LTE-Advanced, LTE-A), or may also include a next generation node B (gNB) in a 5G NR system. The embodiments of the present application do not limit this.

[0060] In addition, in the embodiments of the present application, the network device serves a cell, and the terminal device communicates with the network device through the transmission resources (for example, frequency domain resources, or in other words, spectrum resources) used by the cell. The cell may be a cell corresponding to the network device (for example, a base station). The cell may belong to a macro base station or a base station corresponding to a small cell. Here, the small cells may include: Metro cell, Micro cell, Pico cell, Femto cell, etc. These small cells have the characteristics of small coverage range and low transmit power, and are suitable for providing high-rate data transmission services.

[0061] In addition, multiple cells can operate on the same frequency on a carrier in an LTE system or an NR system. In some special scenarios, the concepts of the above-mentioned carrier and cell can also be considered equivalent. For example, in the carrier aggregation (CA) scenario, when a secondary carrier is configured for a UE, the carrier index of the secondary carrier and the cell identify (Cell ID) of the secondary cell operating on this secondary carrier are carried simultaneously. In this case, the concepts of carrier and cell can be considered equivalent. For example, a UE accessing a carrier is equivalent to accessing a cell.

[0062] 3) MCS table. The MCS table includes at least one of the following contents: MCS index, modulation order, target coding rate, spectral efficiency, etc. One MCS index corresponds to one modulation order, one target coding rate, and one spectral efficiency.

[0063] 4) Number of spatial multiplexing streams. In a wireless communication system, spatial multiplexing technology can be used to transmit multiple layers of data streams in parallel on the same time-frequency resource. The number of spatial multiplexing streams can be represented by rank, and the value of rank can be the number of layers of data streams transmitted in parallel on the same time-frequency resource.

[0064] 5) Channel state information (CSI). Generally speaking, CSI is further divided into periodic CSI (P-CSI), aperiodic CSI (A-CSI), and semi-persistent CSI (SPS-CSI). The meaning of periodic CSI is that the terminal device periodically sends CSI to the network device; aperiodic CSI can be triggered by the network device through downlink control information (DCI). Each time the network device triggers, the terminal device can send CSI to the network device once; semi-persistent CSI can be triggered by the network device, for example, the network device triggers through high-layer signaling. Each time the network device triggers, the terminal device can continuously send CSI to the network device for a period of time. High-layer signaling such as radio resource control (RRC) signaling, etc.

[0065] Among them, the CSI may include one or more of information such as a channel quality indicator (CQI), precoding matrix indicators (PMI), rank indicator (RI), reference signal receiving power (RSRP), channel-state information reference signal resource indicator (CRI), or an indicator of the number of non-zero wideband amplitude coefficients.

[0066] 6) Block error rate (BLER). The BLER is the percentage of blocks in error among all the blocks transmitted.

[0067] 7) The terms "system" and "network" in the embodiments of the present application may be used interchangeably. "Multiple" means two or more, and in view of this, "multiple" in the embodiments of the present application may also be understood as "at least two". "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / ", unless otherwise specified, generally represents an "or" relationship between the associated objects before and after.

[0068] In addition, unless otherwise stated, the ordinal numbers such as "first" and "second" mentioned in the embodiments of the present application are used to distinguish multiple objects and are not used to limit the order, timing, priority, or importance of multiple objects.

[0069] The embodiments of the present application can be applied to an LTE system or a 5G NR system, and can also be applied to a next-generation mobile communication system or other similar communication systems.

[0070] Please refer to Figure 1 , which is an application scenario of the embodiments of the present application. Figure 1 It includes a network device and at least one terminal device. The network device is, for example, an access network device and / or a core network device. Among them, the terminal device can communicate with the network device.

[0071] The communication method provided by the embodiments of the present application will be introduced below with reference to the accompanying drawings of the specification.

[0072] Please refer toFigure 2 , the embodiment of the present application provides a first communication method, which can be applied to a sending device. The sending device can be Figure 1 the network device shown in Figure 1 or the terminal device shown in Figure 1 . In the following introduction, it is assumed that the sending device is

[0073] the network device shown in

[0074] as an example. The method includes S201 to S206.

[0075] Optionally, the network device can determine the first channel state based on the CSI from the terminal device. For example, the network device can send a reference signal to the terminal device, and the terminal device can measure according to the reference signal to obtain a CSI report (report). In the embodiment of the present application, the CSI report is also simply referred to as CSI. The terminal device can send the CSI to the network device. The reference signal includes, for example, a channel state information-reference signal (CSI-RS) and / or a synchronization signal and a physical broadcast channel (PBCH) block (synchronization signal and PBCH block, SSB), etc. After receiving the CSI from the terminal device, the network device can determine the first channel state based on the CSI. The first channel state is, for example, a downlink channel state.

[0076] Or, the network device can determine the first channel state based on the uplink reference signal from the terminal device. For example, the terminal device can send an uplink reference signal to the network device, and the network device can measure based on the uplink reference signal and determine the first channel state according to the measurement result. The uplink reference signal includes, for example, a sounding reference signal (SRS).

[0077] The transmission parameters include, for example, parameters such as MCS and / or the number of spatial multiplexing streams. In the embodiment of the present application, the network device can predict the block error rate of sending data packets with different transmission parameters in the first channel state through a model. For example, after determining the first channel state, the network device can input different transmission parameters and the CSI corresponding to the first channel state as input information into the model, and the model can output the block error rate corresponding to different transmission parameters.

[0078] Optionally, before predicting the block error rate through the model, the network device can also train the model to obtain the model. Please refer to Figure 3 , which is an optional way for the network device to obtain the model. The method for obtaining the model includes S301 to S303.

[0079] S301: The network device collects a second data set.

