Shared bicycle track feature extraction method and system based on Bluetooth sniffing

Abstract

The invention provides a shared bicycle track feature extraction method and system based on Bluetooth sniffing. The method comprises the following steps that a Bluetooth sniffing device obtains a Bluetooth signal of a shared bicycle; extracting feature information of the shared bicycle from the Bluetooth signal; positioning the shared bicycle according to the Bluetooth signal, and obtaining position coordinates of the shared bicycle at different time points; obtaining the travel time of the shared bicycle based on the appearance time and the disappearance time of the Bluetooth signal; obtaining the residence time of the shared bicycle in a certain area based on the duration of the Bluetooth signal; and obtaining the speed change and the total distance in the whole riding process based on the position coordinates of the shared bicycle at different time points. The data quality is improved by adopting filtering and error correction algorithms.

Description

Technical Field

[0001] The present invention relates to the technical field of Bluetooth sniffing, and in particular to a method and system for extracting track features of a shared bicycle based on Bluetooth sniffing. Background Art

[0002] In order to understand the user's riding habits of shared bicycles and better operate and maintain shared bicycles, it is necessary to obtain the shared bicycle trajectory features of a large number of users. Currently, obtaining the shared bicycle trajectory features generally includes the following steps: Bluetooth sniffing device deployment: set up special Bluetooth sniffing devices in the area that needs to be monitored. These devices usually have strong signal receiving capabilities and certain processing capabilities. Bluetooth signal scanning: The device continuously scans the surrounding Bluetooth signals and detects the signals emitted by the Bluetooth device of the bicycle. Feature information extraction: extract the feature information related to the bicycle from the captured Bluetooth signal, such as the MAC address of the Bluetooth device, signal strength, specific identification in the broadcast data packet, etc. Data analysis and processing: analyze and process the collected feature information, which may include data cleaning, deduplication, screening and other operations to improve the quality and availability of the data. However, when performing these analyses, the following points should be noted: Bluetooth signals may be interfered by the environment, resulting in inaccurate data. At the same time, different shared bicycle brands are different, and third-party enterprise feature data is required, and the amount of basic data required is large. Summary of the invention

[0003] The present invention solves the problem that the prior art over-relies on third-party data and causes inaccurate analysis when extracting shared bicycle trajectory features, proposes a shared bicycle trajectory feature extraction method and system based on Bluetooth sniffing, and adopts filtering and error correction algorithms to improve data quality.

[0004] In order to achieve the above objectives, the following technical solutions are proposed: A method for extracting track features of a shared bicycle based on Bluetooth sniffing includes the following steps: S1, the Bluetooth sniffing device obtains the Bluetooth signal of the shared bicycle; S2, extracting the characteristic information of the shared bicycle from the Bluetooth signal; S3, positioning the shared bicycle according to the Bluetooth signal, and obtaining the position coordinates of the shared bicycle at different time points; S4, obtaining the travel time of the shared bicycle based on the appearance time and disappearance time of the Bluetooth signal; S5, obtaining the residence time of the shared bicycle in a certain area based on the duration of the Bluetooth signal; S6, based on the position coordinates of the shared bicycle at different time points, obtain the speed change and total distance during the entire riding process.

[0005] To identify bicycle features based on Bluetooth signal analysis, first, it is necessary to collect Bluetooth feature information of bicycles of different brands and models, including the MAC address range of Bluetooth devices, specific Bluetooth broadcast data packet formats, device names, etc., and build a detailed database. Extract key features such as MAC address, device name, and specific fields in the broadcast data packet from the captured Bluetooth signals. Compare and match the extracted features with the features in the established database to determine the brand and model of the bicycle. By continuously collecting Bluetooth signals and combining signal strength and time information, calculate the movement speed and trajectory of the bicycle. According to the appearance and disappearance time of the Bluetooth signal, determine the length of time the bicycle stays in a specific area, such as at a parking spot, commercial area, etc. Record the time when the user uses the bicycle and analyze whether there is a fixed travel time period, such as morning and evening peak hours, weekends, etc. Combined with trajectory and time information, calculate the user's riding speed and single riding distance to understand the user's riding habits.

