Motorcycle unlocking and locking method based on low-frequency radio frequency, Bluetooth and communication protocols

Abstract

The invention provides a motorcycle unlocking and locking method based on low-frequency radio frequency, Bluetooth and communication protocols, and the method comprises the steps: carrying out the phase compensation of a received signal through an iterative optimization equalizer coefficient until the phase distortion is effectively suppressed; a Kalman filtering algorithm is adopted to filter signals of a receiving end, and a filtered high-quality signal is obtained by establishing a state space model of the signals and updating state estimation by using observation values, so that interference of a multipath effect on a pairing process is reduced; in combination with a retransmission mode of a pairing protocol, dynamically adjusting retransmission parameters by tracking the response time and the success rate of a connection request; a Bluetooth pairing process is simulated, and connection success rates under different environment parameter settings are analyzed through a large number of simulation experiments, so that parameter selection of a pairing protocol is optimized.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method for unlocking and locking a motorcycle using a low-frequency radio frequency, Bluetooth and communication protocol. Background Art

[0002] In multipath propagation environments such as underground garages, the reflection of Bluetooth broadcast signals by concrete walls will cause serious multipath effects. The signals of these reflected paths are superimposed on the signals of the direct paths at the receiving end. Due to the propagation distance and medium differences, the phases and amplitudes of the signals of different paths are different, resulting in phase distortion of the composite signal. During the Bluetooth pairing process, devices exchange necessary authentication and key information through broadcast signals, and then establish a connection based on the negotiated parameters. However, due to the phase distortion caused by the multipath effect, it is difficult for the receiving device to accurately extract the effective information of the broadcast signal, resulting in authentication failure or key negotiation errors, thereby affecting the success rate of the pairing process. In order to improve the connection success rate, it is necessary to introduce a retransmission mechanism for multipath effects in the Bluetooth pairing protocol. However, the design of the retransmission mechanism needs to balance the contradiction between the number of retransmissions, delay and power consumption. Too many retransmissions will significantly increase the pairing delay and device power consumption, while insufficient retransmissions cannot effectively suppress multipath interference and ensure the pairing success rate. Therefore, how to adaptively adjust the retransmission strategy according to the state of the multipath channel and maximize the suppression of multipath interference while meeting the pairing delay and power consumption constraints is one of the key technical issues that need to be solved in the design of the Bluetooth pairing protocol. Summary of the invention

[0003] The present invention provides a motorcycle unlocking and locking method using low-frequency radio frequency, Bluetooth and communication protocol, which mainly includes:

[0004] Obtain vehicle location information to determine whether the vehicle is located in an underground garage. If the vehicle is located in an underground garage, obtain the concrete wall reflection parameters in the vehicle's environment and simulate the propagation path and signal strength distribution of the Bluetooth broadcast signal in a multi-path propagation environment to obtain the signal phase distortion characteristics at the receiving end.

[0005] The phase of the received signal is compensated by iteratively optimizing the equalizer coefficient until the phase distortion is effectively suppressed;

[0006] During the Bluetooth pairing process, the bit error rate and signal-to-noise ratio indicators of the receiving end signal are obtained and analyzed, and the number of retransmissions and the retransmission time interval of the pairing protocol are dynamically adjusted. When the bit error rate of the receiving end signal is greater than the bit error rate threshold, the number of retransmissions and the retransmission time interval are increased;

[0007] The Kalman filter algorithm is used to filter the receiving end signal. By establishing the state space model of the signal and using the observed value to update the state estimation, a high-quality signal is obtained after filtering, which reduces the interference of multipath effect on the pairing process.

[0008] Combined with the retransmission method of the pairing protocol, the retransmission parameters are dynamically adjusted by tracking the response time and success rate of the connection request;

[0009] Based on the concrete wall reflection parameters obtained in the vehicle's environment, a mapping relationship between the Bluetooth broadcast signal quality and the vehicle's position is established. According to different environmental characteristics and locations, the change trend of the signal quality is predicted, and the pairing protocol parameters are adjusted in advance.

[0010] The Bluetooth pairing process is simulated, and through a large number of simulation experiments, the connection success rate under different environmental parameter settings is analyzed to optimize the parameter selection of the pairing protocol.

[0011] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0012] The present invention discloses a method for unlocking and locking a motorcycle using low-frequency radio frequency, Bluetooth and communication protocols. By obtaining the vehicle location information to determine whether it is in an underground garage, the multipath propagation of Bluetooth signals is simulated in combination with the concrete wall reflection parameters and the ray tracing algorithm, and the phase distortion characteristics of the receiving end signal are analyzed. A blind equalization algorithm is used for phase compensation, and a Kalman filter is used to reduce multipath effect interference. The pairing protocol retransmission strategy is dynamically adjusted according to the bit error rate and signal-to-noise ratio, a mapping relationship between signal quality and environmental characteristics is established, and the signal change trend is predicted. The pairing parameters are optimized through simulation experiments to improve the connection success rate and stability of Bluetooth devices in complex environments such as underground garages. The present invention effectively solves the problem of Bluetooth pairing difficulties in special environments such as underground garages, and improves the reliability and user experience of the vehicle-mounted Bluetooth system. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The present invention is a flowchart of a method for unlocking and locking a motorcycle using low-frequency radio frequency, Bluetooth and communication protocols.

[0014] Figure 2 A schematic diagram of a motorcycle unlocking and locking method using low frequency radio frequency, Bluetooth and communication protocol according to the present invention.

[0015] Figure 3 It is another schematic diagram of a motorcycle unlocking and locking method using low frequency radio frequency, Bluetooth and communication protocol of the present invention. DETAILED DESCRIPTION

[0016] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0017] like Figure 1-3 In this embodiment, a method for unlocking and locking a motorcycle using a low-frequency radio frequency, Bluetooth and communication protocol may specifically include:

[0018] Step S101, obtain the vehicle location information and determine whether the vehicle is located in an underground garage environment. If the vehicle is located in an underground garage, obtain the concrete wall reflection parameters in the vehicle environment, simulate the propagation path and signal strength distribution of the Bluetooth broadcast signal in a multi-path propagation environment, and obtain the signal phase distortion characteristics of the receiving end.