[0080] Among them, the second data set includes multiple sample data, and the multiple sample data are sample data constructed by the network device based on the saved data during the operation of the system. Each sample data in the multiple sample data includes channel state information, transmission parameters configured by the network device based on the channel state information for sending data packets, and feedback information from the terminal device on whether the data packet is successfully received. It can be understood that there is a corresponding relationship between the channel state information, transmission parameters, and feedback information included in each sample data. For example, the multiple sample data includes a first sample data, and the channel state information included in the first sample data for indicating the second channel state is, for example, CSI-1. Then, the transmission parameters included in the first sample data are the transmission parameters (such as the first transmission parameter) actually used by the network device to send a data packet (such as data packet 1) to the terminal device 1 in the channel state indicated by CSI-1, and the feedback information included in the first sample data is the information on whether the terminal device successfully receives data packet 1. For example, if the terminal device successfully receives data packet 1, the feedback information included in the first sample data is ACK, and if the terminal device fails to receive data packet 1, the feedback information included in the first sample data is NACK. Among them, the transmission parameters include, for example, MCS and / or RANK, etc.

[0081] Optionally, the channel state information included in each sample data may include information for indicating the downlink channel state, and / or information for indicating the uplink channel state. The information for indicating the uplink channel state is, for example, obtained by a network device measuring the sounding reference signal (SRS) from a terminal device; the information for indicating the downlink channel state includes, for example, CSI, or does not include complete CSI, but includes one or more of parameters such as CQI, PMI, or RI. For example, when the network device sends a data packet to the terminal device through the downlink channel, when the network device selects the transmission parameters for sending the data packet based on the channel state, it may first determine the downlink channel state and select the transmission parameters based on the downlink channel state. Among them, the network device may determine the downlink channel state based on the information for indicating the downlink channel state (such as CSI, etc.) from the terminal device; or, if the uplink and downlink channels are reciprocal, the network device may also determine the uplink channel state and use the uplink channel state as the downlink channel state. For example, in a time-division duplexing (TDD) system, the uplink and downlink channels may be reciprocal.

[0082] Taking the channel state information included in each sample data as CSI as an example, the process of the network device collecting each sample data is introduced below. Please refer to Figure 4 。

[0083] S401: The network device sends a reference signal to the terminal device. Correspondingly, the terminal device receives the reference signal from the network device.

[0084] Among them, the reference signal includes, for example, CSI-RS and / or SSB, etc.

[0085] S402: The terminal device measures based on the reference signal to obtain CSI.

[0086] S403: The terminal device sends CSI to the network device. Correspondingly, the network device receives the CSI from the terminal device.

[0087] S404: The network device determines the transmission parameters for sending the data packet based on the CSI.

[0088] S405: The network device sends the data packet to the terminal device based on the transmission parameters. Correspondingly, the terminal device receives the data packet from the network device.

[0089] S406: The terminal device sends feedback information about the data packet to the network device. Correspondingly, the network device receives the feedback information from the terminal device. The feedback information may indicate whether the data packet is successfully received.

[0090] S407: The network device may construct sample data based on the CSI, the transmission parameters, and the feedback information.

[0091] After the network device executes S401 - S407 multiple times to obtain the second data set, the network device may execute S302 as follows.

[0092] S302: The network device expands the second data set to obtain the first data set.

[0093] Please refer to Figure 5 , the network device may divide the second data set obtained in S301 into two sub - data sets based on the feedback information included in each sample data. For example, the network device may divide the sample data with the feedback information of ACK (also represented by "0" hereinafter) included in the second data set into one sub - data set, called the first sub - data set, and divide the sample data with the feedback information of NACK (also represented by "1" hereinafter) into another sub - data set, called the second sub - data set. After the network device obtains these two sub - data sets, it may use one of the sub - data sets as the data set to be supplemented and the other sub - data set as the matching data set, and then expand the sample data in the data set to be supplemented to obtain the supplemented data set corresponding to the data set to be supplemented. For example, the network device may use the first sub - data set as the data set to be supplemented, the second sub - data set as the matching data set, and expand the sample data in the first sub - data set to obtain the first supplemented data set corresponding to the first sub - data set; and the network device may also use the second sub - data set as the data set to be supplemented, the first sub - data set as the matching data set, and expand the sample data in the second sub - data set to obtain the second supplemented data set corresponding to the second sub - data set.

[0094] Taking the example of a network device augmenting sample data 1 in the first sub-dataset based on the second sub-dataset, the information included in sample data 1 is as follows: CSI-2, MCS-1, and 0. The network device determines at least one sample data in the second sub-dataset whose included CSI has a similarity greater than the first value with CSI-2. The network device can increase the corresponding number of sample data 1 based on the number of at least one sample data determined from the second sub-dataset. The network device can store the increased sample data 1 in the first augmented dataset, and replace the transmission parameters and feedback information included in each increased sample data 1 in the first augmented dataset with the transmission parameters and feedback information included in at least one sample data determined by the network device from the second sub-dataset. For example, the network device determines that the CSIs included in 3 sample data in the second sub-dataset have a similarity greater than the first value with CSI-2. These 3 sample data are sample data A, sample data B, and sample data C respectively. The information included in sample data A is as follows: CSI-A, MCS-2, 1. The information included in sample data B is as follows: CSI-B, MCS-3, 1. The information included in sample data C is as follows: CSI-C, MCS-4, 1. The network device can copy 3 sample data 1, replace the transmission parameters included in the 3 copied sample data 1 with MCS-2, MCS-3, and MCS-4 respectively, and replace the feedback information in these 3 sample data 1 with 1.

[0095] Optionally, the network device can determine the similarity between CSI2 and the CSI included in each sample data in the second sub-dataset, and then determine at least one sample data whose included CSI has a similarity greater than the first value with CSI-2. To determine the similarity between two CSIs, there can be different methods, which are introduced by examples as follows.

[0096] A. Determine the similarity between two CSIs according to the Euclidean distance.

[0097] In method A, the network device can normalize the two CSIs to obtain the corresponding feature vectors respectively, and then determine the Euclidean distance between these two feature vectors. If the Euclidean distance between these two feature vectors is less than the second value, it indicates that the similarity of these two feature vectors is relatively high. For example, it indicates that the similarity of these two feature vectors is greater than the first value; or, if the Euclidean distance between these two feature vectors is greater than or equal to the second value, it indicates that the similarity of these two feature vectors is relatively low. For example, it indicates that the similarity of these two feature vectors is less than or equal to the first value.

