Data packet transmission control method and device for low-power dual-mode Bluetooth chip

Abstract

The invention relates to the technical field of data packet transmission, and discloses a data packet transmission control method and device for a low-power dual-mode Bluetooth chip. The method comprises: obtaining signal strength parameters, power consumption parameters, transmission rate parameters and environmental parameters of the dual-mode Bluetooth chip, and inputting them into a deep time convolution network for timing feature extraction to obtain a transmission mode switching judgment result; calculating transmission state evaluation data; performing hierarchical adjustment on the chip transmission power, and using a nano-ampere-level LDO circuit to control standby power consumption to obtain dynamic power control parameters; performing compression encoding on the data packet to obtain dual-mode transmission protocol configuration parameters; performing timing sampling to construct a dual-mode state feature sequence; calculating the data packet delivery success rate, end-to-end transmission delay and total system power consumption, and optimizing the transmission control parameters through double-layer deep reinforcement learning to obtain target transmission control parameters. The invention realizes the optimal balance between the transmission performance and power consumption of the dual-mode Bluetooth chip.

Description

Technical Field

[0001] The present invention relates to the technical field of data packet transmission, and in particular to a data packet transmission control method and device for a low-power dual-mode Bluetooth chip. Background Art

[0002] Traditional Bluetooth chips usually work in a single mode, which makes it difficult to balance transmission performance and power consumption requirements in complex application environments, resulting in problems such as reduced transmission reliability and energy waste.

[0003] The dual-mode Bluetooth chips currently on the market often use fixed threshold judgment or simple heuristic strategies when switching modes, lacking in-depth analysis and optimization of real-time transmission status. This method cannot accurately perceive environmental changes and transmission requirements, easily causing unnecessary mode switching, affecting transmission efficiency and system stability. Summary of the invention

[0004] The present invention provides a data packet transmission control method and device for a low-power dual-mode Bluetooth chip, and the present invention achieves an optimal balance between the transmission performance and power consumption of the dual-mode Bluetooth chip.

[0005] In a first aspect, the present invention provides a data packet transmission control method for a low-power dual-mode Bluetooth chip, the data packet transmission control method for the low-power dual-mode Bluetooth chip comprising:

[0006] Obtain the signal strength parameters, power consumption parameters, transmission rate parameters and environmental parameters of the dual-mode Bluetooth chip, and input them into the deep time convolution network for time series feature extraction to obtain the transmission mode switching judgment result;

[0007] Calculating transmission state evaluation data according to the transmission mode switching determination result, wherein the transmission state evaluation data includes: packet loss rate, transmission delay and real-time power consumption;

[0008] Based on the transmission status evaluation data, the chip transmission power is adjusted in stages, and the standby power consumption is controlled by using a nano-ampere level LDO circuit to obtain a dynamic power control parameter;

[0009] The radio frequency sensitivity threshold is set according to the dynamic power control parameter, and the data packet is compressed and encoded to obtain the dual-mode transmission protocol configuration parameter;

[0010] Performing timing sampling based on the dual-mode transmission protocol configuration parameters, calculating the packet time interval, signal strength change rate and link bit error rate, and constructing a dual-mode state feature sequence;

[0011] The data packet delivery success rate, end-to-end transmission delay and total system power consumption are calculated based on the dual-mode state feature sequence, and the transmission control parameters are optimized through double-layer deep reinforcement learning to obtain the target transmission control parameters.

[0012] In a second aspect, the present invention provides a data packet transmission control device for a low-power dual-mode Bluetooth chip, the data packet transmission control device for the low-power dual-mode Bluetooth chip comprising:

[0013] The acquisition module is used to obtain the signal strength parameters, power consumption parameters, transmission rate parameters and environmental parameters of the dual-mode Bluetooth chip, and input them into the deep time convolution network for time series feature extraction to obtain the transmission mode switching judgment result;

[0014] A calculation module, used to calculate transmission state evaluation data according to the transmission mode switching determination result, wherein the transmission state evaluation data includes: packet loss rate, transmission delay and real-time power consumption;

[0015] An adjustment module, used to perform hierarchical adjustment on the chip transmission power based on the transmission status evaluation data, and to control the standby power consumption by using a nanoampere-level LDO circuit to obtain a dynamic power control parameter;

[0016] An encoding module, used to set a radio frequency sensitivity threshold according to the dynamic power control parameter, and compress and encode the data packet to obtain a dual-mode transmission protocol configuration parameter;

[0017] A sampling module, used to perform timing sampling based on the dual-mode transmission protocol configuration parameters, calculate the packet time interval, signal strength change rate and link bit error rate, and construct a dual-mode state feature sequence;

[0018] The optimization module is used to calculate the data packet delivery success rate, end-to-end transmission delay and total system power consumption based on the dual-mode state feature sequence, and optimize the transmission control parameters through double-layer deep reinforcement learning to obtain the target transmission control parameters.