[0019] The vehicle position coordinates and acceleration data are obtained according to the on-board positioning sensor, and the Euclidean distance value is calculated for the boundary point coordinates stored in the preset underground garage space boundary database. If the distance value is less than the distance threshold, it is determined that the vehicle is in the underground garage space; for the vehicle position coordinates, the wall node coordinates and material parameters are obtained from the preset underground garage three-dimensional structure model, and the triangulation algorithm is used to construct the reflection surface element, and the electromagnetic wave refractive index distribution map of the garage space is obtained through the reflection surface element; according to the on-board Bluetooth signal transmission source coordinates and the receiver coordinates, a ray tracer is used to calculate the propagation distance of the direct path and the reflection path, and the amplitude attenuation and phase delay of the multi-path signal are calculated; for the multi-path signal, a data grouper is used to group the signal path, and the distorted signal path is screened out by the phase delay threshold, and the phase distortion feature vector is established according to the amplitude and delay of the signal path to obtain the received signal distortion coefficient.

[0020] Exemplarily, the vehicle position coordinates, acceleration data, and light intensity data are obtained from the vehicle positioning sensor. According to the three-dimensional grid coordinates and environmental feature data stored in the preset underground garage space boundary database, the support vector machine is used to calculate the Euclidean distance value between the vehicle position coordinates and the boundary point. If the distance value is less than the distance threshold, the vehicle is determined to be in the underground garage space based on the light sensor data and the acceleration data. According to the vehicle positioning coordinates, the surrounding wall node coordinates and material parameters are obtained in the preset underground garage three-dimensional structure model. For the reflection coefficient database of the concrete wall at different incident angles, the triangulation algorithm is used to construct the wall reflection surface element, and the electromagnetic wave refractive index distribution map of the garage space is obtained from the reflection surface element, and the scattering coefficient value of the electromagnetic wave on the concrete surface is calculated in combination with the wall roughness parameter. For the vehicle Bluetooth signal transmission source coordinates and receiver coordinates, combined with the electromagnetic wave refractive index distribution map, a ray tracer is used to calculate the propagation distance of the direct path and the reflected path. According to the free space propagation loss formula and the reflection loss formula, the amplitude attenuation and phase delay of the multipath signal are calculated, and the signal space distribution data of the receiving point is obtained by complex superposition. The free space propagation loss is calculated by the following formula:

[0021]

[0022] , d represents the distance from the transmitting point to the receiving point, and λ represents the wavelength of the signal.

[0023]

[0024] , φ represents phase delay, d represents the distance from the transmitting point to the receiving point, that is, the length of the reflection path, and λ represents the wavelength of the signal. According to the spatial distribution data of the signal at the receiving point, a data grouper is used to group the signal path into primary and secondary signals. The distorted signal path is screened out based on the phase delay threshold. For the amplitude, delay, and incident angle of each signal path, a reverse neural network is used to establish a phase distortion feature vector, and the distortion coefficient of the received signal is obtained from the feature vector. The vehicle position coordinate data is collected by the on-board positioning sensor, including latitude and longitude, and altitude information. The light sensor collects the brightness value of the vehicle's surrounding environment. The brightness value in a normal road environment is above 50,000 lux, and the brightness value in an underground garage is usually less than 200 lux. The acceleration sensor collects the acceleration change of the vehicle when driving on a slope. When the vehicle enters an underground garage, the acceleration data will show obvious downhill characteristics. The three-dimensional structural model of the underground garage is represented by a grid. Each grid point contains spatial coordinates and corresponding material properties. The wall material database records the reflection coefficients of different concrete types at various incident angles. The reflection coefficient of a typical garage concrete wall at an incident angle of 45 degrees is 0.65. The surface roughness parameter depends on the wall construction process. The roughness coefficient of the plastered concrete is about 0.02. The Bluetooth signal transmitter is located on the top of the vehicle, and the receivers are distributed on the garage roof. The transmission power is 4 dBm, and the receiver sensitivity is negative 90 dBm. The signal propagation path is determined by ray tracing. The direct path loss is only related to the propagation distance. The reflection path also needs to consider the additional loss caused by the number of reflections. At a propagation distance of 5 meters, the direct path loss is about 40 dB, and the additional loss of a single reflection is about 3 dB. The phase distortion characteristics are affected by the multipath effect. When there are multiple reflection paths, the propagation delay differences of different paths lead to distortion after signal superposition. The propagation delay is related to the path length and the dielectric constant of the material. The relative dielectric constant of concrete is about 6, which causes the propagation speed of electromagnetic waves to decrease. For a 5-meter reflection path, the delay is about 20 nanoseconds. If the delay difference between the two paths exceeds 10 nanoseconds, it is considered that significant phase distortion occurs. The signal spatial distribution data records the amplitude and phase information of the signals of each path at the receiving point. The superimposed signal characteristics are obtained through complex operations. When the amplitude fluctuation exceeds 6 decibels or the phase deviation exceeds 45 degrees, it is judged to be severely distorted. The input layer of the inverse neural network contains parameters such as signal amplitude, delay, and arrival angle. The hidden layer adopts a 4-layer structure, each layer contains 8 neurons, and the output layer gives a distortion coefficient between 0 and 1. A distortion coefficient greater than 0.8 indicates that the signal quality is significantly degraded. The data grouper divides the received signals into a primary signal group and a secondary signal group according to the order of arrival time. The primary signal group contains the three earliest arriving paths and is used to determine the overall characteristics of the signal. The secondary signal group is used to evaluate the degree of multipath interference.

[0025] Step S102, performing phase compensation on the received signal by iteratively optimizing the equalizer coefficients until the phase distortion is effectively suppressed.

[0026] A sampling signal is obtained from the receiving end, and the in-phase component and the orthogonal component of the sampling signal are extracted through an orthogonal down-converter; an input vector is constructed according to the in-phase component and the orthogonal component, and a convolutional neural network is used to extract phase features from the input vector to obtain an initial phase value and a phase distortion amplitude; for the initial phase value and the phase distortion amplitude, a standard sine wave and a cosine wave are selected as reference signals according to a preset pilot signal library, and a spectral phase subtraction method is used to obtain a phase deviation angle; if the phase deviation angle is greater than the preset phase deviation angle, an equalizer is used to iteratively compensate the signal, and the tap coefficient is updated according to the gradient increment, and the compensated phase value is obtained from the incremental iteration.