[0098] B. Determine the similarity between two CSIs according to the metric matrix obtained by training. For example, the metric distance between two CSIs can be determined according to the metric matrix obtained by training, and the similarity between the two CSIs can be determined based on this metric distance.

[0099] Among them, the metric matrix can be obtained by training based on multiple sample data collected in S201 by using the monotonic characteristics of MCS. For the training process, please refer to Figure 6 .

[0100] S601: The network device determines the first type of sample data corresponding to N CSIs.

[0101] The network device determines the values of MCS corresponding to different CSIs from the second dataset collected in S301. If the value of MCS corresponding to a certain CSI includes all the candidate values of MCS, the network device can determine the sample data including this CSI as the first type of sample data. Taking all the candidate values of MCS as MCS = [0, 1,..., 28] as an example, if the network device determines M sample data including the same CSI (such as CSI-1) from the multiple sample data, and there are 29 sample data among the M sample data, and the values of MCS included in these 29 sample data are different, the network device determines the M sample data as the first type of sample data corresponding to CSI-1.

[0102] S602: The network device determines the MCS turning point corresponding to each first type of sample data, and determines the sample data where each MCS turning point is located.

[0103] After the network device determines the first type of sample data corresponding to N CSIs, it can use the monotonic characteristics of MCS to determine the MCS turning point corresponding to each CSI among the N CSIs, that is, when determining that the feedback information changes from 0 to 1, determine the value of MCS included in the sample data with the feedback information of 0 (or 1), and determine the sample data where each MCS turning point is located. Taking the determination of the MCS turning point corresponding to CSI-1 as an example, among the M sample data corresponding to CSI-1, the value of MCS included in sample data 1 is 9, and the included feedback information is 0. The value of MCS included in sample data 2 is 10, and the feedback information is 1. The network device can determine MCS = 9 (or MCS = 10) as the MCS turning point corresponding to CSI-1, and determine sample data 1 as the sample data where the MCS turning point is located. It can be understood that there is one MCS turning point corresponding to each CSI, and the sample data where the MCS turning point is located may include multiple. The information included in the multiple sample data is the same, because the CSIs included in the multiple sample data are the same, and the transmission parameters are also the same. Therefore, the feedback information included in the multiple sample data is also the same. The network device can use a certain sample data among the multiple sample data as the sample data where the MCS turning point is located.

[0104] S603: The network device constructs the L matrix and the D matrix.

[0105] After obtaining the MCS turning points corresponding to each CSI and the sample data where each MCS turning point is located, the network device may construct an L matrix based on the difference between the MCS turning points corresponding to every two CSIs, and construct a D matrix based on the feature distance of the sample data where the MCS turning points corresponding to every two CSIs are located.

[0106] For example, the constructed L matrix is as follows:

[0107]

[0108] l ij =(a i -b j ) 2 (Formula 2)

[0109] where n is the number of the first type of sample data, a i is the MCS turning point corresponding to the i-th CSI, and b j is the MCS turning point corresponding to the j-th CSI.

[0110] The constructed D matrix is as follows:

[0111]

[0112] d ij =dist(x i ,y i ) (Formula 4)

[0113] dist(x,y)=(x-y) T W(x-y) (Formula 5)

[0114] where W is the metric matrix, and x and y are sample feature vectors respectively. It can be seen from Formulas 3 to 5 that the value of each element in the D matrix is related to the metric matrix.

[0115] S604: The network device trains the distribution of the D matrix based on the distribution of the L matrix to obtain the metric matrix.

[0116] As introduced above, the sample data includes CSI, transmission parameters, and feedback information. Therefore, the feature distance between two sample data is positively correlated with the similarity of the CSI included in the two sample data. For example, the smaller the feature distance between two sample data, the more similar the CSI included in the two sample data. Conversely, if the CSI included in the two sample data is more similar, it indicates that the feature distance between the two sample data is smaller. Moreover, if the CSI included in the two sample data is more similar, it indicates that the values of the MCS included in the two sample data are also closer, that is, the distance between two MCS turning points is smaller. Thus, it can be concluded that if the distance between two MCS turning points is smaller, it indicates that the feature distance between the sample data where the two MCS turning points are located is smaller, that is, there is an association relationship between the distance between MCS turning points and the feature distance between the corresponding sample data where the MCS turning points are located. Therefore, optionally, the network device can train the distribution of the D matrix based on the distribution of the L matrix to obtain a metric matrix. Optionally, the network device can make the metric matrix satisfy the following relationship:

[0117]

[0118]

[0119]

[0120] Among them, l ij is related to W. To make W satisfy the conditions of formula 6, q ij can be made to be infinitely close to p ij , that is, when q ij is infinitely close to p ij , the obtained W is the target W.

[0121] Optionally, during the process of the network device training the distribution of the D matrix based on the distribution of the L matrix, the amplitude of each adjustment to the metric matrix can also be determined, for example, determining the descent gradient of the metric matrix. Optionally, the descent gradient of the metric matrix can satisfy the following relationship:

[0122]

[0123] Among them, H is an n×n matrix, and the element of H at the (i, j) position can satisfy the following relationship:

[0124]

[0125] Among them, KL is an index to measure the matching degree of two distributions. The larger KL is, the greater the distribution difference and the lower the matching degree. H i,* represents the sum of the i-th row of the H matrix, and H *,jDenotes the sum of the j-th column of the H matrix.

[0126] After determining the descent gradient of the metric matrix, the metric matrix can be adjusted based on this descent gradient, and the adjusted metric matrix can satisfy the following relationship:

[0127]

[0128]

[0129] where α is the learning rate, is the projection operator of the positive semi-definite cone, and its explicit expression is as follows:

[0130]

[0131] Optionally, when the distribution similarity between matrix D and matrix L is greater than a preset value through the adjusted metric matrix W, or when the adjustment amplitude of the metric matrix W is less than a preset amplitude, the currently obtained metric matrix W is output, and the output metric matrix is the trained metric matrix.