[0019] The third aspect of the present invention provides a computer device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned data packet transmission control method of the low-power dual-mode Bluetooth chip.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for controlling data packet transmission of a low-power dual-mode Bluetooth chip.

[0021] In the technical solution provided by the present invention, the timing feature extraction is performed by a deep time convolutional network, and the multi-head attention mechanism is combined to realize the accurate perception of the transmission state of the dual-mode Bluetooth chip and the mode switching judgment, so that the mode switching is more accurate and efficient. Based on the dynamic power control scheme of the nano-ampere LDO circuit, with the efficient conversion of the DC / DC circuit, the power consumption of the chip is significantly reduced, especially the current consumption in the standby state is reduced to the order of 2μV. The segmented compression coding strategy is adopted to optimize the compression ratio of 2:1 and 4:1 for the control field and the data field, respectively, to reduce the protocol overhead and improve the data transmission efficiency. Through the double-layer deep reinforcement learning framework, the adaptive optimization of the transmission parameters is realized, in which the classic Bluetooth mode and the low power mode adopt different update strategies respectively, so that the parameter optimization is more accurate. The anomaly detection mechanism based on CRC check is designed, and the comprehensive evaluation of the signal strength change rate and the link bit error rate is combined to enhance the anti-interference ability and transmission reliability of the system. By dynamically adjusting the RF sensitivity threshold, the intelligent adjustment of the receiving sensitivity is realized, and the signal reception quality is improved. The experience playback mechanism based on the hierarchical storage structure improves the training efficiency of deep reinforcement learning and accelerates the response speed of the system to environmental changes. By adopting a multi-dimensional reward function evaluation indicator and comprehensively considering the transmission power, link quality and bit error rate, the optimal balance between transmission performance and power consumption is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0023] Figure 1 A schematic diagram of the steps of a method for controlling data packet transmission of a low-power dual-mode Bluetooth chip in an embodiment of the present invention;

[0024] Figure 2 It is a structural schematic diagram of a data packet transmission control device of a low-power dual-mode Bluetooth chip in an embodiment of the present invention;

[0025] Figure 3 It is a schematic block diagram of the structure of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The embodiment of the present invention provides a data packet transmission control method and device for a low-power dual-mode Bluetooth chip. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the data packet transmission control method of the low-power dual-mode Bluetooth chip in the embodiment of the present invention includes:

[0028] Step S1, obtaining the signal strength parameters, power consumption parameters, transmission rate parameters and environmental parameters of the dual-mode Bluetooth chip, and inputting them into the deep time convolution network to extract the timing features, and obtaining the transmission mode switching determination result;

