A channel frequency response compression method, device and medium

Abstract

The application provides a channel frequency response compression method, device and medium, and relates to the field of communication. The method comprises the following steps: obtaining an observation result obtained by observing a wireless channel between a transmitter and a receiver; the observation result comprises a plurality of CFR powers corresponding to a plurality of observation points; for each CFR power, using a fitting algorithm to perform curve fitting on the CFR power according to an initial trigonometric function quantity Q corresponding to the CFR power, to obtain an initial fitting result, the initial fitting result comprises a CFR fitting result corresponding to each CFR power, the CFR fitting result comprises Q trigonometric functions; Q is an integer greater than or equal to 1; performing iterative fitting according to a residual error between the observation result and the initial fitting result to obtain a target fitting result with a residual error less than a preset residual error threshold; and taking the trigonometric functions included in the target fitting result as a compression result of the observation result. The application uses trigonometric functions as base functions to compress the CFR, and can improve the compression ratio of the CFR compression.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of communication, and in particular to a channel frequency response compression method, device and medium. BACKGROUND

[0002] With the rapid development of Internet of Things technology, wireless sensing has become one of the emerging sensing technologies today. Compared with traditional sensing technology, wireless sensing can achieve passive positioning, behavior recognition and other functions of the target by analyzing the fluctuation of wireless signals without carrying any equipment, and can be widely applied in smart factories, smart homes and other fields. In the process of wireless sensing, the transmission of channel frequency response (CFR) and other related information is essential.

[0003] The existing compression of CFR is mainly based on the frequency domain sparsity of the signal and the spatial correlation generated by the antenna array as the theoretical basis, and by using two-dimensional discrete cosine transform and Karhunen-Loeve transform to construct a sparse basis, the discrete values of the signal are obtained by random sampling in a manner far less than the Nyquist sampling rate, and the sparse basis is used to fit the discrete values, thereby realizing the compression of CFR. However, this compression method has insufficient compression ratio, and the compressed CFR cannot meet the demand of large-scale transmission of CFR. SUMMARY

[0004] To solve the problem of insufficient compression ratio of channel frequency response in the prior art, the present application provides a channel frequency response compression method, device and medium, which can improve the compression ratio of CFR and reduce the loss in transmission and storage of CFR.

[0005] To achieve the above purpose, the present application adopts the following technical solutions:

[0006] In a first aspect, the present application provides a channel frequency response compression method, which comprises:

[0007] Obtaining an observation result obtained by observing a wireless channel between a transmitter and a receiver; the observation result comprises a plurality of channel frequency response (CFR) powers corresponding to a plurality of observation points;

[0008] For each CFR power, using a fitting algorithm to perform curve fitting on the CFR power according to the number Q of initialization trigonometric functions corresponding to the CFR power, to obtain an initialization fitting result, the initialization fitting result comprises a CFR fitting result corresponding to each CFR power, and the CFR fitting result comprises Q trigonometric functions; Q is an integer greater than or equal to 1;

[0009] According to the residual between the observation result and the initial fitting result, iterative fitting is performed to obtain a target fitting result with a residual less than a preset residual threshold, and the trigonometric function included in the target fitting result is taken as the compression result of the observation result.

[0010] As a possible implementation of the first aspect of the application, according to the residual between the observation result and the initial fitting result, iterative fitting is performed to obtain a target fitting result with a residual less than a preset residual threshold, including:

[0011] The iterative fitting process is performed: it is judged whether the residual between the observation result and the fitting result corresponding to the current iterative fitting process is less than or equal to a preset residual threshold; wherein the fitting result corresponding to the first round of iterative fitting process is the initial fitting result;

[0012] If the residual is less than the preset residual threshold, the fitting result corresponding to the previous iterative fitting process is taken as the target fitting result;

[0013] If the residual is greater than the preset residual threshold, Q is increased, and the CFR power is curve fitted using a fitting algorithm to obtain a fitting result of the next round; based on the obtained fitting result of the next round, the iterative fitting process is repeatedly performed until a target fitting result with a residual less than the preset residual threshold is obtained.

[0014] As a possible implementation of the first aspect of the application, the curve fitting of the CFR power using the fitting algorithm includes:

[0015] The trigonometric function set is initialized according to the Q, and the residual between the trigonometric function set and the observation result is taken as an objective function, and the CFR power is curve fitted using a fitting algorithm based on the objective function; the trigonometric function set includes Q trigonometric functions

[0016] As a possible implementation of the first aspect of the application, the objective function is:

[0017]

[0018] Wherein, E(A) is the residual between the trigonometric function set and the observation result, |H(f k ,t)| 2 is the CFR power, H(f k ,t) is the observation result, Γ(x k ,A) is the trigonometric function set, A is the trigonometric function parameter, t is the time, and k is a positive integer greater than or equal to 1.

[0019] As a possible implementation of the first aspect of the application, the fitting algorithm includes: LM algorithm, Gauss-Newton method, gradient descent method.

[0020] As a possible implementation manner of the first aspect, the number Q of initial trigonometric functions corresponding to the CFR power is determined in the following manner.

[0021] For each CFR power, the CFR power is converted from the time domain to the frequency domain to obtain a spectrum of the CFR power, and the number Q of initial trigonometric functions corresponding to the CFR power is determined according to the number of spectral peaks in the spectrum.

[0022] In the second aspect, the application provides a channel frequency response compression device, which comprises: a communication unit configured to obtain an observation result obtained by observing a wireless channel between a transmitter and a receiver; the observation result comprises a plurality of channel frequency response (CFR) powers corresponding to a plurality of observation points;

[0023] a processing unit configured to, for each CFR power, perform curve fitting on the CFR power using a fitting algorithm according to the number Q of initial trigonometric functions corresponding to the CFR power to obtain an initial fitting result; the fitting algorithm corresponds to an initial fitting order of Q; the initial fitting result comprises a CFR fitting result corresponding to each CFR power, and the CFR fitting result comprises Q trigonometric functions; Q is an integer greater than or equal to 1;

[0024] The processing unit is further configured to perform iterative fitting according to a residual between the observation result and the initial fitting result to obtain a target fitting result with a residual less than a preset residual threshold, and use the trigonometric functions included in the target fitting result as a compression result of the observation result.