[0006] Preferably, the S2 specifically comprises the following steps: S21, establishing a Bluetooth feature information database of shared bicycles of various brands and models; S22, extracting Bluetooth features from the captured Bluetooth signal and comparing them with the Bluetooth feature information database to obtain the brand and model of the shared bicycle.

[0007] Preferably, S2 uses a filtering algorithm to extract feature information of shared bicycles.

[0008] Feature extraction is the core link of Bluetooth signal analysis. The present invention adopts filtering and error correction algorithms to improve data quality. The present invention uses signal processing technology to extract parameters that can characterize signal characteristics from preprocessed signals. Time domain features such as signal strength (RSSI), pulse width, periodicity, etc. The frequency domain features can be observed through the spectrum diagram obtained by Fourier transform, and the frequency distribution of the signal can be observed. Statistical features such as mean, variance, skewness, kurtosis, etc. are used to describe the overall characteristics of the signal. Wavelet features use wavelet transform to extract the multi-scale characteristics of the signal. After collecting N (N is an odd number) RSSI values, these N RSSI values ​​are arranged in order of size, and the RSSI value in the middle is taken as the filter output.

[0009] Preferably, the filtering algorithm comprises the following steps: after collecting an odd number of eigenvalues, all eigenvalues ​​are arranged in order from small to large, and the middle eigenvalue is taken as the filtering output.

[0010] Preferably, S3 uses a triangulation positioning algorithm to obtain the position coordinates of the shared bicycle at different time points.

[0011] Trajectory tracking mainly relies on the Bluetooth signal strength (Received Signal Strength Indication, RSSI) and the location information of multiple Bluetooth sniffing devices. When the Bluetooth signal of a shared bicycle is received by multiple sniffing devices, the signal strength will weaken as the distance increases. By measuring the strength difference of the same Bluetooth signal received by different sniffing devices, triangulation or other positioning algorithms can be used to estimate the location of the shared bicycle. By continuously recording these estimated positions over time, the movement trajectory of the shared bicycle can be depicted. For example, in the triangulation positioning algorithm, assume that three Bluetooth sniffing devices A, B, and C are arranged in an area. The Bluetooth signal of the shared bicycle is received by these three devices, and the RSSI values ​​they receive are RSSI_A, RSSI_B, and RSSI_C, respectively. The coordinates of devices A, B, and C are known to be (x_A, y_A), (x_B, y_B), and (x_C, y_C), respectively. According to the signal propagation model, the approximate distance d_A, d_B, and d_C between the shared bicycle and each device can be calculated. Then, through geometric calculations, such as trilateration, the location coordinates (x, y) of the shared bicycle are determined.

[0012] Preferably, an error correction algorithm is used to obtain the position coordinates of the shared bicycle at different time points.

[0013] The specific process of the error correction algorithm is as follows: use the state equation to represent the motion state of the shared bicycle; input the position and time step at the previous moment into the state equation to obtain the predicted value of the shared bicycle position at the current moment, and combine the measured value of the current moment position with the uncertainty of the predicted value to calculate the estimated value as the shared bicycle position coordinates at the current moment.

[0014] The present invention obtains the position coordinates of the shared bicycle at different time points by using an error correction algorithm to improve data quality. The specific process of the error correction algorithm is as follows: The first step is to build a system model: the real position of a shared bike changes over time, but its movement has certain rules and can be described by a state equation. For example, simple uniform linear motion can be expressed as: position = previous moment position + speed × time step.

[0015] The second step is measurement: the position of the shared bicycle is measured through some kind of sensor (such as Bluetooth signal), but the measurement results are noisy.

[0016] The third step is filtering process: (i) Prediction step: According to the system model, the position at the current moment is predicted based on the optimal estimate at the previous moment.

[0017] (ii) Update step: After obtaining the new measurement value, the uncertainty of the predicted value and the measured value is combined and represented by the covariance matrix to calculate a better estimate. The Kalman gain is used to weigh the importance of prediction and measurement.

[0018] Step 4: Keep looping: As new measurements come in, repeat the prediction and update steps to continuously improve the estimate of the object's position.