[0029] It is understandable that the execution subject of the present invention may be a data packet transmission control device of a low-power dual-mode Bluetooth chip, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0030] Specifically, the signal receiving module of the dual-mode Bluetooth chip is used to obtain the received signal strength and link quality indicators such as RSSI (received signal strength indication) and link bit error rate (BER) in real time. These data reflect the current communication quality, and are filtered by a bandpass filter to remove the interference of high-frequency and low-frequency noise, and extract effective information about the signal strength. The filtered signal strength data constitutes a signal strength parameter vector, which contains the effective signal strength information received at different time points. In terms of the collection of power consumption parameters, attention is paid to the standby current and working current of the dual-mode Bluetooth chip in different working modes. The standby current refers to the current consumption when the chip is in an idle state, while the working current is the current consumed during the actual transmission process. In order to accurately monitor these two parameters, current detection is performed through a nano-ampere LDO (low-dropout linear regulator) circuit. The LDO circuit can efficiently provide low-power power, and its accuracy is sufficient to capture tiny current changes. The standby current and working current data obtained by this circuit form a power consumption parameter vector, which provides the power consumption characteristics of the chip in different working modes. For the collection of transmission rate parameters, attention is paid to the throughput and transmission delay of the dual-mode Bluetooth chip in the actual transmission process. Throughput reflects the amount of data successfully transmitted per unit time, while transmission delay is the time from the sender initiating a data packet to the receiver successfully receiving the data packet. In order to count these parameters, the sliding window method is used to collect and calculate the data in real time to obtain a stable transmission rate parameter vector. The size of the sliding window is adjusted according to the real-time requirements of the system to ensure that the acquired data is representative in time while avoiding excessive lag. The signal strength parameter vector, power consumption parameter vector and transmission rate parameter vector are input into the input layer of the deep temporal convolutional network, and the input data is standardized by batch normalization to eliminate the problem of different scales in the data and obtain a standardized feature matrix. The standardized feature matrix is ​​input into the first temporal convolutional layer of the deep temporal convolutional network for processing. The convolution kernel size of the first temporal convolutional layer is 1×3, the step size is 1, the padding is 1, and the ReLU activation function is used. The function of the convolutional layer is to extract local features of the input time series data and capture the temporal change trend of signal strength, power consumption and transmission rate. After convolution processing, the first feature map is obtained, which represents the local features of the input data in time series. The first feature map is input into the dilated convolution layer of the deep temporal convolutional network for processing. The dilation rate of the dilated convolution layer is 2, the convolution kernel size is 1×3, and batch normalization and ReLU activation functions are used for processing. Dilated convolution captures temporal features over a longer time range by increasing the receptive field of the convolution kernel, improving the model's ability to model long-term dependencies. After processing with dilated convolution, a temporal feature map is obtained, which represents the feature expression of the input data over a longer time scale. The temporal feature map is processed by a multi-head attention layer.The multi-head attention mechanism captures time series feature information of different dimensions by calculating multiple attention heads in parallel. The dimension of each attention head is 64, and there are 8 attention heads in total. The multi-head attention layer assigns different weights to different parts of the input data, so as to focus on the most relevant time period or data features and reduce the impact of irrelevant information. The attention weighted features are input into the fully connected layer for feature dimensionality reduction. Through the fully connected layer, the network maps the extracted high-dimensional features to low-dimensional representations, and uses the Softmax classifier to calculate the probability values ​​of the classic Bluetooth mode and the low-power Bluetooth mode. The Softmax classifier determines which transmission mode the current system should switch to by mapping the input features to a probability distribution. According to the probability value, the final transmission mode switching decision result is determined.

[0031] Step S2, calculating transmission status evaluation data according to the transmission mode switching determination result, the transmission status evaluation data including: packet loss rate, transmission delay and real-time power consumption;