[0027] Exemplarily, a received signal with a sampling rate of 24 MHz is obtained from an 8-bit analog-to-digital converter at the receiving end, the in-phase component and the orthogonal component of the received signal are extracted by an orthogonal down-converter, an input vector is constructed for the orthogonal signal sequence according to a 64-tap shift register, and a four-layer convolutional neural network is used to extract the phase feature from the input vector to obtain the initial phase value and the phase distortion amplitude. For the initial phase value and the distortion amplitude, a standard sine wave and a cosine wave are selected from a preset pilot signal library as reference signals, the phase spectrum features of the received signal and the reference signal are calculated according to a 1024-point fast Fourier transform, the spectral phase subtraction method is used to obtain the phase deviation angle, and the signal phase deviation value is reduced by a phase deviation compensator. According to the compensated phase deviation value, four groups of equalizer units composed of complex multipliers and adders are used to iteratively compensate the signal, each group of equalizer units includes 16 tap functions, the tap coefficients are updated based on the gradient increment, the incremental step size parameter is set to 0.01, and the compensated phase value is obtained from the incremental iteration. The signal root mean square error is calculated from the compensated phase value. The original received signal is used as the reference signal. The phase difference of the signal before and after compensation is calculated using a differential accumulator. If the phase difference in the compensation result is greater than the preset half-value, the compensation process is repeated by increasing the number of iterations until the phase difference is less than half-value to obtain the final phase compensation result. The received signal is sampled by 8-bit analog-to-digital conversion to obtain a discrete digital sequence. The sampling rate of 24 MHz ensures complete reconstruction of the signal. A local 90-degree phase difference oscillator is used in the orthogonal down-conversion process to convert the RF signal into a baseband in-phase component and an orthogonal component. In the example, the 2 MHz signal is down-converted to obtain a 0-100 kHz baseband signal. The 64-tap shift register delays the input signal, with each tap interval of 41.67 nanoseconds and a total delay span of 2.67 microseconds, covering the coherence time of the signal to form a signal vector for convolutional neural network processing. The network adopts a four-layer structure, each layer contains 16 convolution kernels, which are used to extract the phase characteristics of the signal. In practical applications, a 2-degree phase deviation can be accurately detected. The preset pilot signal library contains standard sine and cosine waveforms, with a frequency interval of 1 kHz, a frequency range of 0-100 kHz, and a 1024-point Fourier transform to achieve phase spectrum analysis. The frequency resolution is about 98 Hz, and the phase resolution is better than 1 degree. The phase deviation is extracted by complex multiplication. When a 45-degree phase deviation is measured, the compensator outputs a corresponding reverse phase shift. The equalizer adopts a four-group parallel structure, each group contains 16 taps, and the signal amplitude and phase are adjusted by complex multiplication. The gradient increment step size of 0.01 provides sufficient tracking speed while ensuring convergence stability. In actual measurements, 10 iterations can reduce the phase error from 45 degrees to within 5 degrees. The tap coefficient is updated in real time with the phase deviation. When the signal phase changes suddenly, the equalizer completes the coefficient adjustment within 50 microseconds.In the differential accumulation calculation, 100 sampling points are used as the accumulation window to compare the phases of the signals before and after compensation. When the phase difference is greater than 1 / 2, that is, 180 degrees, it is determined that the compensation is insufficient and the number of iterations needs to be increased. In practical applications, for 2 MHz carrier modulation signals, the constellation diagram of the signal before compensation shows severe distortion and divergent point distribution. After 4-5 iterations, the degree of signal point aggregation is significantly improved and the constellation diagram has a clear outline. The compensation process is based on pure digital implementation, with a processing delay of less than 1 millisecond. The computational complexity increases linearly with the number of taps. In actual communication links, even if there is a 90-degree phase jump, the compensator can stabilize the phase error within 5 degrees within 200 microseconds, effectively overcoming the phase distortion caused by multipath propagation.

[0028] Step S103, during the Bluetooth pairing process, obtain and analyze the bit error rate and signal-to-noise ratio indicators of the receiving end signal, dynamically adjust the number of retransmissions and the retransmission time interval of the pairing protocol, and increase the number of retransmissions and the retransmission time interval when the bit error rate of the receiving end signal is greater than the bit error rate threshold.

[0029] The amplitude value of the received signal is obtained according to the Bluetooth receiver, and bit matching is performed. The signal-to-noise ratio is calculated from the root mean square value of the background noise of the received signal to obtain the bit error rate value; for the bit error rate value, the data waveform is used as an input feature, the bit error rate change trend is predicted by a wavelet neural network, and a reference retransmission number and a retransmission time interval are selected from a Bluetooth pairing protocol parameter table of a preset gear to obtain real-time pairing parameters; a Bluetooth pairing request packet is generated according to the real-time pairing parameters, a data frame is constructed, a receiving buffer with a fixed byte size is set to record the data packet sequence number, the number of continuity interruption points is calculated from the constructed sliding window, and a data packet loss count value is obtained; if the bit error rate corresponding to the data packet loss count value exceeds the bit error rate threshold, the retransmission number is increased on the basis of the reference retransmission number, and the time interval is increased on the retransmission time interval, and a periodic retransmission trigger pulse is generated through a hardware timer until the receiving end receives a complete pairing data packet.