[0019] Preferably, the travel time of the shared bicycle is the difference between the time when the Bluetooth signal disappears and the time when the Bluetooth signal appears.

[0020] Travel time analysis principle: Travel time analysis is based on the appearance and disappearance time of Bluetooth signals. When the Bluetooth sniffing device first detects the Bluetooth signal of a bicycle, the time at this time is recorded as the start time of the trip. When the Bluetooth signal disappears, the time at this time is recorded as the end time of the trip. By calculating the difference between these two time points, the travel time of the bicycle can be obtained. For example, if the Bluetooth sniffing device detects the bicycle signal at 8 am and loses the signal at 9 am, then the travel time is 1 hour.

[0021] Preferably, the residence time of the shared bicycle in a certain area is the duration during which the Bluetooth signal remains stable at one location and the intensity variation is within a set value.

[0022] The analysis of dwell time is also based on the duration of the Bluetooth signal. When the Bluetooth signal remains stable at a location and the intensity does not change much, the bicycle is considered to be in a dwelling state. The dwell time is the time from when the Bluetooth signal appears stably at that location to when it begins to change significantly (such as signal strength weakening or disappearing). For example, if the Bluetooth signal of a bicycle is stable at a certain location from 8:10 to 8:30, then the dwell time is 20 minutes.

[0023] Preferably, the S6 specifically comprises the following steps: S61, the riding distance is calculated using the Euclidean distance formula between two points; S62, calculating the speed between adjacent points by dividing the distance by the time difference; S63, integrating the speeds and distances between all adjacent points to obtain the total distance and speed changes during the entire riding process.

[0024] Cycling speed and distance analysis principle: In order to calculate the riding speed and distance, first obtain the position coordinates of the bicycle at different time points through trajectory tracking. Assume that the position coordinates of the bicycle at time t1 are (x1, y1), and the position coordinates at time t2 are (x2, y2). The distance can be calculated using the Euclidean distance formula between two points: d = sqrt{(x2-x1)^2+(y2-y1)^2} The speed is calculated by dividing the distance by the time difference: v = frac{d}{t2-t1} By continuously calculating the speed and distance between adjacent time points, the speed change and total distance during the entire riding process can be obtained.

[0025] A shared bicycle trajectory feature extraction system based on Bluetooth sniffing adopts the above-mentioned shared bicycle trajectory feature extraction method based on Bluetooth sniffing, including Alibaba Cloud platform and several queue cluster servers. The feature information of the shared bicycle is sent to the Alibaba Cloud platform through a base station. The Alibaba Cloud platform sends the feature information to the queue cluster server in the form of a queue. The queue cluster server stores the feature information on a hard disk, and a user queries the feature information through a terminal.

[0026] The present invention also proposes a shared bicycle trajectory feature extraction system based on Bluetooth sniffing, which deploys servers in the form of clusters in the physical architecture and adopts the powerful Alibaba Cloud Internet of Things system to ensure the high availability of the system in the physical architecture; the information of shared bicycles entering and leaving the parking site is sent to the Alibaba Cloud platform through the base station, and each queue cluster server shares the data of the cloud platform; the client server sends the message to the queue cluster server in the form of a queue; the client of the queue cluster server stores the message to the hard disk. Users can query the information through the terminal, browser or mobile phone.

[0027] The beneficial effects of the present invention are as follows: the Bluetooth sniffing device obtains the shared bicycle information without the cooperation of the third-party bicycle enterprise, and the cost is low; the filtering and error correction algorithms are used to improve the data quality, and the statistics are more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a flow chart of the method of the present invention.

[0029] Figure 2 It is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0030] Embodiment 1: This embodiment proposes a method for extracting track features of a shared bicycle based on Bluetooth sniffing, which specifically includes the following steps: S1, using several Bluetooth sniffing devices to simultaneously obtain the Bluetooth signals of shared bicycles; S2, extracting feature information from the Bluetooth signal of the shared bicycle; feature extraction is the core link of Bluetooth signal analysis, and the present invention adopts filtering and error correction algorithms to improve data quality. The present invention uses signal processing technology to extract parameters that can characterize signal characteristics from the preprocessed signal. Time domain features such as signal strength (RSSI), pulse width, periodicity, etc. The frequency domain features can be observed through the spectrum obtained by Fourier transform, and the frequency distribution of the signal can be observed. Statistical features such as mean, variance, skewness, kurtosis, etc. are used to describe the overall characteristics of the signal. Wavelet features use wavelet transform to extract the multi-scale characteristics of the signal. After collecting N (N is an odd number) RSSI values, these N RSSI values ​​are arranged in order of size, and the RSSI value in the middle is taken as the filter output.