[0025] In the third aspect, the application provides a channel frequency response compression device, which comprises a processor and a communication interface; the communication interface and the processor are coupled, and the processor is configured to run a computer program or instructions to implement a channel frequency response compression method as described in the first aspect and any possible implementation manner of the first aspect.

[0026] In the fourth aspect, the application provides a computer readable storage medium, which stores instructions, and when the instructions are run on a terminal, the terminal performs a channel frequency response compression method as described in the first aspect and any possible implementation manner of the first aspect.

[0027] In the fifth aspect, the application provides a computer program product comprising instructions, and when the computer program product is run on a channel frequency response compression device, the channel frequency response compression device performs a channel frequency response compression method as described in the first aspect and any possible implementation manner of the first aspect.

[0028] In a sixth aspect, an embodiment of the present application provides a chip, which comprises a processor and a communication interface, the communication interface being coupled to the processor, and the processor being configured to run computer programs or instructions to implement the channel frequency response compression method described in the first aspect and any possible implementation manner of the first aspect.

[0029] Specifically, the chip provided in the embodiments of the present application further comprises a memory configured to store the computer programs or instructions.

[0030] The technical solutions provided in the present application have at least the following beneficial effects: the present application uses trigonometric functions as base functions to compress CFR, which can improve the compression ratio of CFR compression; using compressed CFR for transmission and storage can save the resources occupied by CFR in transmission and storage. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 A wireless sensing system schematic diagram provided for the embodiments of the present application;

[0032] Figure 2 A communication system schematic diagram provided for the embodiments of the present application;

[0033] Figure 3 A channel frequency response compression method flowchart provided for the embodiments of the present application;

[0034] Figure 4 Another channel frequency response compression method flowchart provided for the embodiments of the present application;

[0035] Figure 5 A compression result comparison diagram provided for the embodiments of the present application;

[0036] Figure 6 A compression result schematic diagram provided for the embodiments of the present application;

[0037] Figure 7 A structure schematic diagram of a channel frequency response compression device provided for the embodiments of the present application;

[0038] Figure 8 Another structure schematic diagram of a channel frequency response compression device provided for the embodiments of the present application;

[0039] Figure 9 A structure schematic diagram of a chip provided for the embodiments of the present application. DETAILED DESCRIPTION

[0040] The channel frequency response compression method and device provided in the embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0041] The term "and / or" in this document merely describes an associated relationship, which means that there can be three relationships, for example, A and / or B can mean that A exists alone, A and B exist together, and B exists alone.

[0042] The terms "first" and "second" and the like in the description of the present application and in the claims of the present application are used for distinguishing between similar objects, or for distinguishing between the same object for different purposes, and are not necessarily used to describe a specific order or sequence.

[0043] In addition, the terms "comprise", "have" and any variations thereof in the description of the present application are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that comprises a list of steps or units is not limited to the listed steps or units, but can optionally further include other steps or units not listed or can optionally further include other steps or units inherent to such processes, methods, products or devices.

[0044] It should be noted that in the embodiments of the present application, the words "exemplary" or "for example" are used to mean serving as an example, instance, or illustration. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of the present application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Rather, the use of the words "exemplary" or "for example" is intended to present concepts in a concrete manner.

[0045] In the following, some related terms and technologies involved in the embodiments of the present application are explained and described.

[0046] 1) Channel state information (CSI)

[0047] In the field of wireless communication, channel state information is the channel property of a communication link. It describes the attenuation factor of a signal on each transmission path, i.e. the value of each element in the channel gain matrix H, such as signal scattering (Scattering), environmental attenuation (fading, multipath fading or shadowing fading), distance attenuation (power decay of distance) and the like.

[0048] Channel state information can describe the multipath effect of a channel with channel impulse response (CIR). The multipath propagation of a signal is considered to be time delay spread in time domain, and causes selective fading of the signal in frequency domain. Therefore, the channel state information can also be represented from amplitude-frequency characteristics and phase-frequency characteristics with channel frequency response (CFR) of the wireless channel. The frequency response of the channel can be represented by the following formula. The CFR can be understood as the response of the signal characteristics in different frequency ranges, generally including amplitude / frequency and phase / frequency responses. The CIR can be understood as the signal energy value of the signal arriving at the receiving end after passing through different times (different propagation paths cause different propagation times). Under the condition of infinite bandwidth, the CFR and the CIR are Fourier transforms of each other.

[0049] CSI is generally used to reflect the current status of the wireless channel. In the wireless fidelity (WiFi) protocol, a CSI matrix corresponding to each orthogonal frequency division multiplexing (OFDM) subcarrier group is obtained by measuring the OFDM subcarrier group. The number of rows of the CSI matrix is the number of transmitting antennas, and the number of columns of the CSI matrix is the number of receiving antennas. Each element of the CSI matrix is a complex number including a real part and an imaginary part. Therefore, in the case of a large number of antennas and a large number of subcarriers, for example, the number of subcarriers is 114, and the number of transmitting antennas and the number of receiving antennas are both 4. The CSI of each subcarrier is a matrix, and the number of rows and the number of columns of the matrix are both 4. Each element in the matrix is a complex number. The real part and the imaginary part of the complex number are represented by 8 bits, respectively. Thus, if the CSI on 114 subcarriers is transmitted, 3648 bytes are used for transmission. Even if the CSI on 114 subcarriers is processed by the compression algorithm in the institute of electrical and electronics engineers (IEEE) 802.11ac, the data amount of the processed data is in the order of thousands of bytes, and a large amount of transmission resources is occupied.

[0050] 2) Data compression

[0051] Data compression, also known simply as compression, refers to reducing the amount of data in order to reduce storage space, improve transmission, storage and processing efficiency, or reorganizing data according to a certain algorithm to reduce data redundancy and storage space, under the premise of not losing useful information. Data compression includes lossy compression and lossless compression. Data compression or source coding is a process of representing information with fewer data elements than the uncoded, according to a specific coding mechanism. For any form of communication, compression data communication can only work properly when the sender and receiver of the information can understand the coding mechanism. Data compression can be achieved because most real-world data has statistical redundancy. Compression is important because it can help reduce the consumption of expensive resources such as hard disk space / network bandwidth, and compression also consumes information processing resources, which are also expensive resources. Therefore, the design of data compression mechanism needs to be a compromise between compression ability (compression speed, compressed data size), distortion degree (quality loss), required computing resources, and other factors to be considered.

