WiFi sensing anti-interference method and WiFi device

Abstract

The invention relates to the field of communication perception integration, in particular to a WiFi (Wireless Fidelity) perception anti-interference method, which is used for solving the problem that the perception performance is reduced due to interference in wireless networks of the same frequency band along with the increase of the number of WiFi equipment. According to the method, an unknown interference level is measured statistically and the quality of channel state information is improved by utilizing interference measurement based on variance and a channel state information optimization method based on maximum ratio combination, so that the WiFi sensing performance is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication perception integration, and in particular to a WiFi perception anti-interference method and a perception device. Background Art

[0002] As the number of WiFi devices increases, interference is common in wireless networks in the same frequency band, which affects WiFi perception performance.

[0003] Current WiFi sensing research does not consider the impact of interference. In fact, perception is more sensitive to interference than communication. Even if data transmission is not affected at a certain interference level, interference may still affect perception performance. Summary of the invention

[0004] The purpose of the present invention is to solve the problem of decreased perception performance caused by interference in wireless networks in the same frequency band as the number of WiFi devices increases in ISAC.

[0005] To achieve the above technical objectives, the specific technical solutions are as follows.

[0006] In a first aspect, the present invention proposes a WiFi perception anti-interference method, the method comprising the following steps: obtaining the transmission and reception signals of the pilot symbol in each subcarrier, and obtaining the reception signal of the data symbol, constructing a transmission signal matrix and a reception signal matrix, the matrix size is M×N, N is the number of symbols, M represents the number of subcarriers; based on the i-th reception signal of the j-th subcarrier and the i-th transmitted signal Calculate the channel state information of the i-th symbol of the j-th subcarrier Based on the fact that the noise distribution is a Gaussian distribution with a mean of 0, the estimated mean of the channel state information of the jth subcarrier is obtained Then obtain the variance of the channel state estimation mean; determine the highest quality symbol number K based on the variance and a preset threshold; based on the channel state information and the mean channel state information estimate Get quality metrics Construct channel state information quality matrix And select K symbols with the highest quality from the quality matrix Q; based on the K symbols with the highest quality, use maximum ratio combining to obtain optimal channel state information for each subcarrier; wherein i=1, 2, ..., N, j=1, 2, ..., M.

[0007] In one implementation of the above technical solution, the preset threshold is the L variance threshold Variance threshold in L There is a variance greater than When the value is Otherwise, K is the set value; where l∈[1,L], L is the preset value, for The corresponding preset value.

[0008] In one implementation of the above technical solution, the estimated mean value of the channel state information of the j-th subcarrier is and the variance of the mean of the channel state estimation, the acquisition steps include: the channel state information estimation value on the jth subcarrier is based on It follows a Gaussian distribution with a mean of 0. Expectations based on The expectation of The variance is used to estimate the interference level of the symbols on the subcarrier; wherein, is the noise of the i-th symbol of the j-th subcarrier.

[0009] In one implementation of the above technical solution, the quality metric value

[0010] In one implementation of the above technical solution, the step of obtaining the optimal channel state information of each subcarrier by using maximum ratio combining includes: constructing the transmission signal on the jth subcarrier by using the K symbols with the highest quality in each subcarrier Get x j The corresponding received signal Solving optimization problems Get the optimal weight w, w = [w1, w2, ..., w K ] T , Λ is a diagonal matrix reflecting the uncorrelated characteristics of interference between different symbols; based on the weights w and y j , the optimal channel state information of the jth subcarrier is

[0011] In one implementation of the above technical solution, the method further includes: obtaining a channel state information vector with the highest quality from each data packet, which is represented as h=[h1, h2, ..., h M ], where h j Represents the optimized channel state information on the j-th subcarrier; within a set time, a set of h is extracted from multiple data packets, and the channel state information matrix of each activity is constructed after preprocessing as the input of the perception model for classification or regression.

[0012] In a second aspect, the present invention proposes a WiFi device that uses any of the above methods to perform anti-interference processing on channel state information to improve perception performance.

