Method for predicting and avoiding wireless communication interference based on listening to wireless device communications

Abstract

The application discloses a method for predicting and avoiding wireless communication interference based on listening to wireless device communication, which comprises the following steps: S1, after parameter initialization, the algorithm predicts the time of channel interference; S2, the algorithm is used for overflow processing of the channel interference prediction result; S3, threshold limit detection is performed on the processed channel prediction result, and the detection result is updated and tracked; S4, each group of tracking is evaluated through the algorithm; S5, the score of each group of tracking is calculated, and the optimal tracking set is selected; S6, the tracking is pruned through the algorithm, and useless tracking is deleted. The application is based on the method for tracking and predicting multiple interference sources in the TDMA protocol, the average signal strength of all TDMA time slots is continuously detected to evaluate the signal interference from external devices, and the interference signals of different interference sources are identified.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wireless signal transmission, in particular to a method for predicting and avoiding wireless communication interference based on listening to wireless device communication. BACKGROUND

[0002] With the development of the Internet of Things, the number of devices with wireless network functions has also increased dramatically. However, the capacity of each Wireless Sensor Network (WSN) is limited, so each Internet of Things device can potentially interfere with other devices in its own WSN and devices in other WSNs. Furthermore, wireless signals between different technologies can also interfere with each other, such as 2.4 GHz wireless network signals and low-power Bluetooth network signals.

[0003] Signal interference and conflicts between WSNs can cause the loss of data packets transmitted between Internet of Things devices, and data packet loss can further cause data packet retransmission in the network, which in turn can cause data packet loss in the network, forming a vicious cycle. As the data packet loss rate increases, the time for devices to send data also increases, which increases the communication time of Internet of Things devices and greatly increases the power consumption of devices. This problem is particularly serious for low-power Internet of Things devices. In addition, signal conflicts between wireless networks can also cause data packets to violate real-time constraints and be unable to be sent to target machines within a specified time, which is particularly important for safety-critical applications and industrial control systems.

[0004] To reduce conflicts between wireless networks transmitting data packets, it is essential to detect the characteristics of device conflicts in the network, so as to have a basic understanding of the frequency bands that may exist in the network environment. A typical method is to measure the traffic size and data packet transmission error rate, packet loss rate, etc. in a certain channel to determine whether the channel has a high occupancy rate, and then exclude the channel from the list. However, this method only considers past and present signal interference in the wireless channel and cannot actively predict future interference that may occur in the channel. It simply avoids using channels with high occupancy rates. However, if the signal interference in the signal presents a specific frequency pattern, it can actually be predicted by this characteristic. For a wireless sensor network with the ability to self-schedule network device communication (such as a TDMA-based wireless protocol), interference can be predicted to avoid transmitting data packets when interference occurs, thereby reducing the impact of signal transmission. Fortunately, in some wireless sensor networks composed of low-power devices, there is a specific pattern of interference.

[0005] In the prior art, the traditional way of detecting channel interference is to determine the interference source in the channel according to the received signal strength indicator (RSSI) in the channel. The principle is to detect those signal sources that deviate significantly from the normal signal strength based on the average value of the signal strength of the TDMA wireless protocol. However, this technology only detects the signal strength in the channel, but does not detect the data packets affected by the interference. It can only detect and locate the existence of the interference source, but cannot identify the pattern characteristics of the interference signal, and thus cannot predict the occurrence of interference and avoid scheduling.

[0006] Meanwhile, another technology is the multiple hypothesis tracking algorithm (MHT), which is commonly used in radar and optical tracking. During the operation of the algorithm, a large number of hypothesis features are generated, and these hypothesis features are used to fit the known interference features. By comparing the prediction of the hypothesis features with the actual features, the closest hypothesis feature can be selected as the identification result, and the hypothesis that deviates significantly from the actual hypothesis is pruned. The multiple hypothesis tracking algorithm often needs to maintain a large number of hypothesis features and fit the actual features for calculation, which requires a large amount of calculation. In addition, the traditional multiple hypothesis tracking algorithm usually only tracks one target (which is caused by their application scenarios, such as radar), and cannot detect multiple targets at the same time. In the interference of the wireless channel, there are often multiple interference sources that need to be dealt with at the same time.