[0031] S3, positioning the shared bicycle according to the Bluetooth signals simultaneously acquired by several Bluetooth sniffing devices, and obtaining the corresponding position coordinates of the shared bicycle at different time points; The present invention obtains the position coordinates of the shared bicycle at different time points by using an error correction algorithm to improve data quality. The specific process of the error correction algorithm is as follows: The first step is to build a system model: the real position of a shared bike changes over time, but its movement has certain rules and can be described by a state equation. For example, simple uniform linear motion can be expressed as: position = previous moment position + speed × time step.

[0032] The second step is measurement: the position of the shared bicycle is measured through some kind of sensor (such as Bluetooth signal), but the measurement results are noisy.

[0033] The third step is filtering process: (i) Prediction step: According to the system model, the position at the current moment is predicted based on the optimal estimate at the previous moment.

[0034] (ii) Update step: After obtaining the new measurement value, the uncertainty of the predicted value and the measured value is combined and represented by the covariance matrix to calculate a better estimate. The Kalman gain is used to weigh the importance of prediction and measurement.

[0035] Step 4: Keep looping: As new measurements come in, repeat the prediction and update steps to continuously improve the estimate of the object's position.

[0036] S4, calculates the travel time of the shared bicycle based on the appearance time and disappearance time of the Bluetooth signal; Travel time analysis principle: Travel time analysis is based on the appearance and disappearance time of the Bluetooth signal. When the Bluetooth sniffing device first detects the Bluetooth signal of the bicycle, the time at this time is recorded as the start time of the trip. When the Bluetooth signal disappears, the time at this time is recorded as the end time of the trip. By calculating the difference between these two time points, the travel time of the bicycle can be obtained. For example, if the Bluetooth sniffing device detects the bicycle signal at 8 am and loses the signal at 9 am, then the travel time is 1 hour.

[0037] S5, based on the duration of the Bluetooth signal, obtaining the residence time of the shared bicycle in a certain area; The analysis of dwell time is also based on the duration of the Bluetooth signal. When the Bluetooth signal remains stable at a location and the intensity does not change much, the bicycle is considered to be in a dwelling state. The dwell time is the time from when the Bluetooth signal appears stably at that location to when it begins to change significantly (such as signal strength weakening or disappearing). For example, if the Bluetooth signal of a bicycle is stable at a certain location from 8:10 to 8:30, then the dwell time is 20 minutes.

[0038] S6, integrates the corresponding position coordinates of the shared bicycles at different time points to obtain the speed change and total distance during the entire riding process.

[0039] To identify bicycle features based on Bluetooth signal analysis, first, it is necessary to collect Bluetooth feature information of bicycles of different brands and models, including the MAC address range of Bluetooth devices, specific Bluetooth broadcast data packet formats, device names, etc., and build a detailed database. Extract key features such as MAC address, device name, and specific fields in the broadcast data packet from the captured Bluetooth signals. Compare and match the extracted features with the features in the established database to determine the brand and model of the bicycle. By continuously collecting Bluetooth signals and combining signal strength and time information, calculate the movement speed and trajectory of the bicycle. According to the appearance and disappearance time of the Bluetooth signal, determine the length of time the bicycle stays in a specific area, such as at a parking spot, commercial area, etc. Record the time when the user uses the bicycle and analyze whether there is a fixed travel time period, such as morning and evening peak hours, weekends, etc. Combined with trajectory and time information, calculate the user's riding speed and single riding distance to understand the user's riding habits.