[0030] Exemplarily, the amplitude value of the received signal is obtained from the Bluetooth receiver, the 8-byte synchronization sequence is used to perform bit matching on the data packet header, the signal-to-noise ratio is calculated according to the root mean square value of the background noise of the received signal, the received byte sequence is verified by the Hamming code error corrector, and the number of bit errors in every 1000 bits is calculated from the verification result to obtain the bit error rate value. According to the real-time calculated signal-to-noise ratio and bit error rate, a 4096-point data waveform is used as an input feature, and the bit error rate change trend within 100 milliseconds is predicted by a wavelet neural network, and the benchmark retransmission number and retransmission time interval are selected from the Bluetooth pairing protocol parameter table of 10 preset gears to obtain real-time pairing parameters. A Bluetooth pairing request packet is generated according to the pairing parameters, a data frame is constructed according to the 8-byte synchronization header and the 32-byte data segment, a 256-byte receiving buffer is set to record the received data packet sequence number, and the number of continuity interruption points is calculated from the sliding window composed of 8 data packets to obtain the data packet loss count value. The link status is determined by comparing the packet loss count value with the bit error rate threshold. If the bit error rate threshold exceeds one thousandth, the number of retransmissions is increased by 2 times on the current benchmark number of retransmissions, and the retransmission time interval is increased by 5 milliseconds. The hardware timer generates a periodic retransmission trigger pulse until the receiving end receives the complete pairing data packet. The Bluetooth receiving signal enters the analog-to-digital converter after passing through the amplifier. The amplitude value is quantized to 256 levels. The synchronization sequence uses the 01010101 alternating bit mode. The packet header position is located in the data stream through 8-byte shift comparison. The background noise uses the signal variance calculation of the idle time slot to obtain the root mean square value. In actual measurements, the bit error rate increases significantly when the signal-to-noise ratio is lower than 10 decibels. The Hamming code adds 3 check bits for every 7 bits of data. The error position is calculated and corrected at the receiving end through the check matrix. The wavelet neural network uses 4096 sampling data points as input to predict the trend of the bit error rate. In practical applications, the process of the bit error rate rising from 0.1% to 1% is usually completed within 50-100 milliseconds. The prediction sends out an early warning signal 20 milliseconds in advance. The protocol parameter table is divided into 10 gears according to the channel quality. The number of retransmissions varies from 1 to 8 times, and the time interval increases from 5 milliseconds to 50 milliseconds. In the multipath fading channel, the retransmission interval should be greater than the channel coherence time. The data packet adopts a fixed format, the 8-byte synchronization header is used for packet start positioning, the 32-byte data segment contains the device address, pairing parameters, and encryption information, and the 256-byte receiving buffer adopts a circular queue structure to store the sequence number of the received packet. The sliding window size is 8 packets. When the sequence number is detected to be discontinuous, it is recorded as a break point. In actual measurements, the packet loss rate of a normal channel is below 0.1%, and it can reach 5% in harsh environments.The link status judgment uses a bit error rate of one thousandth as the threshold. After exceeding the threshold, the number of retransmissions is increased by 2 times based on the original number of retransmissions, and the time interval is increased by 5 milliseconds to avoid continuous loss of data packets under unfavorable channel conditions. The hardware timer has an accuracy of better than 1 microsecond, generating a periodic retransmission trigger signal, and the retransmission continues until the receiving end passes the checksum to confirm the data integrity. The actual measurement shows that under a 20 dB signal-to-noise ratio, the pairing success rate is increased from 70% to 95% after increasing the number of retransmissions and intervals. Through the comprehensive bit error rate prediction and dynamic retransmission mechanism, the pairing reliability is improved in the scenario of channel quality fluctuation. In actual deployment, it is found that in multipath fading environments such as underground garages and elevator cars, the original pairing success rate is only 40%, which is increased to more than 85% after adopting this solution. The system response delay increases by about 50 milliseconds, which is within the acceptable range for user experience. Test data shows that when the channel quality suddenly changes from good to poor, the prediction mechanism adjusts the transmission parameters 20 milliseconds in advance to effectively avoid pairing interruption.

[0031] Step S104, using the Kalman filter algorithm to filter the receiving end signal, by establishing a state space model of the signal, using the observed value to update the state estimate, to obtain a high-quality signal after filtering, and reduce the interference of the multipath effect on the pairing process.

[0032] The complex sequence of received signals of the signal sampler is obtained, and a state vector including signal amplitude, phase angle and frequency offset is constructed according to the complex sequence of received signals; for the state vector, a sliding window is used to calculate the covariance matrix of the signal and noise, and a state optimal estimation sequence of the Kalman filter is obtained through the covariance matrix; the received signal is corrected according to the state optimal estimation sequence, and multipath reflection is judged by comparing the amplitudes of adjacent sampling points and the phase difference of the corrected received signal; for the multipath reflection judgment result, a recursive least squares filter is used to process the corrected received signal, and if the signal residual is less than the noise baseline, the filtered signal sequence is output.

[0033] Exemplarily, a complex sequence of received signals with a sampling rate of 8 MHz is obtained from a signal sampler, and a 4-dimensional state vector containing signal amplitude, phase angle, and frequency offset is constructed for every 256 sampling points. An 8th-order autoregressive model is used to predict the state variables at the next moment, and a transfer relationship between signal states is established through a delay line, and a 4-dimensional state space description is obtained from the transfer relationship. According to the 4-dimensional state space description, a 64-point sliding window is used to calculate the covariance matrix of the signal and noise, an observation matrix is ​​constructed for the preset multipath channel parameters, and a covariance update value and a filter gain value are calculated by a Kalman filter recursive device, and a state optimal estimation sequence is obtained from the observation data. The received signal is corrected according to the state optimal estimation sequence, and a 16-point sliding average is used to process the corrected signal. The multipath reflection is judged by comparing the amplitudes and phase differences of adjacent sampling points, a 0.1 microsecond delay threshold is set to judge the multipath component, and the interference suppression parameter is determined from the multipath judgment result. For the interference suppression parameters, a 16-tap recursive least squares filter is used to process the received signal, and the step size parameter is set to 0.01. The filter weight is updated. If the signal residual is less than the negative 60 dB noise benchmark, the filtered signal sequence is output, and the signal-to-noise ratio index is obtained from the signal residual. The received signal is converted into a digital sequence by an 8 MHz sampler. Every 256 sampling points constitute an observation cycle. In an indoor multipath environment, the signal amplitude fluctuation range is usually within 20 dB, the phase angle can change up to 90 degrees, and the frequency offset is affected by the Doppler effect at around 100 Hz. These features together constitute a 4-dimensional state vector to describe the instantaneous characteristics of the signal. The 8th-order autoregressive model uses the time correlation of the signal to predict state changes. The prediction error is less than 5% in a stationary scenario and up to 15% in a mobile scenario. In actual applications, it is observed that signal amplitude mutations usually occur within 0.1 milliseconds, and phase mutations are on the order of 0.01 milliseconds. The prediction model needs to track these changes quickly. The 64-point sliding window provides sufficient statistical samples while ensuring real-time performance. The noise covariance reflects the stability of the background noise. The typical indoor environmental noise floor is minus 90 decibels. The diagonal elements of the covariance matrix reflect the noise level of each state component, and the non-diagonal elements characterize the correlation between them. The Kalman gain is adaptively adjusted with the signal-to-noise ratio. The gain value is about 0.3 when the signal-to-noise ratio is 20 decibels. The 16-point sliding averager corresponds to a 2-microsecond time window, which matches the indoor multipath delay. In actual measurements, the delay difference between the main path and the reflected path is usually 0.1-0.5 microseconds, the amplitude difference is 5-15 decibels, and the phase difference can reach 180 degrees. Setting a 0.1-microsecond delay threshold can separate 90% of the multipath components. The 16 taps of the recursive least squares filter cover the main multipath delay range. The step size of 0.01 strikes a balance between convergence speed and stability, and the convergence time is about 50 microseconds.In actual measurements under specific environments, the multipath effect in crowded areas causes signal distortion. The unprocessed signal bit error rate exceeds 5%. After filtering, the bit error rate drops to below 0.1%, the signal-to-noise ratio increases by 15 decibels, the cluster radius of the signal constellation diagram drops from 0.3 to 0.05, and the interval between adjacent points increases from 0.5 to 0.9, indicating that the signal quality has been significantly improved. The underground garage scene test shows that the reflection of the concrete wall causes serious multipath interference. The delay of the original signal extends to 0.8 microseconds. After processing, the main energy is concentrated within 0.2 microseconds, which is equivalent to increasing the channel bandwidth by 4 times, the data transmission rate increases from 2 megabits per second to 8 megabits per second, and the packet error rate drops from 10% to 0.1%.