[0032] Specifically, according to the judgment result of the transmission mode switching, it is divided into a classic Bluetooth mode parameter matrix and a low-power Bluetooth mode parameter matrix, and these two matrices represent the transmission characteristics of the two modes respectively. By combining the relevant parameters in the two matrices, a hybrid mode feature vector is obtained to describe the overall transmission characteristics of the current system. The hybrid mode feature vector is segmented in time. In order to better capture the timing changes in the transmission process, the hybrid mode feature vector is segmented according to a certain time window to form a data packet statistical matrix, which records the transmission state and various characteristics in each time window. The data packet statistical matrix is ​​classified by transmission state, and divided into three main states according to different situations in the transmission process: normal transmission, packet loss retransmission and transmission timeout. These states can reflect the problems encountered in different transmission processes. For example, normal transmission means that the data packet can be transmitted smoothly, packet loss retransmission means that there is a data packet loss and needs to be retransmitted, and transmission timeout means that the data packet fails to complete the transmission within the predetermined time. Through the state counter, different states are cumulatively counted to obtain the frequency of occurrence of each state. By analyzing the counting results, it is evaluated which problems (such as packet loss or timeout) have a greater impact on transmission performance under different transmission modes. Based on the state counting results, the packet loss rate is calculated. This indicator represents the ratio of the number of packets lost in a unit time window to the total number of transmitted packets. The packet loss rate is an important indicator for measuring the quality of Bluetooth communication. A high packet loss rate means that the communication environment is poor or there is signal interference during the transmission process. By calculating the packet loss rate, the transmission strategy is dynamically adjusted to reduce data loss and improve transmission stability. At the same time, in order to evaluate the delay during the transmission process, the sending timestamp and receiving timestamp of each data packet are recorded. By calculating the difference between these timestamps, the transmission delay of each data packet is obtained. The transmission delay reflects the time it takes for the data packet to go from the sender to the receiver and is an important indicator for measuring transmission efficiency. The transmission delay sequence is input into the median filter for outlier filtering, and the extreme values ​​caused by network jitter or other interference factors are removed to obtain the transmission delay. The signal transmission power in the mixed mode feature vector is sampled, and the power consumption is evaluated in combination with the standby current value of the nano-ampere LDO circuit. The standby current is the current consumed by the chip in a low-power state. It is usually small, but it still needs to be accurately measured. The power consumption sampling sequence is constructed through the power consumption sampling data, and then the product of the sequence and the current working voltage is accumulated to calculate the power consumption value at each moment. These calculation results are converted through the power consumption classification mapping table to obtain accurate real-time power consumption data. Based on the packet loss rate, transmission delay and real-time power consumption, the transmission status evaluation data is generated to comprehensively reflect the key performance indicators in the current transmission process, including data transmission quality, timeliness and energy efficiency.By real-time monitoring and evaluation of these indicators, dynamic adjustments are made according to the actual transmission status to optimize the various parameters of Bluetooth communication, ensuring the lowest power consumption and optimal transmission performance under different environmental and load conditions.

[0033] Step S3: Based on the transmission status evaluation data, the chip transmission power is adjusted in stages, and the standby power consumption is controlled by using a nanoampere-level LDO circuit to obtain a dynamic power control parameter;

[0034] Specifically, according to the packet loss rate and transmission delay obtained in the transmission state evaluation data, threshold analysis is performed on these data. By setting the loss rate threshold and the delay threshold, it is determined whether the transmission quality of the current network meets the predetermined performance standard. If the packet loss rate or the transmission delay exceeds the set threshold, it indicates that the communication quality is poor and the transmission performance needs to be improved by adjusting the transmit power. By performing threshold analysis on the loss rate and delay data, a power adjustment reference matrix is ​​obtained, which describes how to optimize the transmission effect by adjusting the transmit power under different transmission states. The power adjustment reference matrix is ​​divided into two main transmission states according to different transmit power levels: a high-power transmission state and a low-power transmission state. These two transmission states correspond to different communication environments and quality requirements, respectively. When the packet loss rate is high or the transmission delay is long, the high-power transmission state is switched to enhance signal coverage and transmission stability; when the signal is strong and the transmission environment is good, the low-power transmission state is switched to reduce power consumption and extend the service life of the device. Through division, the transmit power is flexibly adjusted to adapt to different communication needs and minimize power consumption. On this basis, a power classification mapping vector is obtained, which specifically describes the power setting under each transmission mode. Based on the power classification mapping vector, current detection is performed through a nano-ampere LDO circuit. The LDO circuit (low dropout linear regulator) has the characteristics of high precision and low power consumption. It can provide a stable power supply for the chip and detect small current changes. In this process, the operating voltage of the LDO circuit is set to 1.0V, and the standby current threshold is 2μV. Through current detection, the standby power consumption of the chip is monitored in real time, and the standby power consumption control sequence is obtained, and the power consumption data of the chip under different states is recorded. The standby power consumption control sequence is sampled and detected regularly to obtain the power consumption data sequence. Through regular sampling, the dynamic characteristics of power consumption over time are captured, and the power consumption fluctuations of the chip under different working states are reflected. In order to improve the accuracy of power consumption control, the power consumption data sequence is input into the high-frequency clock circuit for operating frequency adjustment. By adjusting the operating frequency, the balance between power consumption and computing efficiency is optimized, so that the chip can maintain efficient performance while minimizing unnecessary energy consumption when performing tasks, and obtain the clock control parameters. According to the clock control parameters, a linear regression analysis is performed on the power consumption data sequence to reveal the relationship between power consumption and variables such as time and working state, which helps the system predict future power consumption trends and obtain a power consumption prediction model. According to the output of the power consumption prediction model, the voltage conversion is performed through the DC / DC circuit. The DC / DC circuit can adjust the voltage to the optimal level according to the power consumption prediction results and the current working conditions, and optimize the power consumption performance of the system. The voltage value after voltage conversion forms a voltage control sequence, which is combined with the aforementioned power classification mapping vector, and the final dynamic power control parameters are obtained by searching the preset dynamic power control table.These parameters can automatically adjust the power consumption and transmission power of the Bluetooth chip according to changes in the actual transmission environment, ensuring the best balance between performance and power consumption in different working modes.