[0007] The traditional channel interference detection scheme either only focuses on the channel where the interference occurs or can only detect interference for a single target. In a low-power wireless sensor network, the interference generated in the network has certain identifiable interference patterns, so by detecting and classifying interference signals, the time of interference occurrence can be predicted. With this information, the time of wireless network data packet transmission can be controlled to avoid the time when interference occurs, thereby reducing the retransmission of data packets due to interference, improving data transmission efficiency, and greatly reducing the power consumption required for wireless transmission of Internet of Things devices. The purpose of this work is to track periodic multi-source interference in a wireless sensor network, and additional information about the radio frequency environment can be used to identify other devices, detect network intrusion, or synchronize with specific interference to monitor behavior.

[0008] At present, there is no effective solution to the problems in the related art. SUMMARY

[0009] In view of the problems in the related art, the present application proposes a method for predicting and avoiding wireless communication interference based on listening to wireless device communication to overcome the above technical problems existing in the prior art.

[0010] Therefore, the specific technical solution adopted by the present invention is as follows:

[0011] A method for predicting and avoiding wireless communication interference based on listening to wireless device communications, the method comprising the following steps:

[0012] S1. After parameter initialization, the algorithm predicts the timing of channel interference.

[0013] S2. Overflow processing is performed on the channel interference prediction results using an algorithm;

[0014] S3. Perform threshold limitation detection on the processed channel prediction results and update and track the detection results;

[0015] S4. Evaluate each tracking group using an algorithm;

[0016] S5. Calculate the score for each tracking group and select the optimal tracking set;

[0017] S6. Prune the tracking using an algorithm and delete useless tracking.

[0018] Furthermore, after parameter initialization, the algorithm uses a channel model and a Kalman filter to predict when channel interference may occur.

[0019] Furthermore, the overflow processing of the channel interference prediction results using the algorithm includes the following steps:

[0020] S21. Detect the number of time slots present in the time division multiple access frame;

[0021] S22. Verify whether the output of the channel interference prediction exceeds the number of time slots;

[0022] S23. If the result exceeds the limit, adjust the time slot around the time slot according to the rate of change.

[0023] Furthermore, the adjustment method for the surrounding adjustment is as follows:

[0024] when hour, Then no overflow handling is required;

[0025] when hour, Then it overflows downwards;

[0026] when hour, It will then overflow upwards.

[0027] In the formula, Where is the number of time slots, mod is the modulo function, and S represents the time slot number where interference occurs in each round of prediction. represents the adjusted time slot number.

[0028] Further, the threshold limiting detection on the processed channel prediction result and the update tracking on the detection result include the following steps:

[0029] S31, branching out a new tracking tree through new detection data;

[0030] S32, calculating the Mahalanobis distance of each group of tracking and detection signals to complete threshold detection;

[0031] S33, updating each group of tracking using the detected signal data.

[0032] Further, the calculation formula of the Mahalanobis distance is:

[0033]

[0034] In the formula, is the measurement difference, is the measurement variance matrix; T represents the mark of matrix transposition in mathematics; represents the square of the Mahalanobis distance.

[0035] Further, the evaluation method in the evaluation of each group of tracking by the algorithm includes using the number of likelihood ratios and the corresponding target probability and error probability within the time step.

[0036] Further, the selection method in the selection of the optimal tracking set by calculating the score of each group of tracking is the standard integer linear programming method.

[0037] Further, the calculation formula of the target probability is:

[0038]

[0039] In the formula, is the target probability, is the measurement difference, is the measurement variance matrix, is the measurement variance matrix determinant of is transpose of the matrix, exp() is the exponential of the value in the parentheses.

[0040] Further, the pruning of the tracking by the algorithm and the deletion of useless tracking include the following steps:

[0041] S61, taking the score of each tracking backtracking N time slots;

[0042] S62, compare the scores of the remaining each track under the k-N time slot and the score obtained in the above step, find the track with the maximum difference value;

[0043] S63, finally delete all the branches of the track in the k-N time slot to the k time slot, and prune the track tree.

[0044] The beneficial effects of the present application are:

[0045] 1. The method for tracking and predicting multiple interference sources in the TDMA protocol in the present application, by continuously detecting the average signal strength of all TDMA time slots, the signal interference from external devices is evaluated, the interference signals of different interference sources are identified, and the improved multi-hypothesis tracking algorithm (MHT) is used to detect the interference sources, so as to help the device to judge the mode characteristics of the interference source, predict and avoid the interference time, reduce the communication interference, and improve the transmission efficiency.