[0034] Step S105 , combining the retransmission mode of the pairing protocol, dynamically adjusting the retransmission parameters by tracking the response time and success rate of the connection request.

[0035] A pairing request record is obtained from a pairing request buffer, wherein the pairing request record includes timestamps of a request sending time and a request receiving time; a delay distribution feature is extracted through a convolutional neural network according to the timestamp to obtain a link quality value; a benchmark retransmission number and a benchmark retransmission interval are selected from a retransmission parameter configuration table according to the link quality value, a sliding window is used to count the number of successful pairing records, and a pairing success rate is obtained by dividing the number of successful pairing records by the total number of requests; a gear is adjusted according to the pairing success rate and the retransmission parameter, and an increment of the number of retransmissions and an increment of the retransmission interval are calculated through a fuzzy rule base, wherein the fuzzy rule base includes a delay rule and a success rate rule; if the pairing success rate is lower than a pairing success rate threshold, an increment value is superimposed on the basis of the benchmark retransmission value, a retransmission trigger pulse is generated through a hardware timer, and a connection status flag is read from a pairing status register to determine a pairing result.

[0036] Exemplarily, a pairing request record is obtained from a pairing request buffer, and a microsecond timestamp is used to record the request sending time and receiving time. A delay sequence is constructed based on 100 request timestamps. The delay distribution characteristics are extracted through a four-layer convolutional neural network. The average delay value and the delay jitter standard deviation are obtained from the Gaussian distribution fitting, and the link quality value is obtained through weighted combination. According to the link quality value, a reference number of retransmissions and a reference retransmission interval are selected from a five-level retransmission parameter configuration table, and a 1000-millisecond sliding window is used to count the number of successful pairing records. The real-time pairing success rate is obtained by dividing the number of successful records by the total number of requests, and the corresponding retransmission parameter adjustment gear is obtained from a quantitative index table. According to the pairing success rate and the retransmission parameter adjustment gear, a retransmission parameter adjustment value is generated through a fuzzy rule base, and the rule base contains 9 delay rules and 9 success rate rules. The triangular membership is used to calculate the retransmission number increment and the retransmission interval increment, and the upper limit of the retransmission number is set to 8 times, and the upper limit of the retransmission interval is set to 50 milliseconds. According to the calculated retransmission parameter increment, the current retransmission configuration parameters are updated. If the pairing success rate is lower than the 60% threshold, the increment value is superimposed on the baseline retransmission value, and a trigger pulse with a period interval equal to the retransmission time interval is generated by the hardware timer. The connection status flag is read from the pairing status register to determine the pairing result. The pairing request delay record reflects the network transmission quality. The microsecond timestamp provides accurate measurement. In the actual measurement, the normal transmission delay is within 5 milliseconds, the network congestion delay can reach 50 milliseconds, and the delay distribution shows a right-skewed feature. The convolutional neural network extracts delay features through a 4-layer structure. The first layer identifies burst delays, the second layer extracts periodic changes, the third layer detects trends, and the fourth layer integrates features. The five-level retransmission parameter configuration table divides the transmission environment according to the link quality. From the best to the worst, it corresponds to 1, 2, 4, 6, and 8 retransmissions, and the time interval increases from 10 milliseconds to 50 milliseconds. The actual measurement shows that 90% of the pairing requests are completed within 2 retransmissions in the office environment, while 4-6 retransmissions are more common in the underground garage environment. The 1000 millisecond statistical window contains about 100 pairing requests, providing a stable success rate estimate. The fuzzy rule base controls the retransmission parameters through the delay rule and the success rate rule. The delay rule covers the 5-50 millisecond range, and the success rate rule covers the 40%-95% range. When the delay is 20 milliseconds and the success rate is 70%, the rule output recommends increasing the number of retransmissions by 2 times and increasing the retransmission interval by 15 milliseconds. The triangular membership function smoothly transitions in the parameter change range to avoid parameter mutations. The hardware timer provides a retransmission trigger benchmark with a timing accuracy better than 1 microsecond. The pairing status register reflects the connection progress in real time. When the success rate is lower than 60%, the parameter adjustment is triggered. The actual measurement shows that in a high interference environment, the number of retransmissions is increased to 8 times and the interval is extended to 50 milliseconds, and the pairing success rate is increased from 45% to 85%. Through actual scenario testing, the average delay in the office environment is 5 milliseconds, the standard deviation is 2 milliseconds, and the link quality value is 0.9. The benchmark retransmission parameters can achieve a 95% pairing success rate.The actual environment is affected by the flow of people, with an average delay of 15 milliseconds, a standard deviation of 8 milliseconds, and a link quality drop to 0.6. The number of retransmissions needs to be increased to 4 times to maintain an 85% success rate. The underground garage scenario is the most severe, with an average delay of 35 milliseconds, a standard deviation of 20 milliseconds, and a link quality of only 0.3. Even with the maximum retransmission parameters, the success rate is only 75%. Data analysis shows that the delay distribution characteristics are strongly correlated with the pairing success rate. For every 5 millisecond increase in the delay standard deviation, the success rate drops by about 10%. Dynamic adjustment of the retransmission parameters can effectively combat network quality fluctuations. The test found that when the transmission quality deteriorates in a short period of time, rapidly increasing the number of retransmissions is more effective than extending the retransmission interval. In a continuous interference environment, appropriately extending the retransmission interval can provide more stable performance.

[0037] Step S106, based on the obtained concrete wall reflection parameters in the vehicle environment, establish a mapping relationship between the Bluetooth broadcast signal quality and the vehicle position, predict the change trend of the signal quality according to different environmental characteristics and positions, and adjust the pairing protocol parameters in advance.