[0035] Step S4: setting a radio frequency sensitivity threshold according to the dynamic power control parameter, and compressing and encoding the data packet to obtain a dual-mode transmission protocol configuration parameter;

[0036] Specifically, first. According to the dynamic power control parameters, it is classified into classic Bluetooth mode and low-power Bluetooth mode, so that appropriate power control strategies can be adopted in different modes to obtain the RF sensitivity parameter matrix. The change of RF sensitivity directly affects the reception quality of the data packet. The sensitivity threshold is set according to different Bluetooth working modes to ensure data transmission stability and low power consumption. The RF sensitivity parameter matrix is ​​input into the received signal strength indicator (RSSI) to sample and analyze the RF signal. RSSI is an important indicator for measuring the strength of the received signal. The result of its sampling and analysis generates a signal strength distribution sequence. The sequence reflects the fluctuation of signal strength in different time windows and can help the system identify changes in signal quality. Through these signal strength data, the receiving sensitivity is accurately adjusted to adapt to different wireless environments. In order to optimize the receiving process, the signal strength distribution sequence is automatically gain controlled, and the signal strength is kept within a suitable range by adjusting the gain of the received signal. After gain control, the gain control parameter is generated, which describes the gain adjustment amplitude during the signal amplification process. According to the gain control parameter, the data packet is segmented and encoded, and the data packet is divided into three parts: frame header, control field and data field. Each part sets a different compression ratio according to its importance and transmission requirements. The control field contains the necessary control information in the network protocol, while the data field carries the actual data to be transmitted. Through the segmented coding strategy, different compression algorithms are used for different types of data to minimize the size of the data and reduce the power consumption during transmission while ensuring data integrity. When encoding the data packet, different compression algorithms are applied, among which the shift compression algorithm is used to compress the control field by 2:1. This algorithm reduces the amount of data in this part by half by removing the redundant information in the control field. For the data field, the run-length encoding algorithm is used, which compresses it by 4:1 by identifying the continuous repetitive pattern in the data field. After the compression algorithm is processed, the overall size of the data packet is greatly reduced, and the compressed and encoded data stream will be more compact than the original data stream, effectively saving bandwidth resources and reducing energy consumption during transmission. The frame format of the compressed encoded data stream is reorganized, and the compressed control field and data field are encapsulated according to the Bluetooth protocol specification to ensure that the data can be correctly transmitted under the framework of the Bluetooth protocol. Synchronization words and CRC check codes are added to each data packet. The synchronization word is used to ensure that the receiving end can correctly align the data stream, and the CRC check code is used to verify whether there is an error in the data transmission process. The output of this stage is a protocol frame sequence that meets the requirements of the Bluetooth protocol. The protocol frame sequence is mapped and associated with the RF sensitivity parameter matrix. This mapping relationship can help the system select the most suitable transmission power level, receiving sensitivity level and data packet format according to the current RF conditions, forming a mapping table containing all transmission parameters.Prioritize the parameters in the transmission parameter configuration table to determine which parameters are most critical to the current transmission process and which parameters can be appropriately relaxed or adjusted. The sorting process helps the system make flexible adjustments in a dynamic environment, thereby ensuring that power consumption is minimized and system performance is improved while meeting transmission quality requirements. The prioritized parameters form the dual-mode transmission protocol configuration parameters.