[0052] 3) Wireless sensing

[0053] Wireless sensing is a technology that uses the signal reflected by radio waves on the target to be detected (such as the human body) to perceive the action of the target to be detected. Wireless sensing technology combines sensing and communication into one, with three distinctive features - "three no": (1) sensorless, the perception of people and the environment no longer needs to deploy special sensors, which is different from the wireless sensor network where sensors are responsible for sensing and wireless signals are responsible for communication; (2) wireless, there is no need to deploy wired lines for communication and sensors; (3) contactless, compared to various wearable smart devices on the market, wireless sensing takes a further step, without the need for users to wear any devices.

[0054] The sensing objects of wireless sensing technology include environment, articles and people, and the potential applications are very rich. Taking sensing people as an example, wireless sensing technology can be used for passive people sensing. “Passive” here refers to that people do not need to carry any electronic devices, in order to distinguish from the traditional wireless positioning system, which locates people by locating the electronic devices carried by people. Such a way is also called device-free or non-invasive. Passive people detection can be widely used in various ubiquitous computing applications, providing better user location-based services. For example, automatically playing product descriptions when visitors approach a product in a museum, counting the most popular products in a supermarket, or counting the number of passengers in an elevator and car, etc. Non-sensor sensing can also be used as a new type of human-computer interaction, which remotely controls electronic devices (computers, game consoles, smart hardware, etc.) by identifying the behavior (posture, action and gesture, etc. small movements) of people to complete specific functions or provide interactive motion games; It can also be used for intelligent medical monitoring to detect the sleep quality of people and the accidental falls of the elderly, etc. The mode of passive sensing also meets the needs of security and protection applications. In security-related applications such as monitoring in classified areas, personnel intrusion detection, disaster emergency response, and protection of important articles, it is necessary to timely detect whether personnel (staff or intruders) who do not carry any wireless communication devices appear in sensitive areas and monitor them.

[0055] Reference Figure 1 The wireless sensing system includes a transmitter 101 and a receiver 102. In actual application, the transmitter 101 can be one or multiple. The receiver 102 can be one or multiple. Figure 1 Only one transmitter and one receiver are shown in the figure. The transmitter 101 and the receiver 102 can be separate physical devices, or can be arranged in the same physical device. The wireless signal received by the receiver 102 includes a direct signal 104 and a reflected signal 105 reflected by the target to be detected 103. When the target to be detected 103 moves, the reflected signal 105 also changes. Correspondingly, the superimposed wireless signal received by the receiver 102 also changes. At this time, the receiver 102 detects that the wireless channel has changed. Usually, the change of the wireless channel is quantified as the change of the CSI in the communication protocol, which is specifically embodied as the change of the amplitude of the CSI and / or the change of the phase of the CSI. That is, the receiver 102 senses whether there is a target to be detected around or the motion status of the target to be detected based on the determined CSI. Therefore, wireless passive sensing technology can be widely applied to wireless sensing applications such as intrusion detection, old people care, gesture recognition, respiratory sleep monitoring, indoor people counting, etc.

[0056] The channel frequency response compression method provided by the embodiment of the present application can be applied toFigure 2 The communication network shown can be a 5th generation (5G) mobile communication network, and can also be a 4th generation (4G) (such as an evolved packet system (EPS) mobile communication network, and can also be a global system of mobile communication (GSM) system, a code division multiple access (CDMA) system, a wideband code division multiple access (WCDMA) system, a general packet radio service (GPRS), a long term evolution (LTE) system, an LTE frequency division duplex (FDD) system, an LTE time division duplex (TDD), a universal mobile telecommunications system (UMTS), a worldwide interoperability for microwave access (WiMAX), a wireless local area network (WLAN) system or a WiFi system, etc., and the present application is not limited thereto.

[0057] Referring to Figure 2 , Figure 2 A communication system applicable to an embodiment of the present application is shown in the figure, which can include one or more network devices 201 (only one is shown) and one or more terminals 200 connected to each network device 201. It can be understood that, Figure 2 This is only a schematic diagram and does not constitute a limitation on the applicable scenarios of the technical solutions provided by the present application.

[0058] The network device 201 can be a transmission reception point (TRP), a base station, a relay station, or an access point, etc. The network device 201 can be a network device in a 5G communication system or a network device in a future evolved network, such as a 5G macro base station or a 5G micro base station; it can also be a wearable device or a vehicle-mounted device, etc. The terminal 200 can be a user equipment (UE), an access terminal, a UE unit, a UE station, a mobile station, a mobile station, a remote station, a remote terminal, a mobile device, a UE terminal, a wireless communication device, a UE agent, or a UE apparatus, etc. The access terminal can be a cellular phone, a cordless phone, etc.

[0059] It can be understood that, in the above communication system, when wireless sensing is performed, either the network device 201 or the terminal 200 can act as the receiver 102 and the other as the transmitter 101 to perform wireless sensing. For example, when the terminal 200 performs downlink communication, the network device 201 acts as the transmitter 101 and the terminal 200 acts as the receiver 102; when the terminal 200 performs uplink communication, the network device 201 acts as the receiver 102 and the terminal 200 acts as the transmitter 101.

[0060] For example, a typical application scenario of an embodiment of the present application is a 5G microcell station environment in an indoor scenario. In the 5G microcell station in the indoor scenario, the position and behavior of the target in the scene are sensed through the transmission and reception of signals. The transmitter transmits a wireless signal, and the receiver estimates the channel frequency response through the trained preamble symbol after receiving the signal. Considering the multipath effect in the indoor scenario, it is assumed that the signal propagates through L paths from the transmitter to the receiver, and at this time, CRH(f, t) can be represented as:

[0061]

[0062] where ε1, ε2, and ε3 can be understood as the time offset caused by the symbol timing offset, the sampling frequency offset, and the center frequency offset introduced by the non-strict synchronization of the transmission and reception ends, f is the signal frequency point, a l (f, t) is the complex attenuation and initial phase of the lth propagation path, is the propagation distance d l (t) causes the corresponding cumulative phase, and c is the propagation speed of the signal in the air. As can be seen from equation (1), the propagation path length d lChanges in (t) will cause changes in the propagation delay, resulting in phase accumulation. In indoor environments, the length of the reflection paths caused by static objects such as walls, furniture, etc. can be approximated as constant (here, higher order reflection signals can be neglected because their energy is relatively weak), while the length of the reflection paths caused by the human target will change over time due to the movement, behavior, etc. of the human body. Therefore, by dividing the propagation paths into static paths and dynamic paths, H(f, t) can be re-expressed as:

[0063] H(f, t) = exp(-j2πε) x (H s (f, t) + H d (f, t)) Equation (2)

[0064] where ε = f(ε1+ε2)+ε3, H s (f, t) is the sum of static reflection paths, and H d (f, t) is the sum of dynamic reflection paths. H a (f, t) can be expressed as:

[0065]

[0066] P d is a set of dynamic reflection paths, is the attenuation of the dynamic path. Without loss of generality, it is assumed that the path length of the lth path caused by the target changes at a constant speed v k over a short period of time, Equation (3) can be further expressed as:

[0067]

[0068] where is the initial path length of the reflection path caused by the target. Obviously, CFR is related to v k , and existing technology research has shown that v k is directly related to human activity, so the accumulation of the phase is directly related to the behavior of the target, but it is difficult to directly calculate v k from the phase of CFR due to the phase error ε = f(ε1+ε2)+ε3. However, in addition to phase accumulation, v k is also related to power changes, and the power of CFR can be expressed as:

[0069] |H(f, t)| 2 = H D + H S + H DC Equation (5)

[0070] where the cross term is:

[0071]

[0072] The direct current term is:

[0073]

[0074] The direct current term is:

[0075]

[0076] wherein, and is the initial phase.

[0077] It can be seen that, with the change of the dynamic path length, the phase accumulation is also constantly changing, thereby causing the overall power of the CFR to change. Based on this, it can be concluded that the behavior of the target is directly related to the change of the CFR power, and the CFR power is also expressed as the sum of multiple trigonometric functions, so the compression of the CFR power can be realized by using trigonometric function fitting.

[0078] And using the compressed CFR power does not lose the information of the change of the propagation path caused by the target, and using the compressed CFR power for wireless sensing can still obtain accurate wireless sensing results. It can be understood that the above trigonometric functions can include: sine function (sine, sin) and cosine function (cosine, cos), which can be converted with each other (the phase difference ).

[0079] The communication system and service scenarios described in the embodiments of the present application are used to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application.

[0080] With the rapid development of Internet of Things technology, wireless sensing has become one of the emerging sensing technologies today. Compared with traditional sensing technology, wireless sensing can achieve passive positioning, behavior recognition and other functions of the target by analyzing wireless signal fluctuations without the target carrying any device, and can be widely applied in smart factories, smart homes and other fields. In recent years, with the large-scale deployment of 5th Generation Mobile Communication Technology (5G) networks, 5G small stations will also be gradually widely applied; it is predicted that by 2025, 78% of small cell base stations will be used in dense environments to fill coverage or capacity gaps. Therefore, the market prospect of wireless sensing based on 5G small stations is broad. In the sensing process, the transmission of CFR and other related information is essential.

[0081] The existing channel state information (CSI) compression is mainly realized by compressive sensing. For compression of channel frequency response (CFR) in the CSI, the frequency domain sparsity of a signal and the spatial correlation generated by an antenna array are mainly used as the theoretical basis, a sparse basis is constructed by using two-dimensional discrete cosine transform and Karhunen-Loeve transform, and in the case of far less than the Nyquist sampling rate, the discrete values of the signal are obtained by random sampling, and the sparse basis is used for fitting, so as to realize the compression of the CFR. However, the compression ratio of this compression method is insufficient, and the compressed CFR cannot meet the demand of large-scale transmission of the CFR. Especially in the 5G small station scene, because the 5G small stations are distributed relatively densely and the number is relatively large, the number of CFRs to be transmitted is huge, and the use of the above method for compression of the CFR will cause redundant information such as phase, thereby causing unnecessary waste of resources in storage and transmission.

[0082] As shown in Figure 3 The present application provides a channel frequency response compression method, which comprises:

[0083] S100. Obtain an observation result obtained by observing a wireless channel between a transmitter and a receiver; the observation result comprises a plurality of channel frequency response (CFR) powers corresponding to a plurality of observation points;

[0084] In a possible implementation, the present application can observe the wireless channel between the transmitter and the receiver, and estimate the CFR amplitude on the receiver side through the trained preamble symbol to calculate a plurality of CFR powers corresponding to a plurality of observation points.

[0085] It can be understood that, assuming that there are K observation points on each antenna of the receiver and there is an observation result on each observation point, there will be K observation results on the receiver side. The above K observation results can be observation results obtained by observing the wireless channel between the transmitter and the receiver in the time domain (i.e. observation results of K frequency bands of the antenna at a certain fixed time point); or the above K observation results can be observation results obtained by observing the wireless channel between the transmitter and the receiver in the frequency domain (i.e. observation results of K time points of the antenna at a certain fixed frequency band).

[0086] S200. For each CFR power, according to the number Q of initial trigonometric functions corresponding to the CFR power, a curve fitting algorithm is used to fit the CFR power to obtain an initial fitting result; the initial fitting result comprises a CFR fitting result corresponding to each CFR power, and the CFR fitting result comprises Q trigonometric functions; Q is an integer greater than or equal to 1;

[0087] In a possible implementation, the number of initial trigonometric functions Q corresponding to the CFR power in S200 can be determined in the following manner.

[0088] For each CFR power, the CFR power is converted from the time domain to the frequency domain to obtain a spectrum of the CFR power, and the number of initial trigonometric functions Q corresponding to the CFR power is determined according to the number of spectral peaks in the spectrum. It can be understood that the CFR power can be converted from the time domain to the frequency domain by Laplace transformation, Fourier transformation, or z-transformation, which is not limited in the present application. It can be understood that in some scenarios that require to save computing power, the number of initial trigonometric functions Q can also be specified in advance, and the iterative fitting is performed based on the initial fitting.

[0089] As a possible implementation of the present application, the fitting algorithm described above can include fitting algorithms such as LM algorithm, Gauss-Newton method, gradient descent method, etc. that can solve the nonlinear least squares problem. These fitting algorithms can all find the global minimum of the functional in a given data interval, and by constructing the Jacobian matrix and linearization at each step, a linear equation set with parameters as independent variables is obtained, and the fitting result is obtained by solving.

[0090] S300. According to the residual between the observation result and the initial fitting result, the iterative fitting is performed to obtain a target fitting result with a residual less than a preset residual threshold, and the trigonometric function included in the target fitting result is taken as the compression result of the observation result.