[0132] After the network device obtains the metric matrix W, it can determine the metric distance between two CSIs through this metric matrix W. If the metric distance between two CSIs is less than a preset distance, the network device can determine that the similarity between these two CSIs is greater than the aforementioned first value.

[0133] Optionally, the network device can also sequentially determine the similarity between CSI-2 (CSI-2 is the CSI included in the aforementioned sample data 1) and each sample data in the second subset, then sort the sample data in the second subset in descending order of similarity, determine a preset number of sample data from the second subset according to the sorting result, and add a preset number of sample data 1 to the first augmented dataset. Replace the transmission parameters and feedback information included in each sample data 1 in the first augmented dataset with the transmission parameters and feedback information included in the preset number of sample data determined from the second subset. It can be understood that the sample data determined by the network device from the second subset are the preset number of sample data ranked in the front.

[0134] Optionally, the network device can also determine at least one sample data from the second subset whose CSI-CSI-2 similarity is greater than a first value, and determine whether the number of the at least one sample data is less than or equal to a preset number. If the number of the at least one sample data is less than or equal to the preset number, the network device can increase the corresponding number of sample data 1 in the first supplementary data set based on the number of the at least one sample data, store the increased preset number of sample data 1 in the first supplementary data set, and the network device replaces the transmission parameters and feedback information included in each sample data 1 in the first supplementary data set with the transmission parameters and feedback information included in the at least one sample data. If the number of the at least one sample data is greater than the preset number, the network device can also sort the similarities between the CSI included in the at least one sample data and CSI-2 in descending order, determine a preset number of sample data from the second subset according to the sorting result, the network device can increase a preset number of sample data 1 in the first supplementary data set, and replace the transmission parameters and feedback information included in each sample data 1 in the first supplementary data set with the transmission parameters and feedback information included in the at least one sample data. For example, if the preset number is 5 and the number of sample data whose CSI-CSI-2 similarity determined by the network device from the second subset is greater than the first value is 4, the network device can increase 4 sample data 1 in the first supplementary data set, and replace the transmission parameters and feedback information included in each of the 4 sample data 1 with the transmission parameters and feedback information included in the sample data whose CSI-CSI-2 similarity determined from the second subset is greater than the first value. If the number of sample data whose CSI-CSI-2 similarity determined by the network device from the second subset is greater than the first value is 7, the network device can increase 5 sample data 1 in the first supplementary data set, and sort the 7 sample data in descending order according to the CSI-CSI-2 similarity, and the network device replaces the transmission parameters and feedback information included in the 5 sample data 1 with the transmission parameters and feedback information included in the first 5 sample data determined from the second subset.

[0135] The network device sequentially performs the above steps on each sample data in the first subset and the second subset, and then can obtain the first supplementary data set corresponding to the first subset and the second supplementary data set corresponding to the second subset. The network device can use the data set formed by the second data set, the first supplementary data set, and the second supplementary data set as the training data set for model training, that is, the first data set.

[0136] Optionally, the network device can further expand the obtained first data set by using the monotonic characteristic of MCS. The monotonic characteristic of MCS means that if the feedback information corresponding to the value of a certain MCS is 0, the feedback information corresponding to the values of other MCSs smaller than the value of this MCS is also 0; if the feedback information corresponding to the value of a certain MCS is 1, the feedback information corresponding to the values of other MCSs greater than the value of this MCS is also 1. For example, the network device can determine at least one sample data corresponding to each CSI in the first data set, determine the values of MCSs not included in the at least one sample data corresponding to each CSI, and determine the feedback information included in the sample data corresponding to the MCS whose difference from the value of this MCS is 1. According to this feedback information, determine the value of the MCS not included. For example, the value of MCS = 6 is not included in the sample data corresponding to CSI-a. The network device can determine the feedback information included in the sample data corresponding to MCS = 5 and MCS = 7 included in the sample data corresponding to CSI-a. If the feedback information included in the sample data corresponding to MCS = 5 and the feedback information included in the sample data corresponding to MCS = 7 are the same, for example, both are 0, then the network device can determine that the feedback information included in the sample data corresponding to MCS = 6 is also 0. At this time, the network device can construct a new sample data, and the information included in this new sample data is: CSI-a, MCS = 9, 0.

[0137] S303: The network device performs model training based on the first data set to obtain a model.

[0138] Please refer to Figure 7 , for the process of training the model based on the first data set in the embodiments of this application.

[0139] S701: The network device determines the distribution probability of MCS corresponding to each CSI based on the first data set.

[0140] After obtaining the first data set for training the model, the network device can regress the distribution probability of the value of MCS corresponding to each CSI according to the first data set. This distribution probability can satisfy the following relationship:

[0141] e t (X) = P(T = t|X) (Formula 14)

[0142] where X is the CSI, T is the value of MCS, and e t (X) is the distribution probability of the value of MCS corresponding to the CSI.

[0143] S702: The network device performs equalization processing on the loss function of the model based on the distribution probability.

[0144] Optionally, the model is a regression model. After the network device regresses to obtain the distribution probability of the MCS value corresponding to each CSI, the network device may perform equalization processing on the loss function of the regression model according to the distribution probability to obtain a target loss function. For example, the network device performs inverse propensity weighting processing on the loss function based on the distribution probability. Optionally, the network device may perform inverse propensity weighting processing on the loss function based on the following relationship.

[0145]

[0146] where f(*,*) is a prediction function for predicting feedback information, and y i is the feedback information included in the sample data (i.e., the sample label), δ(*,*) is the loss function, K is the number of all candidate values of MCS, and L IPS is the target loss function obtained by inverse propensity weighting.

[0147] S703: The network device performs model training based on the target loss function to obtain the model.