[0046] 2. The present application provides a scheme for analyzing and predicting the interference mode according to the signal interference generated between the wireless sensor network devices and the wireless networks, by predicting the interference between the communication of different devices in the network, the device signal transmission time slot can be adjusted to avoid the conflict of the device data transmission, reduce the retransmission caused by the data conflict, and improve the efficiency of data transmission.

[0047] 3. The wireless channel interference detection algorithm proposed in the present application can not only detect, analyze and predict a single interference source, but also can process multiple interference sources at the same time. Compared with the traditional algorithm, the present application efficiently utilizes the monitoring resources in the network and reduces the consumption generated by monitoring the network.

[0048] 4. The present application does not depend on a specific wireless protocol, and only puts forward the requirements meeting the TDMA protocol for the devices in the local network which need to be adjusted to avoid interference. For the interference source, the wireless protocol can be arbitrary, as long as the wireless signal of the interference source has a certain rule and is not completely random, the present application can play a role to avoid interference congestion.

[0049] 5. The key point of the present application is to detect the channel of the signal interference occurring in the wireless network, identify, analyze and predict the interference mode in the channel through the improved multi-hypothesis tracking algorithm (MHT) proposed in the present application, and avoid the conflict of the device data transmission by adjusting the device signal transmission time slot, reduce the retransmission caused by the data conflict, and improve the efficiency of data transmission.

[0050] 6、The improved multi-hypothesis tracking algorithm is used for detecting and analyzing multiple interference sources in a network, and a method for avoiding conflicts in device data transmission is provided by adjusting the device signal transmission time slot according to the algorithm result, the method is not only suitable for interference detection in a wireless sensor network, but also can be applied to network protocol optimization, by introducing active listening and interference detection and data packet retransmission in the network protocol, the device actively optimizes the transmission process, the application can be combined with the network protocol, and there is no need to exist an independent listening point in the network, so that the deployment difficulty is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0052] Fig. 1 is a flow chart of the method for predicting and avoiding wireless communication interference based on listening wireless device communication according to the embodiment of the present application;

[0053] Fig. 2 is a flow chart of the algorithm overall in the method for predicting and avoiding wireless communication interference based on listening wireless device communication according to the embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to further illustrate the embodiments, the present application provides drawings which are part of the disclosure of the present application, mainly used to illustrate the embodiments, and can explain the operating principle of the embodiments in cooperation with the related description of the specification. Those skilled in the art should understand other possible embodiments and advantages of the present application by referring to these contents.

[0055] According to the embodiment of the present application, a method for predicting and avoiding wireless communication interference based on listening wireless device communication is provided.

[0056] The following is an explanation of the abbreviations appearing in the present application:

[0057] Wireless sensor network (Wireless Sensor Networks, WSNs): a kind of Internet of Things, refers to a system of mutual relationship of computing devices, machines, digital machines.

[0058] Time division multiple access (TDMA): A communication technique for sharing a transmission medium (usually a radio frequency band) or network that allows multiple users to use the same frequency at different time slices (time slots).

[0059] Multi-hypothesis tracking (MHT): An algorithm originally used for radar and optical tracking.

[0060] Received signal strength indicator (RSSI): A sign describing the current signal strength received by a device with wireless communication capabilities in a data frame.

[0061] Kalman filter: An algorithm that uses linear system state equations to make optimal estimates of system states from system input and output observations. Since the observations include the effects of noise and interference in the system, the optimal estimates can also be seen as a filtering process.

[0062] Mahalanobis distance: Proposed by Indian statistician P.C. Mahalanobis, it represents the distance between a point and a distribution, and is an effective method for calculating the similarity of two unknown sample sets.

[0063] Maximum weighted independent set (MWIS): A problem in graph theory that seeks the largest independent set of vertices with the maximum weight.

[0064] Integer linear programming (ILP): Integer programming is a type of programming where the variables (all or some) are restricted to be integers. If the variables are restricted to be integers in a linear model, it is called integer linear programming.