[0091] As a possible implementation, in S300, according to the residual between the observation result and the initial fitting result, the iterative fitting is performed to obtain a target fitting result with a residual less than a preset residual threshold, which can include:

[0092] The iterative fitting process is performed: it is judged whether the residual between the observation result and the fitting result corresponding to the current iterative fitting process is less than or equal to the preset residual threshold; wherein the fitting result corresponding to the first round of iterative fitting process is the initial fitting result;

[0093] If the residual is less than the preset residual threshold, the fitting result corresponding to the previous iterative fitting process is taken as the target fitting result;

[0094] If the residual is greater than the preset residual threshold, Q is increased, i.e., the number of trigonometric functions is increased, and the curve fitting of the CFR power is performed using the fitting algorithm to obtain the fitting result of the next round; based on the obtained fitting result of the next round, the iterative fitting process is repeatedly performed until the target fitting result with the residual less than the preset residual threshold is obtained.

[0095] The iterative fitting process can be understood as a multi-round curve fitting process. Each round of curve fitting can be regarded as a process of fitting the CFR power using a fitting algorithm, and multiple curve fittings are performed. When a converged fitting result is obtained, or the number of curve fittings in this round exceeds the preset number of fitting times, the curve fitting in this round is completed, and residual error judgment is performed. Each round of curve fitting can only output the most converged fitting result under the current fitting order (i.e., the number of trigonometric functions used in curve fitting). If this result does not meet the preset residual error requirement, i.e., there is a large deviation between the fitting result and the observation result, the fitting order (i.e., the number of trigonometric functions used in curve fitting) needs to be increased until the residual error requirement is met.

[0096] The application uses trigonometric functions as base functions to compress the CFR, which can improve the compression ratio of the CFR compression. Using the compressed CFR for transmission and storage can save the resources occupied by the CFR in transmission and storage. At the same time, the compression method provided by the application has low computing power requirement and strong universality, and can quickly and efficiently compress the CFR.

[0097] For each CFR power, when the corresponding initial number of trigonometric functions Q is obtained, the CFR power can be fitted using a fitting algorithm, and the order in the fitting is also Q order. In the Q order fitting process, the initial fitting result can be finally obtained through multiple fittings. The initial fitting result is the optimal fitting result that meets the target function under the current order. If the initial fitting result does not meet the preset residual error requirement, the order needs to be increased, the number of trigonometric functions needs to be increased, and the fitting needs to be repeated until a fitting result that meets the preset residual error requirement is obtained. In some scenarios, for example, when the preset residual error requirement is not high, the initial fitting result that meets the preset residual error requirement can be obtained through the initial curve fitting. Therefore, the initial fitting result can be directly used to compress the CFR power.

[0098] As a possible implementation of the present application, the CFR power is curve fitted using a fitting algorithm, including: initializing a set of trigonometric functions according to Q, taking the residual between the set of trigonometric functions and the observation as an objective function, and curve fitting the CFR power using the fitting algorithm based on the objective function; the set of trigonometric functions includes Q trigonometric functions. When performing the first round of iterative fitting, i.e., Q order iterative fitting, Q trigonometric functions are needed, and Q trigonometric function parameters can be initialized, and each trigonometric function parameter can be randomly selected; when performing the second round of iterative fitting, i.e., Q+1 order fitting, Q+1 trigonometric functions can be initialized, and the Q trigonometric function parameters obtained in the first round of iterative fitting are not used. It can be understood that in some cases, when performing the second round of iterative fitting, Q+1 trigonometric functions are needed, one trigonometric function parameter can be initialized, and the remaining Q trigonometric function parameters are all the trigonometric function parameters in the last round of iterative fitting, i.e., Q order iterative fitting. The Q+1 order fitting can be continued on the basis of the Q order fitting, and the fitting efficiency of the present application when performing curve fitting can be improved.

[0099] The channel frequency response compression method provided by the embodiments of the present application will be described in detail below taking a 5G cellular station (also referred to as a 5G small station) communication system in an indoor scene as an example.

[0100] Referring to Figure 4 , Figure 4 The channel frequency response compression method provided by the present application is shown in the flowchart Figure 2 ,

[0101] The method includes:

[0102] S401. Collecting CFR data;

[0103] S402. Calculating the CFR power;

[0104] S403. Obtaining the fitting order based on the CFR power;

[0105] S404. Curve fitting the CFR power using a fitting algorithm;

[0106] S405. Calculating the fitting residual of the fitting result and the CFR power;

[0107] S406. Determining whether the iterative fitting is terminated based on the fitting residual; if the iterative fitting is not terminated, performing S404, if the iterative fitting is terminated, performing S407, and taking the fitting result of the current round of iterative fitting as the target fitting result;

[0108] S407. Compressing the CFR power based on the target fitting result.

[0109] As a possible implementation, S401 can include: in an indoor scene, a 5G microcell station perceives the position, behavior, etc. of a target in the scene through transceiving signals. The transmitter transmits wireless signals, and the receiver estimates CFR data through a trained preamble symbol after receiving the wireless signals.

[0110] As a possible implementation, S402 can include: the CFR data can be represented as H(f, t), and the power of the CFR is calculated. The CFR power can be represented as |H(f k , t)| 2 .

[0111] As a possible implementation, S403 can include: performing a fast Fourier transform on the CFR power to convert from the time domain to the frequency domain to obtain a spectrum graph of the CFR power, and determining the starting order of iterative fitting, that is, the number of trigonometric functions Q, according to the number of spectral peaks in the spectrum graph.

[0112] As a possible implementation, S404 can include: using the LM algorithm to perform curve fitting on the CFR power.

[0113] Suppose there are K observation points on each antenna of the receiver, and there is an observation result on each observation point. The CFR power on the Kth observation point can be represented as (x k , |H(f k , t)| 2 ); where x k represents an observation point, and 1≤k≤K.

[0114] Using a fitting algorithm to perform curve fitting on the CFR power, the Qth fitting result at x k is Γ(x k , A), where A=[A1, A2, …, AQ] Q ] T , A q =[a q , g q , θ q ] T , q∈[1, Q]

[0115] Q is the number of sine waves and the fitting order, and 1≤q≤Q; a q represents the amplitude of the qth sine wave, g q represents the frequency of the qth sine wave, and θ q represents the phase of the qth sine wave.