[0148] It can be understood that the trained model can be trained based on one transmission parameter or multiple transmission parameters, that is, the trained model can predict the block error rate corresponding to one transmission parameter or the block error rate corresponding to multiple transmission parameters. If the model is trained based on multiple transmission parameters, the transmission parameters included in the sample data collected by the network device are the multiple transmission parameters.

[0149] S203: The network device determines the transmission parameter for sending the data packet based on the predicted block error rate.

[0150] In S202, the model can predict the block error rate of the network device for sending data packets based on different transmission parameters in the first channel state. For example, at least one block error rate is predicted, and the transmission parameters corresponding to the at least one block error rate can be different. The network device can determine the transmission parameter corresponding to the minimum block error rate among the at least one block error rate (for example, the second transmission parameter), so that the second transmission parameter can be used as the transmission parameter for the data packet, which can minimize the block error rate of the data packet.

[0151] S204: The network device sends the first data packet to the terminal device based on the second transmission parameter. Correspondingly, the terminal device receives the first data packet from the network device.

[0152] S205: The terminal device sends feedback information to the network device. Correspondingly, the network device receives the feedback information from the terminal device.

[0153] Among them, the feedback information may indicate whether the first data packet is successfully received. For example, the feedback information is hybrid automatic repeat request (HARQ) information, such as an acknowledgement (ACK) or a negative acknowledgement (NACK).

[0154] S206: The network device constructs sample data pairs based on the CSI corresponding to the first channel state, the second transmission parameter, and the feedback information from the terminal device to optimize the model.

[0155] To ensure system performance, the network device cannot evenly configure the transmission parameters for sending data packets for collecting modeling data. For example, some transmission parameters may not be selected during the operation of the system (that is, the network device has not sent data packets to the terminal device using the transmission parameter), resulting in the network device being unable to collect sample data containing the unselected transmission parameter. That is, the sample data collected by the network device is unbalanced. It is difficult to guarantee the accuracy of predicting the block error rate corresponding to the unselected transmission parameter based on the model trained with the unbalanced sample data. The network device may not be able to select relatively optimized transmission parameters based on the block error rate predicted by the model, resulting in a relatively high block error rate of the data packets sent to the terminal device. Therefore, if the data set used by the network device for training the model is the second data set obtained in S301, the network device may not be able to select relatively optimized transmission parameters based on the block error rate predicted by the model. Therefore, in the embodiment of the present application, after collecting the second data set, the second data set can be expanded to obtain the first data set. The sample data in the first data set is more sufficient and more balanced than the second data set. For example, the number of sample data in the first data set is relatively larger than that in the second data set, and the first data set may include sample data corresponding to transmission parameters that have not been selected by the network device in a certain channel state. Enabling the block error rate predicted by the model trained based on the first data set to select more optimal transmission parameters helps to reduce the block error rate of the sent data packets.

[0156] Moreover, in the process of model training in the embodiment of the present application, by using inverse propensity score (IPS) for the cost function, the contributions of different values of the transmission parameter to model training can be made similar, obtaining an unbiased model and improving the accuracy of model prediction.

[0157] Please refer to Figure 8 , the embodiment of the present application provides a second communication method. This method can be applied to a receiving device, and the receiving device can be Figure 1 the network device shown, or it can also be Figure 1 the terminal device shown. In the following introduction process, it is taken that the receiving device is Figure 1 the terminal device shown as an example.

[0158] S801: The terminal device determines a first channel state.

[0159] The terminal device can receive a reference signal from the network device, perform measurements based on the reference signal and determine the first channel state according to the measurement results, or the terminal device can receive CSI from the network device to determine the first channel state. The reference signal is, for example, a reference signal sent by the network device to the terminal device in S201, and the first channel state can be, for example, a downlink channel state.

[0160] S802: The terminal device predicts, through a model, a block error rate of sending a data packet in a first channel state using different transmission parameters.

[0161] The method for the terminal device to obtain the model can refer to S202. Figure 3 The manner shown, as well as the manner in which the terminal device predicts the block error rate through a model, can refer to the manner in which the network device predicts the block error rate through a model in S202, and will not be described in detail here.

[0162] S803: The terminal device determines the transmission parameters for sending the data packet based on the block error rate predicted by the model.

[0163] The process of the terminal device determining the transmission parameter (eg, the second transmission parameter) for sending the data packet may refer to the process of the network device determining the second transmission parameter in S203.

[0164] S804: The terminal device sends the second transmission parameter to the network device. Correspondingly, the network device receives the second transmission parameter from the terminal device.

[0165] S805: The network device sends a first data packet to the terminal device based on the second transmission parameter. Correspondingly, the terminal device receives the first data packet from the network device.

[0166] S806: The terminal device sends feedback information to the network device. Correspondingly, the network device receives ACK / NACK feedback information from the terminal device.

[0167] The feedback information is similar to the feedback information in S205 and will not be described in detail here.

[0168] S807: The terminal device constructs sample data based on the CSI corresponding to the first channel state, the second transmission parameter, and the feedback information sent to the network device to optimize the model.

[0169] In the above technical solution, the transmission parameters are determined by the terminal device, and the determined transmission parameters are recommended to the network device for the network device to send data packets, which can reduce the operating burden of the network device.

[0170] It can be understood that, in the embodiments of the present application, Figure 3 the illustrated embodiments can be executed by the sending device or the receiving device. For example, when the network device sends a data packet to the terminal device, the process of determining the transmission parameters based on the channel state can be executed by the network device or the terminal device. After determining the transmission parameters, the terminal device sends the determined transmission parameters to the network device for the network device to send the data packet to the terminal device according to the received transmission parameters. Among them, if the process of determining the transmission parameters based on the channel state is executed by the network device, then Figure 3 the illustrated embodiments are executed by the network device. If the process of determining the transmission parameters based on the channel state is executed by the terminal device, then Figure 3 the illustrated embodiments are executed by the terminal device.

[0171] The following introduces the devices provided in the embodiments of the present application with reference to the accompanying drawings.