[0040] The present invention also proposes a shared bicycle trajectory feature extraction system based on Bluetooth sniffing, referring to Figure 2In the physical architecture, servers are deployed in clusters, and the powerful Alibaba Cloud IoT system is used to ensure the high availability of the system in the physical architecture. The information of shared bicycles entering and leaving the parking station is sent to the Alibaba Cloud platform through the base station, and each queue cluster server shares the data of the cloud platform. The client server sends the message to the queue cluster server in the form of a queue; the client of the queue cluster server stores the message on the hard disk. Users can query the information through the terminal, browser or mobile phone.

[0041] Embodiment 2: This embodiment improves S6 on the basis of Embodiment 1 and proposes a method for extracting track features of shared bicycles based on Bluetooth sniffing, which specifically includes the following steps: S1, using several Bluetooth sniffing devices to simultaneously obtain the Bluetooth signals of shared bicycles; S2, extracting feature information from the Bluetooth signal of the shared bicycle; feature extraction is the core link of Bluetooth signal analysis, and the present invention adopts filtering and error correction algorithms to improve data quality. The present invention uses signal processing technology to extract parameters that can characterize signal characteristics from the preprocessed signal. Time domain features such as signal strength (RSSI), pulse width, periodicity, etc. The frequency domain features can be observed through the spectrum obtained by Fourier transform, and the frequency distribution of the signal can be observed. Statistical features such as mean, variance, skewness, kurtosis, etc. are used to describe the overall characteristics of the signal. Wavelet features use wavelet transform to extract the multi-scale characteristics of the signal. After collecting N (N is an odd number) RSSI values, these N RSSI values ​​are arranged in order of size, and the RSSI value in the middle is taken as the filter output.

[0042] S3, positioning the shared bicycle according to the Bluetooth signals simultaneously acquired by several Bluetooth sniffing devices, and obtaining the corresponding position coordinates of the shared bicycle at different time points; The present invention obtains the position coordinates of the shared bicycle at different time points by using an error correction algorithm to improve data quality. The specific process of the error correction algorithm is as follows: The first step is to build a system model: the real position of a shared bike changes over time, but its movement has certain rules and can be described by a state equation. For example, simple uniform linear motion can be expressed as: position = previous moment position + speed × time step.

[0043] The second step is measurement: the position of the shared bicycle is measured through some kind of sensor (such as Bluetooth signal), but the measurement results are noisy.

[0044] The third step is filtering process: (i) Prediction step: According to the system model, the position at the current moment is predicted based on the optimal estimate at the previous moment.

[0045] (ii) Update step: After obtaining the new measurement value, the uncertainty of the predicted value and the measured value is combined and represented by the covariance matrix to calculate a better estimate. The Kalman gain is used to weigh the importance of prediction and measurement.

[0046] Step 4: Keep looping: As new measurements come in, repeat the prediction and update steps to continuously improve the estimate of the object's position.

[0047] S4, calculates the travel time of the shared bicycle based on the appearance time and disappearance time of the Bluetooth signal; Travel time analysis principle: Travel time analysis is based on the appearance and disappearance time of the Bluetooth signal. When the Bluetooth sniffing device first detects the Bluetooth signal of the bicycle, the time at this time is recorded as the start time of the trip. When the Bluetooth signal disappears, the time at this time is recorded as the end time of the trip. By calculating the difference between these two time points, the travel time of the bicycle can be obtained. For example, if the Bluetooth sniffing device detects the bicycle signal at 8 am and loses the signal at 9 am, then the travel time is 1 hour.

[0048] S5, based on the duration of the Bluetooth signal, obtaining the residence time of the shared bicycle in a certain area; The analysis of dwell time is also based on the duration of the Bluetooth signal. When the Bluetooth signal remains stable at a location and the intensity does not change much, the bicycle is considered to be in a dwelling state. The dwell time is the time from when the Bluetooth signal appears stably at that location to when it begins to change significantly (such as signal strength weakening or disappearing). For example, if the Bluetooth signal of a bicycle is stable at a certain location from 8:10 to 8:30, then the dwell time is 20 minutes.

[0049] S6, integrates the corresponding position coordinates of the shared bicycles at different time points to obtain the speed change and total distance during the entire riding process.