[0038] The position coordinate data and acceleration data are obtained from the vehicle-mounted inertial navigation unit, and the position coordinate data are used to extract the dielectric constant and loss angle parameters of the current position point; according to the dielectric constant and loss angle parameters, a graph convolutional neural network is used to construct a mapping relationship between position features and environmental parameters to obtain the environmental vector; for the environmental vector, a Gaussian basis function is used to construct a position and signal quality mapping table, and the spatial correlation is calculated through signal strength and bit error rate indicators to obtain a signal quality prediction result; if the signal quality prediction result shows that the signal attenuation exceeds the reference quality threshold, the parameter group of the corresponding attenuation gear is read from the protocol parameter configuration table, the propagation loss is calculated through finite time domain difference, and the number of retransmissions and time interval parameter values ​​in the protocol parameter buffer are updated.

[0039] Exemplarily, centimeter-level position coordinates, velocity vectors and acceleration data are obtained from the vehicle-mounted inertial navigation unit, the dielectric constant, loss angle and surface roughness parameters of the current position point are extracted according to the concrete wall database, and a four-layer graph convolutional neural network is used to construct a mapping relationship between 16-dimensional position features and 32-dimensional environmental parameters, and an environmental vector containing spatial structural features is obtained from the mapping layer. Based on the environmental vector and historical signal data, a Gaussian basis function is used to construct a position and signal quality mapping table, and the spatial correlation is calculated through characteristic indicators composed of signal strength, bit error rate and delay jitter. The reference quality threshold is selected from the five-level signal quality classification table, and the signal quality change in the next 5 seconds is predicted in combination with the vehicle motion trajectory. According to the signal quality prediction results, the garage space is divided into three-dimensional grid units with a grid spacing of 0.5 meters, and the propagation loss of electromagnetic waves between grid units is calculated by finite time domain difference. The multipath components are superimposed according to the reflection coefficient. The signal attenuation parameters are obtained from the grid field strength distribution map, and the propagation loss prediction curve is constructed. According to the propagation loss prediction curve, the retransmission number and time interval parameter values ​​in the paired protocol parameter cache are updated. If the predicted signal attenuation exceeds 90 decibels, the parameter group of the corresponding attenuation gear is read from the protocol parameter configuration table, and the upper limit of the retransmission number is set to 8 times and the upper limit of the retransmission interval is set to 50 milliseconds to obtain the updated configuration. The inertial navigation unit is positioned by fusion of accelerometer and gyroscope. The position accuracy in the underground garage environment is better than 5 cm, the speed measurement error is less than 0.1 meters per second, and the acceleration resolution is 0.01 meters per square second. The concrete wall database records the material parameters under different construction processes. The relative dielectric constant of ordinary concrete is 4-6, the loss tangent value is 0.02-0.05, and the surface roughness is in the range of 0.5-2 mm. The graph convolutional neural network models the garage space as a topological structure of nodes and edges. The 16-dimensional position features include coordinates, speed, acceleration and their historical values. The 32-dimensional environmental parameters describe the geometric and electrical properties of the surrounding walls. The local features are extracted by 4 layers of convolution and then pooled to obtain the global features. The measured accuracy rate reaches 95%. For example, near the corner of the wall, the environmental vector will reflect the superposition characteristics of multiple reflection paths. The Gaussian kernel function maps the spatial position to the signal quality index. The kernel width is set to 3 meters to match the coherence distance of the Bluetooth signal. The signal quality is divided into 5 levels, from level 1 to level 5 corresponding to signal-to-noise ratio greater than 30 decibels, 20-30 decibels, 10-20 decibels, 0-10 decibels, and less than 0 decibels. When the vehicle travels at a constant speed of 15 kilometers per hour, it moves about 21 meters in 5 seconds, and the prediction accuracy exceeds 90%. The 0.5-meter grid spacing provides sufficient spatial resolution within the range of computing resources. The finite time-domain difference calculation uses a time step that is 10 times smaller than the signal wavelength to ensure numerical stability. It was found that the signal attenuation at the corner can reach 30 decibels, while the straight channel is only 15 decibels. The multipath effect is particularly obvious in the corner area, and the reflected signal strength can reach 50% of the direct signal.The test of a typical scenario in an underground garage shows that when a vehicle enters from the entrance ramp, the signal attenuation increases rapidly from 60 decibels to 90 decibels. The prediction algorithm responds 500 milliseconds in advance, adjusting the number of retransmissions from 2 to 6 times, and extending the time interval from 15 milliseconds to 35 milliseconds. When the vehicle enters the signal quality level 5 area, the retransmission parameters reach the maximum value of 8 times and 50 milliseconds, effectively preventing pairing interruption. Experimental data show that it is difficult to cope with the rapidly changing propagation environment relying solely on real-time measurement, and the prediction mechanism significantly improves the adaptability of the protocol. In multi-vehicle scenarios, the environmental feature vector also reflects the occlusion effect, avoiding mutual interference in parameter adjustment. Statistics show that the prediction parameter adjustment scheme reduces the signal interruption rate from 8% to 1.5%, and the average pairing delay from 250 milliseconds to 120 milliseconds.

[0040] Step S107, simulating the Bluetooth pairing process, analyzing the connection success rate under different environmental parameter settings through a large number of simulation experiments, so as to optimize the parameter selection of the pairing protocol.

[0041] The simulated multipath propagation path data is obtained from a ray tracer, a field strength distribution matrix is ​​constructed through the divided space, and a residual convolutional network is used to process the field strength distribution matrix to obtain a spatial propagation feature vector; based on the spatial propagation feature vector, a recursive least squares filter is used to construct a signal quality state space, and a received signal characteristic sequence is output through a state predictor; for the received signal characteristic sequence, successful records are extracted from the pairing log, and a parameter sensitivity value is obtained using a gradient descent method; if the signal quality is lower than the quality threshold, the number of retransmissions is adjusted according to the parameter sensitivity value, and a new parameter configuration is obtained in the parameter space through an exhaustive searcher.