[0172] Figure 9 Fig. shows a sending device 900, which is, for example, a network device, and the sending device 900 can implement the functions of the network device involved above. The sending device 900 can be the network device described above, or can be a chip disposed in the network device described above. The sending device 900 can include a processor 901 and a transceiver 902. Among them, the processor 901 can be used to execute Figure 2 S201, S202, S203, and S206 in the illustrated embodiments, and / or other processes for supporting the technologies described herein. The transceiver 902 can be used to execute Figure 2 S204 and S205 in the illustrated embodiments, and / or other processes for supporting the technologies described herein.

[0173] For example, the processor 901 is used to determine the first channel state;

[0174] The processor 901 is further used to predict, through the model, the block error rate of sending the data packet with different transmission parameters in the first channel state;

[0175] The processor 901 is further used to determine the transmission parameters for sending the data packet based on the block error rate predicted by the model;

[0176] The processor 901 is further used to optimize the model according to the second transmission parameter and the feedback information received by the transceiver 902 to construct sample data.

[0177] Among them, all the relevant contents of each step involved in the above method embodiments can be cited in the function descriptions of the corresponding functional modules, and will not be elaborated here.

[0178] Figure 10A receiving device 1000 is shown. The receiving device 1000 is, for example, a terminal device, and the receiving device 1000 can implement the functions of the terminal device involved above. The receiving device 1000 can be the terminal device described above, or can be a chip disposed in the terminal device described above. The receiving device 1000 can include a processor 1001 and a transceiver 1002. Among them, the processor 1001 can be used to execute Figure 8 S801, S802, S803, and S807 in the embodiments shown, and / or other processes for supporting the technologies described herein. The transceiver 1002 can be used to execute Figure 8 S804, S805, and S806 in the embodiments shown, and / or other processes for supporting the technologies described herein.

[0179] For example, the processor 1001 is used to determine the first channel state;

[0180] The processor 1001 is further used to predict, through a model, the block error rate of sending data packets based on different transmission parameters in the first channel state;

[0181] The processor 1001 is further used to determine the transmission parameters of the data packet to be sent based on the block error rate predicted by the model;

[0182] The processor 1001 is further used to optimize the model by constructing sample data according to the CSI corresponding to the first channel state, the transmission parameters, and the feedback information received by the receiver 1002.

[0183] Among them, all relevant contents of each step involved in the above method embodiments can be cited in the function descriptions of the corresponding functional modules, and will not be elaborated here.

[0184] In a simple embodiment, those skilled in the art can think that the sending device 900 or the receiving device 1000 can also be implemented through the structure of a communication device 1100 as Figure 11A shown. The communication device 1100 can implement the functions of the network device or the terminal device involved above. The communication device 1100 can include a processor 1101. Among them, when the communication device 1100 is used to implement Figure 2 the functions of the network device in the embodiments shown, the processor 1101 can be used to execute Figure 2 S201, S202, S203, and S206 in the embodiments shown, and / or other processes for supporting the technologies described herein. When the communication device 1100 is used to implement Figure 8 the functions of the terminal device in the embodiments shown, the processor 1101 can be used to execute Figure 8S801, S802, S803, and S807 in the illustrated embodiments, and / or other processes for supporting the techniques described herein.

[0185] Among them, the communication device 1100 can be implemented by a field-programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on chip (SoC), a central processor unit (CPU), a network processor (NP), a digital signal processing circuit (DSP), a micro controller unit (MCU), and can also be a programmable logic device (PLD) or other integrated chips. Then, the communication device 1100 can be disposed in the network device or the terminal device of the embodiments of the present application, so that the network device or the terminal device implements the method for transmitting messages provided by the embodiments of the present application.

[0186] In an alternative implementation, the communication device 1100 may include a transceiver component for communicating with a network device or a terminal device. For example, when the communication device 1100 is used to implement Figure 2 the network device in the illustrated embodiment, the transceiver component can be used to execute Figure 2 S204 and S205 in the illustrated embodiment, and / or other processes for supporting the techniques described herein. When the communication device 1100 is used to implement Figure 8 the function of the terminal device in the illustrated embodiment, the transceiver component can be used to execute Figure 8 S804, S805, and S806 in the illustrated embodiment, and / or other processes for supporting the techniques described herein.

[0187] In an alternative implementation, the communication device 1100 may further include a memory 1102, refer to Figure 11B , where the memory 1102 is used to store computer programs or instructions, and the processor 1101 is used to decode and execute these computer programs or instructions. It should be understood that these computer programs or instructions may include the functional programs of the above network device or terminal device. When the functional program of the network device is decoded and executed by the processor 1101, the network device can implement the embodiments of the present application Figure 2The functions of the terminal device in the communication method provided by the illustrated embodiment. When the function program of the terminal device is decoded and executed by the processor 1101, the terminal device can implement the Figure 8 functions of the terminal device in the communication method provided by the illustrated embodiment.

[0188] In another alternative implementation, the function programs of these network devices or terminal devices are stored in a memory external to the communication device 1100. When the function program of the network device is decoded and executed by the processor 1101, part or all of the content of the function program of the above network device is temporarily stored in the memory 1102. When the function program of the terminal device is decoded and executed by the processor 1101, part or all of the content of the function program of the above terminal device is temporarily stored in the memory 1102.

[0189] In another alternative implementation, the function programs of these network devices or terminal devices are set in the memory 1102 stored inside the communication device 1100. When the memory 1102 inside the communication device 1100 stores the function program of the network device, the communication device 1100 can be set in the network device of the embodiment of the present application. When the memory 1102 inside the communication device 1100 stores the function program of the terminal device, the communication device 1100 can be set in the terminal device of the embodiment of the present application.

[0190] In yet another alternative implementation, part of the content of the function programs of these network devices is stored in a memory external to the communication device 1100, and other parts of the content of the function programs of these network devices are stored in the memory 1102 inside the communication device 1100. Or, part of the content of the function programs of these terminal devices is stored in a memory external to the communication device 1100, and other parts of the content of the function programs of these terminal devices are stored in the memory 1102 inside the communication device 1100.