[0050] Cycling speed and distance analysis principle: In order to calculate the riding speed and distance, first obtain the position coordinates of the bicycle at different time points through trajectory tracking. Assume that the position coordinates of the bicycle at time t1 are (x1, y1), and the position coordinates at time t2 are (x2, y2). The distance can be calculated using the Euclidean distance formula between two points: d = sqrt{(x2-x1)^2+(y2-y1)^2} The speed is calculated by dividing the distance by the time difference: v = frac{d}{t2-t1} By continuously calculating the speed and distance between adjacent time points, the speed change and total distance during the entire riding process can be obtained.

[0051] To identify bicycle features based on Bluetooth signal analysis, first, it is necessary to collect Bluetooth feature information of bicycles of different brands and models, including the MAC address range of Bluetooth devices, specific Bluetooth broadcast data packet formats, device names, etc., and build a detailed database. Extract key features such as MAC address, device name, and specific fields in the broadcast data packet from the captured Bluetooth signals. Compare and match the extracted features with the features in the established database to determine the brand and model of the bicycle. By continuously collecting Bluetooth signals and combining signal strength and time information, calculate the movement speed and trajectory of the bicycle. According to the appearance and disappearance time of the Bluetooth signal, determine the length of time the bicycle stays in a specific area, such as at a parking spot, commercial area, etc. Record the time when the user uses the bicycle and analyze whether there is a fixed travel time period, such as morning and evening peak hours, weekends, etc. Combined with trajectory and time information, calculate the user's riding speed and single riding distance to understand the user's riding habits.

[0052] The present invention also proposes a shared bicycle trajectory feature extraction system based on Bluetooth sniffing, referring to Figure 2 In the physical architecture, servers are deployed in clusters, and the powerful Alibaba Cloud IoT system is used to ensure the high availability of the system in the physical architecture. The information of shared bicycles entering and leaving the parking station is sent to the Alibaba Cloud platform through the base station, and each queue cluster server shares the data of the cloud platform. The client server sends the message to the queue cluster server in the form of a queue; the client of the queue cluster server stores the message on the hard disk. Users can query the information through the terminal, browser or mobile phone.

[0053] Embodiment 3: This embodiment improves S3 on the basis of Embodiment 2, and proposes a method for extracting track features of shared bicycles based on Bluetooth sniffing, which specifically includes the following steps: S1, using several Bluetooth sniffing devices to simultaneously obtain the Bluetooth signals of shared bicycles; S2, extracting feature information from the Bluetooth signal of the shared bicycle; feature extraction is the core link of Bluetooth signal analysis, and the present invention adopts filtering and error correction algorithms to improve data quality. The present invention uses signal processing technology to extract parameters that can characterize signal characteristics from the preprocessed signal. Time domain features such as signal strength (RSSI), pulse width, periodicity, etc. The frequency domain features can be observed through the spectrum obtained by Fourier transform, and the frequency distribution of the signal can be observed. Statistical features such as mean, variance, skewness, kurtosis, etc. are used to describe the overall characteristics of the signal. Wavelet features use wavelet transform to extract the multi-scale characteristics of the signal. After collecting N (N is an odd number) RSSI values, these N RSSI values ​​are arranged in order of size, and the RSSI value in the middle is taken as the filter output.

[0054] S3, positioning the shared bicycle according to the Bluetooth signals simultaneously acquired by several Bluetooth sniffing devices, and obtaining the corresponding position coordinates of the shared bicycle at different time points; The present invention obtains the position coordinates of the shared bicycle at different time points by using an error correction algorithm to improve data quality. The specific process of the error correction algorithm is as follows: The first step is to build a system model: the real position of a shared bike changes over time, but its movement has certain rules and can be described by a state equation. For example, simple uniform linear motion can be expressed as: position = previous moment position + speed × time step.

[0055] The second step is measurement: the position of the shared bicycle is measured through some kind of sensor (such as Bluetooth signal), but the measurement results are noisy.

[0056] The third step is filtering process: (i) Prediction step: According to the system model, the position at the current moment is predicted based on the optimal estimate at the previous moment.

[0057] (ii) Update step: After obtaining the new measurement value, the uncertainty of the predicted value and the measured value is combined and represented by the covariance matrix to calculate a better estimate. The Kalman gain is used to weigh the importance of prediction and measurement.