[0037] Step S5, performing timing sampling based on the dual-mode transmission protocol configuration parameters, calculating the packet time interval, signal strength change rate and link bit error rate, and constructing a dual-mode state feature sequence;

[0038] Specifically, timing sampling is performed based on the configuration parameters of the dual-mode transmission protocol. Time series data is obtained from the data transmission process to obtain a timing sampling matrix, which records the sending time and receiving time information of each data packet. Using these timestamp data, the time interval of each data packet is calculated to obtain a time interval statistical sequence. This sequence reflects the transmission time interval between each data packet, which helps to evaluate the transmission stability and load of the network. The time interval statistical sequence is input into the high-speed signal sampling unit to realize real-time sampling of signal strength. By continuously detecting the received signal strength value, a signal strength sampling sequence is obtained. A differential operation is performed on the signal strength sampling sequence, and a signal strength change rate matrix is ​​obtained by calculating the signal strength difference between two adjacent sampling points and dividing it by the time interval between the two sampling points. The signal strength change rate reflects the rate at which the signal strength changes over time. The rapid change of the signal strength means the instability of the link or the influence of external interference. Through this matrix, the drastic fluctuation of the signal quality is identified. The signal strength change rate matrix is ​​grouped and counted according to different sensitivity levels to obtain the signal change characteristic value. The time interval statistical sequence and the signal change characteristic value are combined by parameters to generate an error detection sequence. Through this sequence, error detection is performed in combination with the characteristics of time interval and signal change, so as to identify transmission errors and unstable factors in the link. The error detection sequence is input into the CRC check unit for processing. CRC (cyclic redundancy check) is an error detection algorithm that can effectively identify bit errors that occur during transmission. By performing a cyclic redundancy check on each data packet, the link bit error rate parameter is calculated, which reflects the frequency of errors during data transmission. Feature mapping is performed based on the time interval statistical sequence, signal change characteristic value and link bit error rate parameter, and this information is combined to obtain the dual-mode state feature sequence.

[0039] Step S6: Calculate the data packet delivery success rate, end-to-end transmission delay and total system power consumption based on the dual-mode state feature sequence, and optimize the transmission control parameters through double-layer deep reinforcement learning to obtain the target transmission control parameters.

[0040] Specifically, the dual-mode state feature sequence is input into the policy network layer of the two-layer deep reinforcement learning. The policy network layer contains the classic Bluetooth policy generator and the low-power Bluetooth value evaluator, which adopt the Actor network structure and the Critic network structure respectively. The role of the Actor network is to generate a suitable transmission action strategy, which represents the best operation strategy that the system should take in the current state. The Critic network provides value feedback by evaluating the current strategy, helping the system understand the effects of different strategies, thereby guiding the adjustment of the Actor network. Through the deep learning framework, the system learns how to dynamically select the best Bluetooth transmission mode according to the dual-mode state feature sequence, thereby improving the delivery success rate of data packets, reducing end-to-end transmission delay and optimizing the power consumption performance of the system. The dual-mode transmission action strategy is input into the target network layer for processing. The target network layer contains the dual-mode target policy network and the dual-mode target value network, which are responsible for the target state evaluation under different Bluetooth modes. The target policy network generates strategies that adapt to different modes based on the current state feature sequence, while the target value network provides the system with long-term goal evaluation in each mode by evaluating the Q value of the current state and action combination. This process enables the system to not only handle immediate transmission needs, but also optimize long-term system performance, ensuring that the best transmission decision can always be made in different Bluetooth modes. The dual-mode state feature sequence and target Q value are input into the experience replay buffer for storage. The experience replay buffer records the state transition samples of the classic Bluetooth mode and the low-power Bluetooth mode separately through a hierarchical storage structure. By accumulating a large number of state transition samples in the experience replay buffer, the system can better understand the relationship between state and action in various environments and optimize the transmission strategy through repeated learning. After the data is stored, the time difference error is calculated based on the dual-mode training sample set. The time difference error is an indicator that measures the gap between the current strategy and the target strategy. The error is used to evaluate the effect of the strategy and further adjust the network parameters. At the same time, key indicators such as transmission power, link quality and bit error rate are used as evaluation criteria of the reward function to guide the optimization of the learning process. Through the feedback of these reward signals, the system continuously updates the network parameters in the two Bluetooth modes and obtains a set of dual-mode optimization parameters, which can help the system maintain efficient transmission performance in different Bluetooth modes. In order to improve the stability and convergence speed of the system, the optimized dual-mode parameters are soft-updated. The soft update method can avoid training instability caused by updating too quickly, and can maintain the smoothness of the learning process while ensuring high update efficiency. For the classic Bluetooth mode, the update rate is set to 0.02, while for the low-power Bluetooth mode, the update rate is set to 0.01. Through this different update rate, the system can make more precise adjustments based on the transmission characteristics of each mode, ensuring that the transmission performance and power consumption can be effectively balanced under different conditions.After the soft update, the updated dual-mode target parameters are applied to the current transmission status, and the appropriate Bluetooth working mode is dynamically selected according to the transmission priority and power consumption requirements of the data packet. The system can automatically select the classic Bluetooth mode or the low-power Bluetooth mode according to the real-time transmission requirements and power consumption restrictions to achieve the best transmission control. Through the adaptive transmission control strategy, the Bluetooth chip can achieve efficient and low-power communication in different usage scenarios. Based on the adaptive transmission control strategy, the dynamic power control parameters, sensitivity threshold and compression coding parameters of the Bluetooth chip are generated. These parameters are converted into target transmission control parameters through the parameter mapping table, thus providing a full range of optimization solutions for Bluetooth communication. The dynamic power control parameters can automatically adjust the transmission power according to the changes in the environment, the sensitivity threshold adjusts the receiving sensitivity according to the signal quality, and the compression coding parameters adjust the data compression degree according to the transmission rate requirements.