[0013] Beneficial technical effects of the present invention: The method first measures the interference level of each subcarrier based on the variance method to determine the symbol that can be used to improve the channel state information; and then uses multiple symbol signals for optimization, thereby improving the perceived performance under interference conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0015] Figure 1 , standard WiFi transmission process and WiFipacket structure in time domain and frequency domain.

[0016] Figure 2 , interference of WiFi data packets in time domain and frequency domain.

[0017] Figure 3 , one The method flow diagram of an implementation manner in which the quality of channel state information (CSI) is enhanced to improve the performance of WiFi sensing under interference is shown.

[0018] Figure 4 , one A diagram of the signal collection process when a data packet is correctly demodulated in an implementation manner.

[0019] Figure 5 , one CSI of symbols on subcarriers under two interference levels in this implementation manner.

[0020] Figure 6 , one Schematic diagram of the impact of K on CSI quality in various implementation modes.

[0021] Figure 7 , one Schematic diagram of the CSI enhancement algorithm for the j-th subcarrier in an implementation manner. DETAILED DESCRIPTION

[0022] Wireless sensing technology has emerged as a promising sensing paradigm in the past few years. It uses the transmission and reception of wireless signals to detect and measure physical characteristics and changes in the environment. Common applications include home security monitoring, healthcare monitoring, etc. Current research focuses on extracting changes from the channel state information (CSI) of WiFi signals to capture target movement in the surrounding environment, such as human activity recognition and gesture recognition. These studies are implemented on a set of WiFi devices dedicated to sensing and specific channels, and do not consider the network characteristics of WiFi devices.

[0023] With the development of 6G communication perception integration, some researchers have begun to pay attention to the ISAC design on WiFi systems. In the past two years, some related work has been carried out, and the basic idea is to use communication data packets to achieve perception. For example, the IEEE802.11bf task group attempted to revise the IEEE 802.11 standard to support WiFi perception, mainly through MAC (media access control) layer protocol design. For example, CSI is extracted from the WiFi communication data packet sent by the AP (access point) to identify the target activity, mainly to solve the CSI calibration problem in different WiFi communication modes when the target of the data packet is not the perception device. For another example, by focusing on the millimeter wave band WiFi system based on the IEEE 802.11ad standard, the main lobe signal is first used for communication, and then the side lobe signal is used for perception to achieve communication perception integration.

[0024] The following will be combined with the accompanying drawings to clearly and completely describe how to implement the technical solution of this case. Obviously, the described implementation method is only a part of the implementation method of this case, not all of the implementation methods. Based on the implementation method in this case, all other implementation methods obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0025] 1. Basic knowledge

[0026] (1.1) WiFi transmission process

[0027] Figure 1(a) in Figure 1 describes the transmission process of standard WiFi data bits. The data bits first pass through the channel coding module to resist a certain degree of interference, and then are mapped to constellation points according to the modulation method adopted (such as BPSK (Binary Phase ShiftKeying), QPSK (Quadrature Phase Shift Keying) or different levels of QAM modulation (Quadrature Amplitude Modulation), such as QAM-16, QAM-64 and QAM-256). The constellation points are mapped into OFDM subcarriers after the S / P (serial to parallel) module and output as time-domain OFDM symbols after the IFFT (Inverse Fast Fourier Transform) and P / S (Parallel to Serial) processes. A CP (Cyclic Prefix) is inserted into each OFDM symbol to eliminate inter-symbol interference. The final signal is transmitted through the RF front end.

[0028] Figure 1 (b) in the figure is a simplified WiFi data packet structure after expansion from the time domain and frequency domain. In the frequency domain, OFDM divides the WiFi channel into multiple orthogonal subcarriers to carry data in parallel. Specifically, each 20MHz WiFi channel is divided into 64 subcarriers, including 48 data subcarriers, 4 pilot subcarriers, and 12 empty subcarriers. The pilot subcarriers are used for fine channel estimation. Here, we ignore Figure 1 The empty subcarriers in (b) of FIG. 1 do not affect the implementation of this design. In the time domain, the WiFi signal includes a preamble symbol and a group of data OFDM symbols following it.