[0065] The present application will be further described in conjunction with the accompanying drawings and specific embodiments, as shown in Figs. 1-2 The method for predicting and avoiding wireless communication interference based on the communication of the listening wireless device according to the embodiment of the present application includes the following steps:

[0066] S1. After parameter initialization, the algorithm predicts the time of channel interference occurrence.

[0067] Specifically, after parameter initialization, the algorithm uses the channel model and Kalman filter to predict the time of possible channel interference occurrence.

[0068] Where the algorithm is a loop, after initializing the parameters, the algorithm first uses the channel model and Kalman Filter to predict the time when the channel interference is likely to occur.

[0069] Assume the vector A description of a tracking state at time K is given:

[0070]

[0071] The matrix F is Then the state transition can be described as:

[0072]

[0073] Where W is the channel noise, which is subject to a normal distribution with a mean of 0 and a variance of Q, is the time slot number that the predicted interference at time K will occupy, is the rate of change.

[0074] S2, overflow processing of the results of the channel interference prediction by the algorithm.

[0075] Specifically, the overflow processing of the results of the channel interference prediction by the algorithm includes the following steps:

[0076] S21, detecting the number of time slots existing in the time division multiple access frame;

[0077] S22, verifying whether the output results of the channel interference prediction exceed the number of time slots;

[0078] S23, if the results exceed, performing wrap-around adjustment on the time slots according to the rate of change.

[0079] Where, assuming that there are time slots in a time division multiple access (TDMA) frame, then obviously the output results of the interference prediction should be between 0 and .

[0080] However, the results of the interference prediction can exceed this range, and therefore overflow processing is required, which is relatively simple. Obviously, because the time slot number occupied by the interference changes with the rate of change , it is only necessary to perform wrap-around adjustment on the predicted time slots according to the rate of change.

[0081] The specific adjustment method is as follows:

[0082] When , no overflow processing is required;

[0083] When , then underflow;

[0084] When , then overflow.

[0085] In the formula, is the number of time slots, mod is the remainder function, S, S represents the time slot number of each round of prediction disturbance, represents the adjusted time slot number, cor stands for corrected, and represents the corrected.

[0086] S3, threshold limit detection is performed on the processed channel prediction result, and the detection result is updated and tracked.

[0087] Specifically, the threshold limit detection on the processed channel prediction result and the update and tracking of the detection result include the following steps:

[0088] S31, a new tracking tree is branched out through new detection data;

[0089] Specifically, the "detection data" here refers to the time slots disturbed and recorded in each communication channel.

[0090] The detection method uses the difference between the average disturbance and signal levels (DDSL) of physical signals to determine whether a time slot is disturbed. To calculate the DSL, first, the channel analysis signal (CAS) from physical devices in a time slot needs to be counted, and let U CAS is the average of CAS in the entire time slot, U CAS is the set of CAS signals greater than U in the time slot, I CAS is the set of CAS signals less than U in the time slot, and the number of signals in the two sets is M and P, respectively. Then DSL can be calculated as follows:

[0091]

[0092] When DSL is greater than a certain value λ, it is considered that the time slot is disturbed, and the number of the time slot is recorded, which is the required new detection data. The value of the λ constant is determined by the specific application environment, and is generally between 0.01 and 0.03.

[0093] S32, completing threshold detection by calculating Mahalanobis distance of each group of tracking and detection signals;

[0094] S33, updating each group of tracking by using the detection completed signal data.

[0095] The calculation formula of the Mahalanobis distance is:

[0096]

[0097] In the formula, is a measurement difference, is a measurement variance matrix; T represents a symbol of matrix transposition in mathematics; is a square of the Mahalanobis distance, Here, the square appears only because the calculation result of the right formula is the square of the Mahalanobis distance. If it is required to appear in a non-square format, the square root of both sides of the equation can be modified.

[0098] Specifically, when new detection signal data (time data of channel interference occurrence) is obtained in each round, the new detection data needs to be used to update each tracking to ensure that the next prediction is correct. However, since the results of most channel signal detections have strong randomness and irrelevance, these detection results will not have a significant impact on interference tracking, and even will damage the interference tracking effect. Therefore, the detection input is preprocessed here, and in order to ensure the quality of the detection results of each channel, a threshold detection is performed. The method is to calculate the Mahalanobis distance (Mahalanobis distance) of each tracking and the detected signal.

[0099] S4, evaluating each group of tracking by an algorithm.