[0058] Step 4: Keep looping: As new measurements come in, repeat the prediction and update steps to continuously improve the estimate of the object's position.

[0059] Trajectory tracking mainly relies on the Bluetooth signal strength (Received Signal Strength Indication, RSSI) and the location information of multiple Bluetooth sniffing devices. When the Bluetooth signal of a shared bicycle is received by multiple sniffing devices, the signal strength will weaken as the distance increases. By measuring the strength difference of the same Bluetooth signal received by different sniffing devices, triangulation or other positioning algorithms can be used to estimate the location of the shared bicycle. By continuously recording these estimated positions over time, the movement trajectory of the shared bicycle can be depicted. For example, in the triangulation positioning algorithm, assume that three Bluetooth sniffing devices A, B, and C are arranged in an area. The Bluetooth signal of the shared bicycle is received by these three devices, and the RSSI values ​​they receive are RSSI_A, RSSI_B, and RSSI_C, respectively. The coordinates of devices A, B, and C are known to be (x_A, y_A), (x_B, y_B), and (x_C, y_C), respectively. According to the signal propagation model, the approximate distance d_A, d_B, and d_C between the shared bicycle and each device can be calculated. Then, through geometric calculations, such as trilateration, the location coordinates (x, y) of the shared bicycle are determined.

[0060] S4, calculates the travel time of the shared bicycle based on the appearance time and disappearance time of the Bluetooth signal; Travel time analysis principle: Travel time analysis is based on the appearance and disappearance time of the Bluetooth signal. When the Bluetooth sniffing device first detects the Bluetooth signal of the bicycle, the time at this time is recorded as the start time of the trip. When the Bluetooth signal disappears, the time at this time is recorded as the end time of the trip. By calculating the difference between these two time points, the travel time of the bicycle can be obtained. For example, if the Bluetooth sniffing device detects the bicycle signal at 8 am and loses the signal at 9 am, then the travel time is 1 hour.

[0061] S5, based on the duration of the Bluetooth signal, obtaining the residence time of the shared bicycle in a certain area; The analysis of dwell time is also based on the duration of the Bluetooth signal. When the Bluetooth signal remains stable at a location and the intensity does not change much, the bicycle is considered to be in a dwelling state. The dwell time is the time from when the Bluetooth signal appears stably at that location to when it begins to change significantly (such as signal strength weakening or disappearing). For example, if the Bluetooth signal of a bicycle is stable at a certain location from 8:10 to 8:30, then the dwell time is 20 minutes.

[0062] S6, integrates the corresponding position coordinates of the shared bicycles at different time points to obtain the speed change and total distance during the entire riding process.

[0063] Cycling speed and distance analysis principle: In order to calculate the riding speed and distance, first obtain the position coordinates of the bicycle at different time points through trajectory tracking. Assume that the position coordinates of the bicycle at time t1 are (x1, y1), and the position coordinates at time t2 are (x2, y2). The distance can be calculated using the Euclidean distance formula between two points: d = sqrt{(x2-x1)^2+(y2-y1)^2} The speed is calculated by dividing the distance by the time difference: v = frac{d}{t2-t1} By continuously calculating the speed and distance between adjacent time points, the speed change and total distance during the entire riding process can be obtained.

[0064] To identify bicycle features based on Bluetooth signal analysis, first, it is necessary to collect Bluetooth feature information of bicycles of different brands and models, including the MAC address range of Bluetooth devices, specific Bluetooth broadcast data packet formats, device names, etc., and build a detailed database. Extract key features such as MAC address, device name, and specific fields in the broadcast data packet from the captured Bluetooth signals. Compare and match the extracted features with the features in the established database to determine the brand and model of the bicycle. By continuously collecting Bluetooth signals and combining signal strength and time information, calculate the movement speed and trajectory of the bicycle. According to the appearance and disappearance time of the Bluetooth signal, determine the length of time the bicycle stays in a specific area, such as at a parking spot, commercial area, etc. Record the time when the user uses the bicycle and analyze whether there is a fixed travel time period, such as morning and evening peak hours, weekends, etc. Combined with trajectory and time information, calculate the user's riding speed and single riding distance to understand the user's riding habits.