[0029] (1.2) WiFi perception process

[0030] The principle of WiFi sensing is that the movement of the sensing target will affect the WiFi channel, so the target motion information can be obtained by analyzing the channel changes caused. Most studies implement WiFi sensing by analyzing the CSI extracted from the leading part of the WiFi data packet, such as Figure 1 As shown in (b) of the figure. The preamble is used for synchronization and channel estimation during the demodulation process. Its transmitted signal is a known symbol Where j represents the index of the subcarrier, and when there are M subcarriers, j∈[1,M]. Use Denotes the interference and background noise on the j-th subcarrier of the leading edge, using h j represents CSI, then the received signal corresponding to the jth subcarrier for:

[0031]

[0032] The receiver can obtain the CSI vector [h1, h2, …, h M ] T The basic process of WiFi detection is similar, based on CSI collected from a set of packets over a period of time, which includes CSI preprocessing to remove noise, activity interval detection, and feature extraction from CSI for classification or regression, depending on the application.

[0033] Current WiFi sensing research does not consider the impact of interference, which means that in formula (1) When h represents only negligible background noise, j It can be approximated as

[0034] However, when interference is present, For h j The impact of estimation cannot be ignored. In fact, perception is more sensitive to interference than communication. Even if data transmission can be completed normally under a certain interference level, the perception performance may have been seriously affected.

[0035] (1.3) Interference in WiFi networks

[0036] The interference situation in WiFi networks is complex. Different degrees of interference may exist in each symbol and subcarrier of a WiFi data packet, which can be divided into time domain interference and frequency domain interference.

[0037] In the time domain, since wireless devices use the CSMA / CA mechanism for channel access in the ISM band, the start time of the data packet is random. At the same time, the length of the data packet will also vary.

[0038] For example, in Figure 2 In (a), the target WiFi data packet is the data packet from which we need to extract CSI for sensing. IF data packets represent those data packets that interfere with the target data packet, which may be WiFi data packets or other types of data packets. The target data packet may encounter various interference situations, such as interference from IF data packet 1 in the first segment of the symbol, interference from IF data packet 2 in the middle segment of the symbol, and interference from IF data packet 3 in the last segment of the symbol. We note that multiple IF data packets may interfere with the target data packet at the same time. Considering that the signal power of each IF data packet on the sensing device will be different, a WiFi data packet may experience different degrees of interference in different symbols. The sensing device in this case is a WiFi device.

[0039] In the frequency domain, a WiFi channel may overlap with a ZigBee channel with a 2MHz bandwidth and a Bluetooth channel with a 1MHz bandwidth. It may also partially overlap other WiFi channels. Figure 2 (b) in Figure 1 shows an example where a target 20MHz WiFi channel overlaps with a ZigBee channel, a Bluetooth channel, and other WiFi channels. IF packets in different channels will interfere with different subcarriers of the target WiFi packet. Therefore, WiFi packets may encounter different levels of interference in different subcarriers.

[0040] (II) WiFi perception anti-interference design method

[0041] (2.1) Overview

[0042] This solution improves the performance of WiFi sensing under interference by enhancing the quality of channel state information (CSI). Figure 3 shown.

[0043] The WiFi device that performs the sensing task first collects the transmitted and received signals of each data packet and reconstructs the channel state information. It then performs anti-interference processing to optimize the CSI, including two general methods, namely variance-based interference measurement and maximum ratio combining (MRC)-based CSI optimization, which measure the unknown interference level from a statistical perspective and improve the CSI quality. Based on these two methods, we propose a practical CSI enhancement process to adapt to the complex interference conditions in actual networks. Then, the device can obtain high-quality CSI from each data packet and input it into the CSI preprocessing module to remove outliers caused by burst noise, and obtain the sample index of each activity at the beginning and end of the CSI to realize activity window detection. Finally, it inputs the CSI associated with each activity into the sensing model.

[0044] Figure 3 The gray blocks in the figure represent the newly added modules in the present invention, while the white blocks represent the existing modules currently related to CSI-based WiFi sensing. This solution can be regarded as an extension of the current WiFi sensing work, and the optimized CSI can be used as the input of the current work to improve its sensing performance under interference. Where X is the transmitted signal and Y is the received signal.