[0191] In the embodiment of the present application, the sending device 900, the receiving device 1000, and the communication device 1100 are presented in the form of dividing each function into respective function modules, or can be presented in the form of dividing each function module in an integrated manner. Here, a "module" can refer to an ASIC, a processor and a memory that execute one or more software or firmware programs, an integrated logic circuit, and / or other devices that can provide the above functions.

[0192] Since the sending device 900, the receiving device 1000, and the communication device 1100 provided by the embodiment of the present application can be used to execute Figure 2 the illustrated embodiment or Figure 8 the communication method provided by the illustrated embodiment, the technical effects that can be obtained therefrom can refer to the above method embodiments and will not be elaborated herein.

[0193] Embodiments of the present application are described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 or multiple blocks.

[0194] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described according to the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that includes one or more integrated available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as digital versatile discs (DVDs)), or semiconductor media (such as solid state disks (SSDs)), etc.

[0195] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the embodiments of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these changes and modifications.

Claims

1. A communication method, characterized in that, Applied to a sending device, the method includes: Determine the first channel state; Predict, through a model, the block error rate of sending data packets based on different transmission parameters in the first channel state. Among them, the first data set used to train the model is a data set obtained by augmenting the values of the transmission parameters corresponding to the second data set. The second data set includes first sample data. The first sample data includes information indicating a second channel state, a first transmission parameter corresponding to the second channel state, and feedback information. The feedback information is used to indicate whether the data packet sent based on the first transmission parameter is received successfully; Send the first data packet based on the second transmission parameter corresponding to the lowest block error rate predicted by the model.

2. The method according to claim 1, wherein The method further includes: Divide the second data set into a first sub-data set and a second sub-data set based on the feedback information included in the sample data in the second data set. The feedback information included in the sample data in the first sub-data set is used to indicate that the data packet is received successfully, and the feedback information included in the sample data in the second sub-data set is used to indicate that the data packet is received unsuccessfully; Perform a first operation on the sample data in the first sub-data set based on the second sub-data set to obtain a first augmented data set corresponding to the first sub-data set, and perform the first operation on the sample data in the second sub-data set based on the first sub-data set to obtain a second augmented data set corresponding to the second sub-data set; Determine that the first data set includes the second data set, the first augmented data set, and the second augmented data set.

3. The method according to claim 2, characterized in that, Performing a first operation on the sample data in the first sub-data set based on the second sub-data set includes: Obtain the information indicating the third channel state included in the second sample data. The second sample data is the sample data in the first sub-data set; Determine third sample data from the second sub-data set based on the information indicating the third channel state. Among them, the first similarity between the information indicating the channel state included in the third sample data and the information indicating the third channel state is greater than a first value, and the third sample data includes at least one sample data; Increase the second sample data based on the number of the third sample data; Replace the transmission parameter and the feedback information included in the increased second sample data with the transmission parameter and the feedback information included in the third sample data, respectively.

4. The method according to claim 3, wherein Determining third sample data from the second sub-data set based on the information indicating the third channel state includes: Determine a first feature vector of the information indicating the third channel state, and determine a second feature vector of the information indicating the channel state included in the sample data in the second sub-data set; Determine that at least one sample data corresponding to at least one second feature vector whose Euclidean distance from the first feature vector is less than a second value is the third sample data. Among them, the Euclidean distance between two feature vectors being less than the second value is used to indicate that the first similarity between the two feature vectors is greater than the first value.

5. The method according to claim 3, characterized in that The transmission parameters include a modulation and coding scheme (MCS). Determining third sample data from the second subset of data based on the information indicating the third channel state includes: Determining a third subset of data from the second subset of data, the third subset of data including a plurality of first type sample data, each of the first type sample data including M sample data, the M sample data indicating the same channel state, and the values of MCS included in the M sample data including all candidate values of MCS; Determining the MCS turning point corresponding to the channel state indicated by each of the first type sample data, wherein the MCS turning point corresponding to the channel state is used to indicate the value of MCS when the feedback information corresponding to the channel state changes from a third value to a fourth value, the third value being used to indicate successful reception of a data packet, and the fourth value being used to indicate failure of a data packet; Obtaining a first matrix according to the difference between the turning points corresponding to every two channel states in the third subset of data; Determining fourth sample data among the M sample data corresponding to the channel state indicated by each of the first type sample data, wherein the value of MCS included in the fourth sample data is the MCS turning point corresponding to the channel state indicated by each of the first type sample data; Determining the feature distance between every two fourth sample data based on a pre-constructed first metric matrix to obtain a second matrix; Training the first metric matrix based on the first matrix and the second matrix to obtain a second metric matrix; Determining the third sample data through the second metric matrix.

6. The method according to claim 5, wherein Training the first metric matrix based on the first matrix and the second matrix to obtain a second metric matrix includes: Determining a second similarity between the first matrix and the second matrix; If the second similarity is less than a fifth value, adjusting the first metric matrix until the second similarity is greater than or equal to the fifth value, and then outputting the second metric matrix.

7. The method according to claim 6, wherein Adjusting the first metric matrix includes: Determining the gradient of the first metric matrix; Determining the adjustment amplitude of the first metric matrix based on the gradient; Adjusting the first metric matrix based on the adjustment amplitude.

8. The method according to claim 5, wherein The method further includes: Determining at least one sample data in the first subset of data that includes the same channel state; Determining that the at least one sample data does not include a sixth value of MCS; Determining the feedback information corresponding to MCS with a difference of 1 from the sixth value; Determining that the at least one sample data does not include the feedback information corresponding to MCS based on the feedback information of MCS with a difference of 1 from the sixth value.

9. The method according to any one of claims 1 to 8, characterized in that, The method further includes: Determining the distribution probability of the value of a first parameter corresponding to each channel state indicated by the sample data in the first subset of data, the first parameter including MCS; Performing equalization processing on the loss function of a pre-constructed model based on the distribution probability to obtain a target loss function; Training the model based on the target loss function.