[0116] Based on the initial order Q of fitting, the Q trigonometric functions are initialized, and the CFR power is curve fitted using the Levenberg-Marquarelt (LM) algorithm; in the curve fitting, the minimum residual error between the fitting result and the CFR power is taken as the target, and the following equation is obtained:

[0117]

[0118] The first order derivative of E(A) in equation (9) is obtained as follows:

[0119]

[0120] Wherein, e(A) = |H(f k , t)| 2 - Γ(x k , A), represents the error between the observation value and the fitting result, and J(A) is the Jacobian matrix of e(A), which can be expressed as:

[0121]

[0122] The second order derivative of E(A) is obtained as follows:

[0123]

[0124] Wherein,

[0125] Therefore, the target function in equation (9) can be transformed as follows:

[0126]

[0127] The least square problem is obtained as follows:

[0128] A (n+1) = A (n) - ((J (n) ) T J (n) +S (n) ) -1 (J (n) ) T e(A (n) ) Equation (13)

[0129] The above equation (13) can be used to represent the relationship between the n+1th fitting and the nth fitting under the Q order fitting; wherein, A (n) represents the trigonometric function parameters calculated by the nth fitting under the Q order fitting, S (n) represents the S(A) calculated by the nth iteration under the Q order fitting, J (n) is the Jacobian matrix calculated by the nth fitting under the Q order fitting, and e(A(n) Let H(A) be the error between the CFR fitting result of the nth fitting under Q-order fitting and the CFR power. Without loss of generality, the second-order term S(A) can be ignored, therefore the Hessian matrix H(A) is... (n) =(J (n) ) T J (n) The target search direction can be approximated by the nth fitting under the Q-order fitting of V(A), and the target search direction under the nth fitting under the Q-order fitting is expressed as:

[0130]

[0131] Where, μ (n) Let μ be the damping coefficient of the nth fitting under Q-order fitting. (n) >0, minimize E(A), and obtain the result in A (n) The target search direction is:

[0132] d (n) =-((J) (n) ) T J (n) +μ (n) I) -1 (J (n) ) T e (n) Equation (15)

[0133] Where I is the identity matrix. Let v (n) =(J (n) ) T e (n) For any μ (n) >0, v (n) d (n) =-(J (n) ) T e (n) ((J (n) ) T J (n) +μI) -1 (J (n) ) T e (n) <0 Equation (16)

[0134] This application introduces a damping coefficient μ. (n) Thus constructing a positive definite matrix (J) (n) ) T J (n) +μ (n) I to correct H (n) That is, when μ (n) When H is large enough, (n) It is a full-rank matrix. The Hessian matrix H can be solved. (n) The problem of not being able to iterate when the rank is not full.

[0135] In the LM algorithm, the Armijio search method is used to calculate the step size a, and the inequality is satisfied:

[0136] E(A (n) +ρ m d (n) )≤E (n) +σρ m (v (n) ) T d (n) Formula (17)

[0137] The minimum non-negative integer m, where σ, ρ ∈ (0, 1). Based on the m satisfying formula (17), the step size a can be obtained, where a = ρ m .

[0138] Based on the search direction and step size described above, fitting is performed, and each time fitting is based on the step size and search direction to approach the convergent fitting result; after n times of fitting, the convergent fitting result Γ(x k , A (n) ) is obtained. At the same time, considering that in some special cases, many times of fitting are performed, and the convergent fitting result is also not obtained, therefore, the application can monitor the size of n, when n is greater than the preset fitting times threshold, the fitting process is stopped, and the fitting result is output.

[0139] As a possible implementation, S404 can also include using the Gauss-Newton method or the gradient descent method to perform curve fitting on the CFR power;

[0140] For the Gauss-Newton method, the calculation process only needs to introduce the LM algorithm into the damping coefficient μ (n) Set to 0; that is, formula (15) becomes:

[0141] d (n) = -((J (n) ) T J (n) ) -1 (J (n) ) T e (n) Formula (15) a

[0142] Therefore, formula (16) becomes:

[0143] v (n) d (n) = -(J (n) ) T e (n) ((J (n) ) T J (n) ) -1 (J(n) ) T e (n) <0 Equation (16)a

[0144] The remaining steps and formulas can be referred to the above process and will not be repeated here; using the Gauss-Newton method for curve fitting can significantly reduce the amount of computation required for curve fitting, and the Gauss-Newton method requires far less computation than the LM algorithm, especially when the matrix (J) is involved. (n) ) T J (n) +S (n) When it is reversible, the amount of computation required can be greatly reduced, and fitting results can be obtained quickly and accurately.

[0145] For the gradient descent method, equation (9) can be applied to A. o The first-order Taylor expansion is obtained at the point where

[0146] E(A) = E(A0) + v(A)(A-A0), thus optimizing the objective function is: The objective function can then be further transformed into:

[0147] E(A0) + v(A)(A - A0) = 0. For this objective function, the update of A during curve fitting can be expressed as... Compared to the Newton-Gaussian method, gradient descent can provide faster and more accurate fitting results when the objective function is convex, requiring less computational power.

[0148] As one possible implementation, S405. Calculating the fitting residual between the fitting result and the CFR power may include: assuming the CFR fitting result Γ(x) of the nth iteration. k A (n) The fitting residual between the power and the CFR power is:

[0149]

[0150] As one possible implementation, S406. Determining whether the iterative fitting has terminated based on the fitting residual may include: determining whether the fitting residual is greater than a preset residual threshold.

[0151] If the residual is greater than the preset residual threshold, the fitting order (i.e. the number of trigonometric functions) Q is increased for iterative fitting, and S404, S405 and S406 are repeated until the fitting residual is less than or equal to the preset residual threshold.

[0152] Preferably, the above increases the fitting order (i.e., the number of trigonometric functions) Q to perform the iterative fitting, and repeatedly performing S404, S405 and S406 can include: adding 1 to Q, and then repeatedly performing S404, S405 and S406. As a preferred embodiment, when performing the Q+1 order fitting to perform S404, the parameters of the first Q trigonometric functions can inherit the Q trigonometric function parameters obtained in the Q order fitting, and the search direction and step size obtained in the last fitting of the Q order fitting can be used as the search direction and step size of the first time of the Q+1 order fitting.