[0042] Exemplarily, the attenuation and delay values ​​of the multipath propagation path are obtained from the ray tracer, and the space is divided into three-dimensional grids with a spacing of 0.5 meters. A field strength distribution matrix containing signal strength, phase offset, and propagation delay is constructed, and spatial features are extracted through a four-layer residual convolutional network. Each layer contains 16 feature maps, and the pooling layer uses maximum pooling to obtain a 64-dimensional spatial propagation feature vector from the output layer. According to the 64-dimensional propagation feature vector, a 32-order recursive least squares filter is used to construct an 8-dimensional signal quality state space, including four observations and four predictions of signal-to-noise ratio, bit error rate, phase distortion, and packet error rate. The filter parameters are updated by a step factor of 0.01, and a received signal characteristic sequence consisting of 1000 sampling points is output from the state predictor. According to the characteristic sequence of the received signal, the successful records in the 10-minute pairing log are extracted, and the gradient descent method is used to optimize the parameters of the number of retransmissions and the time interval. The optimization objective function combines the signal quality and the connection success rate. The parameter gradient is calculated by back propagation, and the learning rate is set to 0.1. When the gradient change rate is less than 0.01, the iteration is stopped, and the parameter sensitivity value is obtained from the optimization curve. According to the parameter sensitivity value, the retransmission configuration in the protocol parameter configuration table is updated. If the signal quality is lower than the negative 90 decibel threshold, the number of retransmissions is increased according to the sensitivity ratio based on the original number of retransmissions, and the time interval is adjusted synchronously. The optimal combination is found in the parameter space through the exhaustive searcher, and the upper limit of the number of retransmissions is limited to 8 times and the upper limit of the time interval is 50 milliseconds to obtain the new parameter configuration. Ray tracing simulates multipath propagation by calculating the signal reflection path. In the underground garage scene, the direct path loss is about 60 decibels, and a single reflection increases the loss by 15 decibels. The 0.5-meter grid spacing provides accurate spatial resolution. The feature map of the four-layer residual convolutional network reflects spatial features at different scales. The first layer captures local structures such as corners, the second layer extracts regional features such as channels, the third layer identifies large-scale features such as parking areas, and the fourth layer integrates global information. The 32-order recursive least squares filter processes the 8-dimensional state space. The signal-to-noise ratio in the observation reflects the channel quality, with a typical value of 0-30 dB. The bit error rate describes the transmission reliability, with a normal range of less than 0.1%. The phase distortion represents multipath interference, usually not exceeding 45 degrees. The packet error rate shows the link stability and should be less than 1%. The step factor of 0.01 strikes a balance between tracking speed and stability. The 10-minute pairing log contains about 600 connection attempts. The gradient descent optimization searches for the optimal configuration in the parameter space. When the signal-to-noise ratio is 20 dB, 3 retransmissions with a 15-ms interval provide a 95% success rate. When the signal-to-noise ratio drops to 10 dB, 6 retransmissions with a 25-ms interval are required to achieve a 90% success rate. During the optimization process, the sensitivity of the number of retransmissions to the success rate is significantly improved when the signal-to-noise ratio is low.The parameter configuration is dynamically adjusted according to the scene characteristics. In open areas, the signal-to-noise ratio is higher than minus 60 decibels, and the basic configuration can be used. The signal-to-noise ratio at the corner drops to minus 75 decibels, and the number of retransmissions needs to be increased to 4 times. The signal-to-noise ratio in narrow channels can be as low as minus 90 decibels, and the number of retransmissions is increased to 8 times for the most effective. Exhaustive search evaluates all combinations within the range of 1-8 retransmissions and 10-50 milliseconds intervals. Experimental data show that in multi-vehicle scenarios, signal quality shows significant spatial correlation, and the pairing behavior of adjacent vehicles will interfere with each other. Through spatial multiplexing optimization, different parameter configurations are used in different areas to avoid parameter conflicts. Statistical results show that compared with fixed parameters, the dynamic optimization scheme reduces the average pairing delay from 200 milliseconds to 80 milliseconds, and the success rate increases from 85% to 97%. In high-speed mobile scenarios, signal quality changes rapidly, and traditional parameter optimization is difficult to respond in time, so the prediction mechanism is particularly important. Data shows that when the vehicle speed reaches 20 kilometers per hour, adjusting parameters 200 milliseconds in advance can prevent 90% of pairing failures, verifying the practical value of simulation optimization. Performance tests show that the entire optimization process runs in less than 10 milliseconds on an ordinary processor, meeting real-time processing requirements.

[0043] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A motorcycle unlocking and locking method using low frequency radio frequency, Bluetooth and communication protocol, characterized in that: The method comprises: Obtain vehicle location information to determine whether the vehicle is located in an underground garage. If the vehicle is located in an underground garage, obtain the concrete wall reflection parameters in the vehicle's environment and simulate the propagation path and signal strength distribution of the Bluetooth broadcast signal in a multi-path propagation environment to obtain the signal phase distortion characteristics at the receiving end. The phase of the received signal is compensated by iteratively optimizing the equalizer coefficient until the phase distortion is effectively suppressed; During the Bluetooth pairing process, the bit error rate and signal-to-noise ratio indicators of the receiving end signal are obtained and analyzed, and the number of retransmissions and the retransmission time interval of the pairing protocol are dynamically adjusted. When the bit error rate of the receiving end signal is greater than the bit error rate threshold, the number of retransmissions and the retransmission time interval are increased; The Kalman filter algorithm is used to filter the receiving end signal. By establishing the state space model of the signal and using the observed value to update the state estimation, a high-quality signal is obtained after filtering, which reduces the interference of multipath effect on the pairing process. Combined with the retransmission method of the pairing protocol, the retransmission parameters are dynamically adjusted by tracking the response time and success rate of the connection request; Based on the concrete wall reflection parameters obtained in the vehicle's environment, a mapping relationship between the Bluetooth broadcast signal quality and the vehicle's position is established. According to different environmental characteristics and locations, the change trend of the signal quality is predicted, and the pairing protocol parameters are adjusted in advance. The Bluetooth pairing process is simulated, and through a large number of simulation experiments, the connection success rate under different environmental parameter settings is analyzed to optimize the parameter selection of the pairing protocol.

2. The method according to claim 1, characterized in that The method of obtaining the vehicle location information and determining whether the vehicle is located in an underground garage environment, and if the vehicle is located in an underground garage, obtaining the concrete wall reflection parameters in the vehicle environment, and simulating the propagation path and signal strength distribution of the Bluetooth broadcast signal in a multi-path propagation environment to obtain the signal phase distortion characteristics at the receiving end includes: The vehicle position coordinates and acceleration data are obtained according to the vehicle positioning sensor, and the Euclidean distance value is calculated for the boundary point coordinates stored in the preset underground garage space boundary database. If the distance value is less than the distance threshold, it is determined that the vehicle is in the underground garage space; According to the vehicle position coordinates, the wall node coordinates and material parameters are obtained from the preset underground garage three-dimensional structure model, and the reflection surface element is constructed by using the triangulation algorithm. The electromagnetic wave refractive index distribution map of the garage space is obtained through the reflection surface element; According to the coordinates of the vehicle Bluetooth signal source and the receiver, a ray tracer is used to calculate the propagation distance of the direct path and the reflected path, and the amplitude attenuation and phase delay of the multipath signal are calculated; For multipath signals, a data grouper is used to group the signal paths, and the distorted signal paths are screened out through a phase delay threshold. A phase distortion feature vector is established according to the amplitude and delay of the signal path to obtain the received signal distortion coefficient.