[0065] The present invention also proposes a shared bicycle trajectory feature extraction system based on Bluetooth sniffing, referring to Figure 2 In the physical architecture, servers are deployed in clusters, and the powerful Alibaba Cloud IoT system is used to ensure the high availability of the system in the physical architecture. The information of shared bicycles entering and leaving the parking station is sent to the Alibaba Cloud platform through the base station, and each queue cluster server shares the data of the cloud platform. The client server sends the message to the queue cluster server in the form of a queue; the client of the queue cluster server stores the message on the hard disk. Users can query the information through the terminal, browser or mobile phone.

Claims

1. A method for extracting shared bicycle trajectory features based on Bluetooth sniffing, characterized in that: The following steps are involved: S1, the Bluetooth sniffing device obtains the Bluetooth signal of the shared bicycle; S2, extracting the characteristic information of the shared bicycle from the Bluetooth signal; S3, positioning the shared bicycle according to the Bluetooth signal, and obtaining the position coordinates of the shared bicycle at different time points; S4, obtaining the travel time of the shared bicycle based on the appearance time and disappearance time of the Bluetooth signal; S5, obtaining the residence time of the shared bicycle in a certain area based on the duration of the Bluetooth signal; S6, based on the position coordinates of the shared bicycle at different time points, obtain the speed change and total distance during the entire riding process.

2. According to claim 1, a method for extracting shared bicycle trajectory features based on Bluetooth sniffing is characterized in that: The S2 specifically includes the following steps: S21, establishing a Bluetooth feature information database of shared bicycles of various brands and models; S22, extracting Bluetooth features from the captured Bluetooth signal and comparing them with the Bluetooth feature information database to obtain the brand and model of the shared bicycle.

3. According to claim 1, a method for extracting shared bicycle trajectory features based on Bluetooth sniffing is characterized in that: The S2 uses a filtering algorithm to extract feature information of the shared bicycle.

4. According to claim 3, a method for extracting shared bicycle trajectory features based on Bluetooth sniffing is characterized in that: The filtering algorithm comprises the following steps: after collecting an odd number of eigenvalues, all eigenvalues ​​are arranged in order from small to large, and the middle eigenvalue is taken as the filtering output.

5. According to claim 1, a method for extracting shared bicycle trajectory features based on Bluetooth sniffing is characterized in that: The S3 uses a triangulation positioning algorithm to obtain the position coordinates of the shared bicycle at different time points.

6. According to claim 5, a method for extracting shared bicycle trajectory features based on Bluetooth sniffing is characterized in that: An error correction algorithm is used to obtain the location coordinates of shared bicycles at different time points.

7. A method for extracting shared bicycle trajectory features based on Bluetooth sniffing according to any one of claims 1 to 6, characterized in that: The travel time of the shared bicycle is the difference between the time when the Bluetooth signal disappears and the time when the Bluetooth signal appears.

8. A method for extracting shared bicycle trajectory features based on Bluetooth sniffing according to any one of claims 1 to 6, characterized in that: The residence time of the shared bicycle in a certain area is the duration during which the Bluetooth signal remains stable at one location and the intensity variation is within a set value.

9. A method for extracting shared bicycle trajectory features based on Bluetooth sniffing according to any one of claims 1 to 6, characterized in that: The S6 specifically comprises the following steps: S61, the riding distance is calculated using the Euclidean distance formula between two points; S62, calculating the speed between adjacent points by dividing the distance by the time difference; S63, integrating the speeds and distances between all adjacent points to obtain the total distance and speed changes during the entire riding process.

10. A shared bicycle trajectory feature extraction system based on Bluetooth sniffing, using the shared bicycle trajectory feature extraction method based on Bluetooth sniffing according to claim 1, characterized in that: It includes the Alibaba Cloud platform and several queue cluster servers. The characteristic information of the shared bicycles is sent to the Alibaba Cloud platform through the base station. The Alibaba Cloud platform sends the characteristic information to the queue cluster server in the form of a queue. The queue cluster server stores the characteristic information on the hard disk, and the user queries the characteristic information through the terminal.

Images