[0153] A (n+1) = A (n) + a (n) d (n) Equation (19)

[0154] That is, using Equation (19) to replace Equation (13) can accelerate the curve fitting speed of the LM algorithm and reduce the computing power required for curve fitting.

[0155] Referring to Figure 5 and Figure 6 , Figure 5 The compression result comparison diagram of the compression method provided by the embodiment of the present application is shown in FIG. 6. The solid line in the figure is the original CFR power curve, and the dashed line is the fitted CFR power curve (i.e., the compressed CFR power curve). It can be seen that the compressed CFE power curve is consistent with the original CFR power curve in overall trend. And the compressed CFR power curve is composed of 8 trigonometric functions, and the partial screenshots of the 8 trigonometric functions are shown in FIG. 7. Figure 6 Figure 6 The compression result comparison diagram of the compression method provided by the embodiment of the present application is shown in FIG. 6. The solid line in the figure is the original CFR power curve, and the dashed line is the fitted CFR power curve (i.e., the compressed CFR power curve). It can be seen that the compressed CFE power curve is consistent with the original CFR power curve in overall trend. And the compressed CFR power curve is composed of 8 trigonometric functions, and the partial screenshots of the 8 trigonometric functions are shown in FIG. 7.

[0156] The embodiment of the present application can divide the channel frequency response compression device into functional modules or functional units according to the above method examples. For example, each functional module or functional unit can be divided according to each function, or two or more functions can be integrated into one processing module. The above integrated module can be realized in the form of hardware or in the form of software functional module or functional unit. The division of modules or units in the embodiment of the present application is illustrative, and is only a logical functional division. Actual implementation can have another division mode. ​

[0157] As Figure 7 shown, a structural schematic diagram of a channel frequency response compression device provided by an embodiment of the present application is shown, the device comprising:

[0158] The communication unit 202 is configured to obtain an observation result obtained by observing a wireless channel between a transmitter and a receiver; the observation result comprises a plurality of channel frequency response (CFR) powers corresponding to a plurality of observation points.

[0159] The processing unit 201 is configured to, for each CFR power, perform curve fitting on the CFR power according to an initial fitting order Q corresponding to the CFR power, to obtain an initial fitting result; the initial fitting order corresponding to the fitting algorithm is Q; the initial fitting result comprises a CFR fitting result corresponding to each CFR power, and the CFR fitting result comprises Q trigonometric functions; Q is an integer greater than or equal to 1.

[0160] The processing unit 201 is further configured to perform iterative fitting according to a residual between the observation result and the initial fitting result, to obtain a target fitting result with a residual less than a preset residual threshold, and to take the trigonometric functions included in the target fitting result as a compression result of the observation result.

[0161] When implemented by hardware, the communication unit 202 in the embodiment of the present application can be integrated on a communication interface, and the processing unit 201 can be integrated on a processor. The specific implementation manner is shown in Figure 8 .

[0162] Figure 8 Another possible structural schematic diagram of the channel frequency response compression device involved in the above embodiment is shown. The channel frequency response compression device comprises a processor 302 and a communication interface 303. The processor 302 is configured to control and manage the actions of the channel frequency response compression device, for example, to perform the steps performed by the processing unit 201 described above, and / or to perform other processes of the technology described herein. The communication interface 303 is configured to support the communication between the channel frequency response compression device and other network entities, for example, to perform the steps performed by the communication unit 202 described above. The channel frequency response compression device can further comprise a memory 301 and a bus 304, and the memory 301 is configured to store the program code and data of the channel frequency response compression device.

[0163] The memory 301 can be a memory in the channel frequency response compression device, which can include a volatile memory such as a random access memory, and can also include a non-volatile memory such as a read-only memory, a flash memory, a hard disk or a solid state disk, and can also include a combination of the above kinds of memories.

[0164] The processor 302 described above can be a central processing unit, a general purpose processor, a digital signal processor, an application specific integrated circuit, a field programmable gate array, or other programmable logic device, transistor logic, hardware component, or any combination thereof. It can implement or execute various exemplary logical blocks, modules, and circuits described in connection with the disclosure. It can also be a combination of computing components, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0165] The bus 304 can be an Extended Industry Standard Architecture (EISA) bus or the like. The bus 304 can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, Figure 8 In the figure, only one thick line is used to represent the bus, but it does not mean that there is only one bus or only one type of bus.

[0166] Figure 9 FIG. 17 is a structural schematic diagram of a chip 170 provided by an embodiment of the present application. The chip 170 includes one or more (including two) processors 1710 and a communication interface 1730.

[0167] Optionally, the chip 170 further includes a memory 1740, which can include a read-only memory and a random access memory, and provides operation instructions and data to the processor 1710. A part of the memory 1740 can further include a non-volatile random access memory (NVRAM).

[0168] In some embodiments, the memory 1740 stores the following elements, execution modules or data structures, or a subset thereof, or an extended set thereof.

[0169] In an embodiment of the present application, corresponding operations are performed by calling operation instructions (which can be stored in an operating system) stored in the memory 1740.

[0170] The processor 1710 can implement or execute various exemplary logical blocks, units, and circuits described in connection with the disclosure. The processor can be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It can implement or execute various exemplary logical blocks, units, and circuits described in connection with the disclosure. The processor can also be a combination of components that implement computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0171] The memory 1740 can include a volatile memory, such as a random access memory, and can also include a non-volatile memory, such as a read-only memory, a flash memory, a hard disk, or a solid state disk. The memory can also include a combination of the above-mentioned types of memory.

[0172] The bus 1720 can be an extended industry standard architecture (EISA) bus or the like. The bus 1720 can be divided into an address bus, a data bus, a control bus, and the like. For convenience of representation, Figure 9 Only one line is used to represent the bus in the figure, but it does not mean that there is only one bus or only one type of bus.

[0173] Through the description of the above embodiments, those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of functional modules is taken as an example, and in actual application, the above-mentioned functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0174] The embodiment of the present application provides a computer program product containing instructions, when the computer program product runs on a computer, so that the computer executes the channel frequency response compression method in the method embodiment of the foregoing method embodiment.

[0175] The embodiment of the present application also provides a computer readable storage medium, and the computer readable storage medium stores instructions, when the instructions run on a computer, so that the computer executes the channel frequency response compression method in the method flow of the method embodiment.