[0100] Specifically, the evaluation method in the step of evaluating each group of tracking by an algorithm includes using a log-likelihood ratio (LLR) and a corresponding target probability and an error probability in a time step to score each hypothesis.

[0101] The algorithm for evaluating each group of tracking and giving a score includes the following steps:

[0102] The algorithm uses a log-likelihood ratio (LLR) as a score of tracking. To calculate the log-likelihood ratio, a target probability and an error probability of the Kth time slot are required. For the target probability , the calculation method is: ​

[0103]

[0104] where is the measurement difference, is the measurement variance matrix, refers to the determinant of the measurement variance matrix represents the transpose of the matrix; exp() refers to taking the exponential of the value inside the parentheses.

[0105] The calculation method is:

[0106]

[0107] where is the number of time slots.

[0108] Finally, the log-likelihood ratio (LLR), i.e., the score, can be obtained by multiplying the ratio of and :

[0109]

[0110] where the ∏ symbol indicates multiplication, the above formula is multiplied by 0 to K, and log indicates taking the logarithm.

[0111] S5, calculate the score of each group of tracking, and select the optimal tracking set.

[0112] Specifically, the method of selecting the optimal tracking set in the calculation of the score of each group of tracking is a standard integer linear programming method.

[0113] After calculating the score of each tracking, the algorithm also needs to find the optimal tracking set in the tracking tree as the basis for predicting the occurrence of interference. This problem can be classified as a maximum weighted independent set (MWIS) problem, and a standard integer linear programming (ILP) method is used to solve it. This step often consumes the most computing time.

[0114] wherein selecting the optimal tracking set comprises the following steps:

[0115] First, describe the variables needed in the evaluation process, which means taking one tracking from each of the k time slots to form a tracking set numbered , the score of which is​ Let represent the weight variable corresponding to the tracking set, describing whether the set is selected. Therefore, finding the optimal tracking set is an integer linear programming problem, where the objective function is:

[0116]

[0117] M1, M2...M k It is the number of items tracked within the 1st, 2nd...kth time slots.

[0118] The constraint function is a set of equations consisting of the following conditions: for i u Belongs to 1 to M u If u belongs to 1 to k, then:

[0119]

[0120] By solving this integer linear programming problem, we can obtain the following: The group numbering = 1 is The tracking is the optimal tracking we seek. Solving integer linear programming problems can use existing, publicly available solutions (branch and bound methods), which will not be elaborated upon here.

[0121] S6. Prune the tracking using an algorithm and delete useless tracking.

[0122] Specifically, in order to maintain high computational efficiency and avoid excessive invalid tracking, the algorithm needs to prune the tracking tree, removing useless tracking. At this point, one round of the algorithm's operation ends, and the next round begins.

[0123] Specifically, the steps for pruning tracking are as follows:

[0124] Pruning the tracking uses an N-step backtracking method. For a tracking set with time slot k, the method for deleting useless tracking is as follows:

[0125] Take the score after each tracking back N time slots, that is, the tracking score of each tracking in time slot kN.

[0126] Since the optimal tracking for time slot k has already been calculated, we can first find the score of the optimal tracking for time slot kN in time slot k.

[0127] Then, the scores of the remaining tracks in the kNth time slot are compared with the scores obtained in the above steps, and the track with the largest difference is found.

[0128] Finally, delete all traces that branch off from the kNth time slot to the kth time slot, thus completing the pruning of the trace tree.

[0129] To sum up, by means of the technical scheme of the present application, the method for tracking and predicting multiple interference sources in the TDMA protocol, the average signal strength of all TDMA time slots is continuously detected to evaluate the signal interference from external devices, and the interference signals of different interference sources are identified, and the interference sources are detected according to the improved multi-hypothesis tracking algorithm (MHT), so as to help the device to determine the mode characteristics of the interference source, predict and avoid the time of interference, reduce the communication interference, and improve the transmission efficiency. The present application provides a scheme for analyzing and predicting the interference mode according to the signal interference generated between the wireless sensor network devices and between the wireless networks, and by predicting the interference between the communication of different devices in the network, the device signal transmission time slot can be adjusted to avoid the conflict of the device data transmission, reduce the retransmission caused by the data conflict, and improve the efficiency of data transmission. The wireless channel interference detection algorithm provided by the present application can not only detect, analyze and predict a single interference source, but also can process multiple interference sources at the same time. Compared with the traditional algorithm, the present application efficiently utilizes the monitoring resources in the network and reduces the consumption generated by monitoring the network.