[0045] (2.2) Signal acquisition

[0046] According to our design, the sensing device should collect the transmitted and received signals of each data packet to optimize the channel state information under interference. The signal collection process of each data packet is as follows: Figure 4As shown. Since the preamble symbol and the pilot symbol are known to the receiving end, the transmission and reception signals of the preamble symbol can be obtained during the preamble synchronization, and the transmission and reception signals of the pilot symbol and the reception signal of the data symbol can be obtained during the OFDM demodulation. After successful demodulation, the transmitted data signal can be obtained. That is, for each WiFi data packet, compared with the prior art that only uses the preamble symbol to calculate the CSI, the present method also extracts the pilot symbol and the data symbol in addition to improve the quality of the channel state information and improve the performance of WiFi perception.

[0047] It should be noted that we only utilize WiFi packets that can be demodulated correctly, and packets that fail to demodulate will be discarded and not used for sensing.

[0048] The sensing device constructs a transmit signal matrix X and a receive signal matrix Y for each received data packet. X is expressed as:

[0049]

[0050] in represents the transmitted signal of the i-th symbol in the data packet on the j-th subcarrier, M represents the number of subcarriers, and N represents the number of symbols. The structure of Y is the same as that of X. For each have:

[0051]

[0052] in represents the interference and background noise of the i-th symbol on the j-th subcarrier in the data packet. The background noise is usually much smaller than the interference, and we also regard it as a component of the interference.

[0053] In the following sections, use h j To represent the CSI on the jth subcarrier. When the duration of a data packet does not exceed a few milliseconds, it can be considered that the CSI between different symbols in a data packet remains unchanged.

[0054] (2.3) Variance-based interference measurement

[0055] The first step in anti-interference is to measure the interference level of each subcarrier. For formula (3), due to the randomness of unknown interference, it is almost impossible to obtain an accurate h. Therefore, a variance-based method can be used to statistically measure the interference level of each subcarrier. The CSI on the jth subcarrier is estimated by the following formula:

[0056]

[0057] in, is the estimated CSI, including h jand additional interference components

[0058] make represents the CSI estimated from all symbols on the jth subcarrier. It can be observed that is a set of random variables, the variance of each element depends on SINR (signal to noise ratio). For each element calculated by formula (4) It is known that is unknown and is considered a random variable.

[0059] We assume that the interference It obeys a Gaussian distribution with a mean of 0. Since the channel coefficient h j and interference component is independent, so The mean is:

[0060]

[0061] in is equivalent interference. The mean of is zero, which can be verified Yes j is an unbiased estimate of , so we can get The expectation is:

[0062]

[0063] The variance of can be calculated as follows:

[0064]

[0065] When the number of symbols is large enough, the variance of the estimated CSI is negatively correlated with the SINR. Therefore, we can estimate the interference level of the symbol on any subcarrier by calculating the variance of the CSI. Figure 5 The real part of CSI extracted from a set of symbols on one subcarrier is shown in low and high interference cases. Similarly, the imaginary part has similar characteristics. It can be seen that in the case of SINR of 30dB, the variance of CSI is much lower than that of SINR of 10dB, and CSI will have more identifiable features.

[0066] It should be noted that there is usually a CFO (center frequency offset) Δ between the transmitter and the receiver. f , which will cause phase deviation. A solution widely used in WiFi communication scenarios can be used, namely Therefore, we ignore this process in the analysis and replace the is regarded as the received signal after frequency offset compensation.

[0067] (2.4) CSI Optimization Based on Maximum Ratio Combining (MRC)

[0068] In this section, we propose to optimize CSI based on Maximum Ratio Combining (MRC) to improve the CSI quality of each subcarrier by combining the signals of multiple symbols.

[0069] Assumptions is a set of symbols sent on the jth subcarrier that will be used to optimize the CSI quality. We have:

[0070] y j =h j ·x j +n j (8)

[0071] in represents the corresponding received signal, Indicates the corresponding interference. j During these symbols the value is considered to remain constant.