[0176] The computer readable storage medium, for example, can be, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a register, a hard disk, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing, or any other medium from which a processor can read and write information. An exemplary storage medium is coupled to the processor such that the processor can read information from, and write information to, the storage medium. Of course, the disclosure is not limited to a particular storage medium. The processor and the storage medium can be located in an ASIC. In some embodiments, the computer readable storage medium can be any tangible medium that can contain or store program code for use by or in connection with an instruction execution system, apparatus, or device.

[0177] Embodiments of the present application provide a computer program product comprising instructions which, when executed on a computer, cause the computer to carry out the channel frequency response compression method as described in Figures 3 to 4

[0178] Since the channel frequency response compression apparatus, the computer readable storage medium, and the computer program product in the embodiments of the present application can be applied to the above method, the technical effects they can obtain can also be referred to the above method embodiments, which will not be described here again.

[0179] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented by other means. For example, the above-described device embodiments are only illustrative, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed units can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.​

[0180] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0181] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit.

[0182] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any change or replacement within the technical scope disclosed in the present application should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method of channel frequency response compression, characterized by, The method comprises: obtaining observation results obtained by observing a wireless channel between a transmitter and a receiver; the observation results comprise a plurality of channel frequency response (CFR) powers corresponding to a plurality of observation points; for each CFR power, using a fitting algorithm to perform curve fitting on the CFR power according to an initial number Q of trigonometric functions corresponding to the CFR power, to obtain an initial fitting result; the initial fitting result comprises a CFR fitting result corresponding to each CFR power, and the CFR fitting result comprises Q trigonometric functions; the Q is an integer greater than or equal to 1; performing iterative fitting according to residuals between the observation results and the initial fitting result, to obtain a target fitting result with a residual less than a preset residual threshold, and taking the trigonometric functions included in the target fitting result as compression results of the observation results; the initial number Q of trigonometric functions corresponding to the CFR power is determined in the following manner: for each CFR power, convert the CFR power from a time domain to a frequency domain to obtain a frequency spectrum of the CFR power, and determine the initial number Q of trigonometric functions corresponding to the CFR power according to a number of spectral peaks in the frequency spectrum.

2. The method of claim 1, wherein, The iterative fitting according to the residuals between the observation results and the initial fitting result to obtain a target fitting result with a residual less than a preset residual threshold comprises: performing an iterative fitting process: judging whether a residual between the observation results and a fitting result corresponding to a current iterative fitting process is less than or equal to a preset residual threshold; the fitting result corresponding to the first round of iterative fitting process is the initial fitting result; if the residual is less than the preset residual threshold, taking the fitting result corresponding to the current iterative fitting process as the target fitting result; if the residual is greater than the preset residual threshold, increasing the Q, using the fitting algorithm to perform curve fitting on the CFR power, calculating a fitting result of the next round, and repeating the iterative fitting process based on the obtained fitting result of the next round, until a target fitting result with a residual less than the preset residual threshold is obtained.

3. The method of claim 2, wherein, The curve fitting on the CFR power using the fitting algorithm comprises: constructing a trigonometric function group according to the Q initial trigonometric functions, taking a residual between the trigonometric function group and the observation results as an objective function, and using the fitting algorithm to perform curve fitting on the CFR power based on the objective function; the trigonometric function group comprises Q trigonometric functions.

4. The method of claim 3, wherein, The objective function is: Where E(A) is the residual between the trigonometric function set and the observation result. For CFR power, For the observation results, It is a trigonometric function system. This represents the frequency of the k-th subcarrier of the OFDM signal. Let A represent the power of the sum of Q trigonometric functions of the k-th subcarrier, where A is the trigonometric function parameter, t is time, k is a positive integer greater than or equal to 1, and K is the number of subcarriers. , , , The number of sine waves is also the fitting order, and q is an integer. ; Indicates the first The amplitude of a sine wave, Indicates the first The frequency of a sine wave, Indicates the first The phase of a sine wave.

5. The method of claim 3, wherein, The fitting algorithm comprises: LM algorithm, Gauss-Newton method, and gradient descent method.

6. A channel frequency response compression apparatus characterized by comprising: The device comprises: a communication unit configured to obtain observation results obtained by observing a wireless channel between a transmitter and a receiver; the observation results comprise a plurality of channel frequency response (CFR) powers corresponding to a plurality of observation points; The processing unit is configured to perform curve fitting on each CFR power by using a fitting algorithm according to an initial number Q of trigonometric functions corresponding to the CFR power, to obtain an initial fitting result; the fitting algorithm corresponds to an initial fitting order of the Q; the initial fitting result includes a CFR fitting result corresponding to each CFR power, and the CFR fitting result includes Q trigonometric functions; the Q is an integer greater than or equal to 1; The processing unit is further configured to perform iterative fitting according to a residual error between the observation result and the initial fitting result, to obtain a target fitting result with a residual error less than a preset residual error threshold, and to take the trigonometric functions included in the target fitting result as the compression result of the observation result. The initial number Q of trigonometric functions corresponding to the CFR power is determined in the following manner: For each CFR power, the CFR power is converted from a time domain to a frequency domain to obtain a frequency spectrum of the CFR power, and the initial number Q of trigonometric functions corresponding to the CFR power is determined according to a number of spectral peaks in the frequency spectrum.

7. A channel frequency response compression apparatus characterized by comprising: The processing unit is configured to perform curve fitting on each CFR power by using a fitting algorithm according to an initial number Q of trigonometric functions corresponding to the CFR power, to obtain an initial fitting result; the fitting algorithm corresponds to an initial fitting order of the Q; the initial fitting result includes a CFR fitting result corresponding to each CFR power, and the CFR fitting result includes Q trigonometric functions; the Q is an integer greater than or equal to 1; When a computer executes the instruction, the computer executes the method of any one of claims 1-5.

8. A computer-readable storage medium having stored therein instructions, the computer-readable storage medium comprising: The processing unit is configured to perform curve fitting on each CFR power by using a fitting algorithm according to an initial number Q of trigonometric functions corresponding to the CFR power, to obtain an initial fitting result; the fitting algorithm corresponds to an initial fitting order of the Q; the initial fitting result includes a CFR fitting result corresponding to each CFR power, and the CFR fitting result includes Q trigonometric functions; the Q is an integer greater than or equal to 1 9. A chip comprising a processor and a communication interface, the communication interface and the processor being coupled, characterized in that, ​

Images