[0130] The present application does not depend on a specific wireless protocol, and only proposes a requirement meeting the TDMA protocol for the device in the local network which needs to be adjusted to avoid interference. For the interference source, the wireless protocol can be arbitrary, as long as the wireless signal of the interference source has a certain rule and is not completely random, the algorithm can play a role to avoid interference congestion. The technical key point of the present application is to detect the signal interference channel in the wireless network through active monitoring and detection, identify, analyze and predict the interference mode in the channel through the improved multi-hypothesis tracking algorithm (MHT) provided by the present application, and avoid the conflict of the device data transmission by adjusting the device signal transmission time slot, reduce the retransmission caused by the data conflict, and improve the efficiency of data transmission. The present application provides an improved multi-hypothesis tracking algorithm to detect and analyze multiple interference sources in the network, and proposes a method for avoiding the conflict of the device data transmission by adjusting the device signal transmission time slot according to the algorithm result. The present application is not only suitable for interference detection in the wireless sensor network, but also can be applied to the optimization of network protocol. By introducing the active monitoring and detection interference and data packet retransmission in the network protocol, the device actively optimizes the transmission process, the present application can be combined with the network protocol, and there is no need to exist an independent monitoring point in the network, which reduces the deployment difficulty.

[0131] The above only describes the preferred embodiments of the present application and should not be used to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for predicting and avoiding wireless communication interference based on listening to wireless device communications, the method comprising: The method comprises the following steps: ​ S1, after parameter initialization, the algorithm predicts the time when channel interference occurs; S2, the algorithm processes the results of channel interference prediction; S3, threshold limit detection is performed on the processed channel prediction results, and the detection results are updated and tracked; S4, each group of tracking is evaluated by the algorithm; S5, the score of each group of tracking is calculated, and the optimal tracking set is selected; S6, the algorithm prunes the tracking and deletes useless tracking; The evaluation method in the step of evaluating each group of tracking by the algorithm comprises using the log-likelihood ratio and the corresponding target probability and error probability within the time step; The calculation formula of the target probability is: ； wherein is the target probability, is the measurement difference, is the measurement variance matrix, is the measurement variance matrix the determinant of, is the transpose of the matrix, exp() is the exponential of the value within the parentheses; The step of pruning the tracking by the algorithm and deleting useless tracking comprises the following steps: S61, the score of each tracking is taken back N time slots; S62, the score of each tracking at the k-N time slot is compared with the score obtained in the above step, and the tracking with the largest difference value is found; S63, finally, all the branches of the tracking at the k-N time slot to the k time slot are deleted, and the pruning of the tracking tree is completed.

2. The method of claim 1, wherein, After parameter initialization, the algorithm uses the channel model and the Kalman filter to predict the time when channel interference may occur.

3. The method of claim 1, wherein, The step of processing the results of channel interference prediction by the algorithm comprises the following steps: S21, the number of time slots existing in the time division multiple access frame is detected; S22, it is verified whether the output result of channel interference prediction exceeds the number of time slots; S23, if the result exceeds, the time slot is adjusted by wrapping according to the change rate.

4. The method of claim 3, wherein, The adjustment mode of the wrap-around adjustment is: When time, then no overflow handling is required; When Time, then underflow; When time, then overflows upwards; In the formula, is the number of time slots, mod is a remainder function, S represents the time slot number of each round of prediction interference, represents the adjusted time slot number.

5. The method of claim 1, wherein, The step of performing threshold limit detection on the processed channel prediction results and updating the tracking results comprises the following steps: S31, a new tracking tree is branched out through new detection data; S32, the Mahalanobis distance between each group of tracking and the detection signal is calculated to complete threshold detection; S33, the tracking groups are updated using the detected signal data.

6. The method of claim 5, wherein, The calculation formula of the Mahalanobis distance is: ； wherein is the measured difference, is the measured covariance matrix; T denotes the sign of the matrix transposition in mathematics; denotes the square of the Mahalanobis distance.

7. The method of claim 6, wherein, The selection method in the step of calculating the score of each group of tracking and selecting the optimal tracking set is the standard integer linear programming method.

Images