[0072] According to the MRC method, a higher SINR will lead to a higher CSI estimation accuracy. To maximize SINR, we set weights w = [w1, w2, ..., w K ] T , and transform formula (8) into:

[0073] y′ j =w T ·y j =h j ·w T ·x j +w T ·n j (9)

[0074] Then the power of the received target signal is:

[0075]

[0076] Since the interference is unknown, we can only use the interference n j Treat it as a set of random variables and obtain its power statistically:

[0077]

[0078] where Λ is a diagonal matrix, reflecting the uncorrelated nature of interference between different symbols, and

[0079]

[0080] in

[0081] The design goal here is to obtain the weight w to maximize the SINR, that is, P s / p n Then, determining w is equivalent to solving the following optimization problem:

[0082]

[0083] Since the probability of each symbol being interfered is usually the same, and the interference signals have the same distribution in statistics, assuming that the interference level of each symbol is statistically consistent, each element σ in formula (12) is kk have the same value. Based on this assumption, the optimal weight w can be calculated

[0084]

[0085] By using the weight w and formula (9), we can get the optimized CSI on the jth subcarrier:

[0086]

[0087] We theoretically analyze how this CSI optimization process improves CSI quality. We calculate h′ j =w T ·y j Variance of:

[0088]

[0089] in is the variance of the interference, which indicates the energy level of the interference in the environment. It can be seen that the optimized variance is much lower than the value calculated by the original received signal, which is equivalent to The results show that the optimization process can significantly improve the quality of CSI.

[0090] In order to further analyze the impact of K value on the optimization process, we can take the derivative along K to obtain:

[0091]

[0092] This means that as K increases, the interference suppression will decrease. Figure 6As shown in the figure, when the SINR is 10dB, even when K = 5, the CSI quality is significantly improved compared to the original quality when K = 1, and the quality is even higher when K increases to 35. We can also see that as K increases, the improvement in quality becomes smaller and smaller, and when K exceeds 35, the improvement becomes insignificant. Therefore, a balanced K value is needed in actual networks to ensure the effectiveness of interference resistance at a lower computational complexity.

[0093] (2.5) CSI enhancement process for practical systems

[0094] The previous design provides a theoretical method for interference measurement and CSI optimization from a statistical perspective. However, different symbols in the actual network may have inconsistent interference conditions, and how to select symbols for CSI optimization is a challenge. In this section, we generate a CSI quality matrix for each packet to measure interference at a more granular level, and then propose a general algorithm to adapt to different interference conditions.

[0095] Formula (5) and Formula (6) show that the mean CSI on the jth subcarrier is close to the actual CSI for a relatively large number of symbols. At the same time, based on the previous analysis of the variance, we see that the distance between each symbol and the mean can represent the SINR situation. A smaller distance indicates a larger SINR, and the CSI quality calculated by this symbol is higher. Therefore, we further estimate the CSI quality calculated for each symbol based on this observation. For each received packet, we will is considered as the benchmark for the jth subcarrier because it is an unbiased estimate with small variance. We generate a CSI quality matrix for each symbol on each subcarrier, namely:

[0096]

[0097] in and Therefore, the larger It means that the CSI quality extracted from the i-th symbol on the j-th subcarrier is higher.

[0098] Then we propose a general algorithm for CSI optimization of the j-th subcarrier, whose pseudo code is as follows: Figure 7 As shown. The sensing device first uses the variance to roughly measure the SINR level of this subcarrier and determine the required K value. Since the variance is inversely proportional to the SINR, we use the L variance threshold The SINR situation is divided into several levels. The first element The highest value indicates the lowest SINR threshold. As the preset threshold, l∈[1,L], Variance threshold in L There is a variance greater than When the value is Otherwise, K is the set value, where L is the preset value. represents the lth variance threshold The value of K is usually larger. Corresponding The larger the variance, the higher the SINR level. Note that the SINR level and K value associated with the variance are determined empirically. Afterwards, the sensing device selects the K symbols with the highest quality through the CSI quality matrix, calculates the weight w according to equation (14), and then calculates h through equation (15). j . When the variance is higher than the maximum threshold When , we consider that this subcarrier is severely interfered and set K to 100 to improve the CSI quality as much as possible.