10. A communication method, characterized in that, When applied to a receiving device, the method includes: Determining a first channel state; Predict the block error rate of sending data packets in the first channel state based on different transmission parameters through a model. Among them, the first data set for training the model is a data set obtained by augmenting the values of the transmission parameters corresponding to the second data set. The second data set includes first sample data, and the first sample data includes information for indicating a second channel state, the first transmission parameter corresponding to the second channel state, and feedback information. The feedback information is used to indicate whether the data packet sent based on the first transmission parameter is received successfully; Send the second transmission parameter corresponding to the lowest block error rate predicted by the model to the sending device, and the sending device is used to send a first data packet to the receiving device based on the second transmission parameter.

11. The method according to claim 10, wherein The method further includes: Divide the second data set into a first sub-data set and a second sub-data set based on the feedback information included in the sample data in the second data set. The feedback information included in the sample data in the first sub-data set is used to indicate that the data packet is received successfully, and the feedback information included in the sample data in the second sub-data set is used to indicate that the data packet is received unsuccessfully; Perform a first operation on the sample data in the first sub-data set based on the second sub-data set to obtain a first augmented data set corresponding to the first sub-data set, and perform the first operation on the sample data in the second sub-data set based on the first sub-data set to obtain a second augmented data set corresponding to the second sub-data set; Determine that the first data set includes the second data set, the first augmented data set, and the second augmented data set.

12. The method according to claim 11, wherein Performing a first operation on the sample data in the first sub-data set based on the second sub-data set includes: Obtain the information for indicating a third channel state included in the second sample data, and the second sample data is the sample data in the first sub-data set; Determine third sample data from the second sub-data set based on the information for indicating the third channel state. Among them, the first similarity between the information for indicating the channel state included in the third sample data and the information for indicating the third channel state is greater than a first value, and the third sample data includes at least one sample data; Increase the second sample data based on the number of the third sample data; Replace the transmission parameter and the feedback information included in the increased second sample data with the transmission parameter and the feedback information included in the third sample data respectively.

13. The method according to claim 12, wherein Determining third sample data from the second sub-data set based on the information for indicating the third channel state includes: Determine the first feature vector of the information for indicating the third channel state, and determine the second feature vector of the information for indicating the channel state included in the sample data in the second sub-data set; Determine that at least one sample data corresponding to at least one second feature vector whose Euclidean distance from the first feature vector is less than a second value is the third sample data. Among them, the Euclidean distance between two feature vectors being less than the second value is used to indicate that the first similarity between the two feature vectors is greater than the first value.

14. The method according to claim 12, wherein The transmission parameters include a modulation and coding scheme (MCS). Determining third sample data from the second subset of data based on the information indicating the third channel state includes: Determining a third subset of data from the second subset of data, the third subset of data including a plurality of first type sample data, each of the first type sample data including M sample data, the M sample data indicating the same channel state, and the values of MCS included in the M sample data including all candidate values of MCS; Determining the MCS turning point corresponding to the channel state indicated by each of the first type sample data, where the MCS turning point corresponding to the channel state is used to indicate the value of MCS when the feedback information corresponding to the channel state changes from a third value to a fourth value, the third value being used to indicate successful packet reception, and the fourth value being used to indicate packet failure; Obtaining a first matrix according to the difference between the turning points corresponding to every two channel states in the third subset of data; Determining fourth sample data among the M sample data corresponding to the channel state indicated by each of the first type sample data, where the value of MCS included in the fourth sample data is the MCS turning point corresponding to the channel state indicated by each of the first type sample data; Determining the feature distance between every two fourth sample data based on a pre-constructed first metric matrix to obtain a second matrix; Training the first metric matrix based on the first matrix and the second matrix to obtain a second metric matrix; Determining the third sample data through the second metric matrix.

15. The method according to claim 14, wherein Training the first metric matrix based on the first matrix and the second matrix to obtain a second metric matrix includes: Determining a second similarity between the first matrix and the second matrix; If the second similarity is less than a fifth value, adjusting the first metric matrix until the second similarity is greater than or equal to the fifth value, and then outputting the second metric matrix.

16. The method according to claim 15, characterized in that, Adjusting the first metric matrix includes: Determining the gradient of the first metric matrix; Determining the adjustment amplitude of the first metric matrix based on the gradient; Adjusting the first metric matrix based on the adjustment amplitude.

17. The method according to claim 14, characterized in that, The method further includes: Determining at least one sample data in the first subset of data that includes the same channel state; Determining that the at least one sample data does not include a sixth value of MCS; Determining the feedback information corresponding to MCS whose value has a difference of 1 from the sixth value; Determining that the at least one sample data does not include the feedback information corresponding to MCS based on the feedback information of MCS whose value has a difference of 1 from the sixth value.

18. The method according to any one of claims 10 to 17, characterized in that, The method further includes: Determining the distribution probability of the value of a first parameter corresponding to each channel state indicated by the sample data in the first subset of data, the first parameter including MCS; Performing equalization processing on the loss function of a pre-constructed model based on the distribution probability to obtain a target loss function; Training the model based on the target loss function.

19. A communication device, characterized in that, Includes: A processor, a memory, and one or more programs; Wherein, the one or more programs are stored in the memory, and the one or more programs include instructions that, when executed by the processor, cause the communication device to execute the method according to any one of claims 1 to 9, or cause the communication device to execute the method according to any one of claims 10 to 18.

20. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program that, when run on a computer, causes the computer to execute the method according to any one of claims 1 to 9, or causes the computer to execute the method according to any one of claims 10 to 18.

21. A chip system, characterized in that, The chip system includes: A processor and an interface, where the processor is used to call and run instructions from the interface, and when the processor executes the instructions, the method according to any one of claims 1 to 9 is implemented, or the method according to any one of claims 10 to 18 is implemented.

22. A computer program product, characterized in that, The computer program product includes a computer program that, when run on a computer, causes the computer to execute the method according to any one of claims 1 to 9, or causes the computer to execute the method according to any one of claims 10 to 18.

Images