3. The method according to claim 1, characterized in that: The phase compensation of the received signal is performed by iteratively optimizing the equalizer coefficient until the phase distortion is effectively suppressed, including: Acquire a sampling signal from a receiving end, and extract an in-phase component and an orthogonal component of the sampling signal through an orthogonal down converter; Constructing an input vector according to the in-phase component and the orthogonal component, extracting phase features from the input vector using a convolutional neural network to obtain an initial phase value and a phase distortion amplitude; For the initial phase value and the phase distortion amplitude, a standard sine wave and a cosine wave are selected as reference signals according to a preset pilot signal library, and a spectral phase subtraction method is used to obtain a phase deviation angle; If the phase deviation angle is greater than the preset phase deviation angle, an equalizer is used to iteratively compensate the signal, the tap coefficient is updated according to the gradient increment, and the compensated phase value is obtained from the incremental iteration.

4. The method according to claim 1, characterized in that In the Bluetooth pairing process, the bit error rate and signal-to-noise ratio indicators of the receiving end signal are obtained and analyzed, and the number of retransmissions and the retransmission time interval of the pairing protocol are dynamically adjusted. When the bit error rate of the receiving end signal is greater than the bit error rate threshold, the number of retransmissions and the retransmission time interval are increased, including: The amplitude value of the received signal is obtained according to the Bluetooth receiver, bit matching is performed, and the signal-to-noise ratio is calculated from the root mean square value of the background noise of the received signal to obtain the bit error rate value; According to the bit error rate value, the data waveform is used as an input feature, the bit error rate change trend is predicted by a wavelet neural network, and the benchmark retransmission number and retransmission time interval are selected from the Bluetooth pairing protocol parameter table of the preset gear to obtain the real-time pairing parameters; Generate a Bluetooth pairing request packet according to the real-time pairing parameters, construct a data frame, set a receiving buffer with a fixed byte size to record the data packet sequence number, calculate the number of continuity interruption points from the constructed sliding window, and obtain a data packet loss count value; If the bit error rate corresponding to the packet loss count value exceeds the bit error rate threshold, the number of retransmissions is increased based on the baseline number of retransmissions, and the time interval is increased based on the retransmission time interval, and a periodic retransmission trigger pulse is generated through a hardware timer until the receiving end receives a complete paired data packet.

5. The method according to claim 1, characterized in that The Kalman filter algorithm is used to filter the receiving end signal, and by establishing a state space model of the signal, the state estimation is updated using the observed value to obtain a high-quality signal after filtering, thereby reducing the interference of the multipath effect on the pairing process, including: Obtain a complex sequence of received signals from a signal sampler, and construct a state vector including a signal amplitude, a phase angle, and a frequency offset according to the complex sequence of received signals; For the state vector, a sliding window is used to calculate the covariance matrix of the signal and the noise, and the state optimal estimation sequence of the Kalman filter is obtained through the covariance matrix; Correcting the received signal according to the state optimal estimation sequence, and judging multipath reflection by comparing the amplitudes of adjacent sampling points and the phase difference of the corrected received signal; According to the multipath reflection judgment result, a recursive least squares filter is used to process the corrected received signal. If the signal residual is less than the noise reference, the filtered signal sequence is output.

6. The method according to claim 1, characterized in that The retransmission method combined with the pairing protocol dynamically adjusts the retransmission parameters by tracking the response time and success rate of the connection request, including: Obtain a pairing request record from a pairing request buffer, wherein the pairing request record includes timestamps of a request sending time and a request receiving time; Extracting delay distribution features through a convolutional neural network according to the timestamp to obtain a link quality value; According to the link quality value, a reference retransmission number and a reference retransmission interval are selected from the retransmission parameter configuration table, a sliding window is used to count the number of successful pairing records, and a pairing success rate is obtained by dividing the number of successful pairing records by the total number of requests; According to the pairing success rate and retransmission parameter adjustment gear, the retransmission number increment and the retransmission interval increment are calculated through a fuzzy rule base, and the fuzzy rule base includes a delay rule and a success rate rule; If the pairing success rate is lower than the pairing success rate threshold, an incremental value is added to the reference retransmission value, a retransmission trigger pulse is generated by a hardware timer, and a connection status flag is read from a pairing status register to determine the pairing result.

7. The method according to claim 1, characterized in that The method of establishing a mapping relationship between the Bluetooth broadcast signal quality and the vehicle position based on the concrete wall reflection parameters obtained in the environment where the vehicle is located, predicting the change trend of the signal quality according to different environmental characteristics and positions, and adjusting the pairing protocol parameters in advance includes: Acquire position coordinate data and acceleration data from the vehicle-mounted inertial navigation unit, wherein the position coordinate data is used to extract the dielectric constant and loss angle parameters of the current position point; According to the dielectric constant and loss angle parameters, a graph convolutional neural network is used to construct a mapping relationship between position features and environmental parameters to obtain the environmental vector; For the environment vector, a Gaussian basis function is used to construct a position and signal quality mapping table, and the spatial correlation is calculated through the signal strength and bit error rate indicators to obtain a signal quality prediction result; If the signal quality prediction result shows that the signal attenuation exceeds the reference quality threshold, the parameter group corresponding to the attenuation gear is read from the protocol parameter configuration table, the propagation loss is calculated by finite time domain difference, and the retransmission number and time interval parameter values ​​in the protocol parameter buffer area are updated.

8. The method according to claim 1, characterized in that The Bluetooth pairing process is simulated, and a large number of simulation experiments are performed to analyze the connection success rate under different environmental parameter settings to optimize the parameter selection of the pairing protocol, including: The simulated multipath propagation path data is obtained from the ray tracer, the field intensity distribution matrix is ​​constructed by dividing the space, and the residual convolution network is used to process the field intensity distribution matrix to obtain the spatial propagation feature vector; According to the spatial propagation feature vector, a recursive least squares filter is used to construct a signal quality state space, and a received signal characteristic sequence is output through a state predictor; For the received signal characteristic sequence, successful records are extracted from the pairing log, and the parameter sensitivity value is obtained by using the gradient descent method; If the signal quality is lower than the quality threshold, the number of retransmissions is adjusted according to the parameter sensitivity value, and a new parameter configuration is obtained in the parameter space through an exhaustive searcher.

Images