[0099] After anti-interference processing, a high-quality CSI vector can be obtained from each data packet, represented as h = [h1, h2, ..., h M ], where h j Denotes the optimized CSI on the jth subcarrier. Based on a set of h extracted from multiple packets over a period of time, CSI preprocessing is performed to construct the CSI matrix of each activity as the input of the perception model for classification or regression.

[0100] So far, the implementation process of the above technical solution includes: a variance-based method to measure the interference level of each subcarrier; a general CSI optimization method based on MRC, which is optimized by utilizing multiple symbol signals; and a CSI enhancement process to improve CSI quality under complex interference conditions in actual networks. During the implementation process, the interference problem in the integration of WiFi communication perception has been fully considered, and the performance of WiFi perception has been improved. The above technical solution has been verified on the USRP software radio platform and has been proven to be fully feasible.

[0101] Through the description of the above implementation modes, those skilled in the art can clearly understand that the present disclosure can be implemented by means of software plus necessary general hardware, and of course can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions performed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structures used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits, etc. However, for the present disclosure, software program implementation is a better implementation mode in most cases.

[0102] Although the embodiments of the present invention are described above in conjunction with the accompanying drawings, the present invention is not limited to the above specific embodiments and application fields, and the above specific embodiments are only illustrative and instructive, rather than restrictive. A person of ordinary skill in the art can also make many forms under the guidance of this specification and without departing from the scope of protection of the claims of the present invention, all of which belong to the protection of the present invention.

Claims

1. A WiFi perception anti-interference method, characterized in that: The method comprises the following steps: Obtain the transmission and reception signals of the pilot symbol in each subcarrier, and obtain the reception signal of the data symbol, and construct a transmission signal matrix and a reception signal matrix, the matrix size is M×N, N is the number of symbols, and M represents the number of subcarriers; The i-th received signal based on the j-th subcarrier and the i-th transmitted signal Calculate the channel state information of the i-th symbol of the j-th subcarrier Based on the fact that the noise distribution is a Gaussian distribution with a mean of 0, the estimated mean of the channel state information of the jth subcarrier is obtained Then, the variance of the mean of the channel state estimation is obtained; Based on the variance and a preset threshold, determining the number K of symbols with the highest quality; Based on channel state information and the mean channel state information estimate Get quality metrics Construct channel state information quality matrix And select K symbols with the highest quality from the quality matrix Q; Based on the K symbols with the highest quality, using maximum ratio combining to obtain optimal channel state information for each subcarrier; Among them, i=1, 2,…, N, j=1, 2,…, M.

2. The method according to claim 1, characterized in that: The preset threshold is the L variance threshold Variance threshold in L There is a variance greater than When the value is Otherwise, K is the set value; where l∈[1,L], L is the preset value, for The corresponding preset value.

3. The method according to claim 1, characterized in that The mean channel state information estimate of the jth subcarrier and the variance of the mean of the channel state estimation, the acquisition steps include: The estimated channel state information on the jth subcarrier is based on It follows a Gaussian distribution with a mean of 0. Expectations based on The expectation of The variance is used to estimate the interference level of the symbols on the subcarrier; in, is the noise of the i-th symbol of the j-th subcarrier.

4. The method according to claim 1, characterized in that: Quality metrics 5. The method according to claim 1, characterized in that The steps of obtaining the optimal channel state information of each subcarrier by using maximum ratio combining include: Using the K highest quality symbols in each subcarrier, construct the transmitted signal on the jth subcarrier Get x j The corresponding received signal Solving optimization problems Get the optimal weight w, w = [w1, w2, ..., w K ] T , Λ is a diagonal matrix reflecting the uncorrelated characteristics of interference between different symbols; Based on the weights w and y j , the optimal channel state information of the jth subcarrier is 6. The method according to claim 1, characterized in that The method further comprises: A channel state information vector with the highest quality is obtained from each data packet and is represented as h = [h1, h2, ..., h M ], where h j represents the optimized channel state information on the j-th subcarrier; Within a set time, a set of h is extracted from multiple data packets, and after preprocessing, the channel state information matrix of each activity is constructed as the input of the perception model for classification or regression.

7. A WiFi device, characterized in that: Adopt any one of the methods of claims 1 to 6 to perform channel state information anti-interference processing.

Images