[0041] In the embodiment of the present invention, the timing feature extraction is performed by a deep time convolution network, and the multi-head attention mechanism is combined to realize the accurate perception of the transmission state of the dual-mode Bluetooth chip and the mode switching judgment, so that the mode switching is more accurate and efficient. The dynamic power control scheme based on the nano-ampere LDO circuit, combined with the efficient conversion of the DC / DC circuit, significantly reduces the power consumption of the chip, especially the current consumption in the standby state is reduced to the order of 2μV. The segmented compression coding strategy is adopted to optimize the compression ratio of 2:1 and 4:1 for the control field and the data field, respectively, which reduces the protocol overhead and improves the data transmission efficiency. Through the two-layer deep reinforcement learning framework, the adaptive optimization of the transmission parameters is realized, in which the classic Bluetooth mode and the low power mode adopt different update strategies respectively, so that the parameter optimization is more accurate. The anomaly detection mechanism based on CRC check is designed, and the comprehensive evaluation of the signal strength change rate and the link bit error rate is combined to enhance the anti-interference ability and transmission reliability of the system. By dynamically adjusting the RF sensitivity threshold, the intelligent adjustment of the receiving sensitivity is realized, and the signal reception quality is improved. The experience playback mechanism based on the hierarchical storage structure improves the training efficiency of deep reinforcement learning and accelerates the response speed of the system to environmental changes. By adopting a multi-dimensional reward function evaluation indicator and comprehensively considering the transmission power, link quality and bit error rate, the optimal balance between transmission performance and power consumption is achieved.

[0042] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0043] The signal strength and link quality indicators of the dual-mode Bluetooth chip are sampled, and noise is filtered through a bandpass filter to obtain a signal strength parameter vector. The standby current and working current of the dual-mode Bluetooth chip are collected, and current detection is performed through a nano-ampere LDO circuit to obtain a power consumption parameter vector. The data packet throughput and transmission delay of the dual-mode Bluetooth chip are monitored, and data statistics are performed through a sliding window to obtain a transmission rate parameter vector.

[0044] The signal strength parameter vector, power consumption parameter vector and transmission rate parameter vector are input into the input layer of the deep temporal convolutional network, and a standardized feature matrix is ​​obtained after batch normalization.

[0045] The standardized feature matrix is ​​input into the first time convolution layer of the deep time convolution network for processing. The convolution kernel size of the first time convolution layer is 1×3, the step size is 1, the padding is 1, and the ReLU activation function is used to obtain the first feature map;

[0046] The first feature map is input into the dilated convolution layer of the deep temporal convolutional network for processing. The dilation rate of the dilated convolution layer is 2, the convolution kernel size is 1×3, and batch normalization and ReLU activation function are used to obtain the temporal feature map.

[0047] The time series feature map is processed through a multi-head attention layer, which contains 8 attention heads, each with a dimension of 64, to obtain attention weighted features;

[0048] The attention-weighted features are input into the fully connected layer for feature dimension reduction, and the probability values ​​of the classic Bluetooth mode and the low-power Bluetooth mode are calculated through the softmax classifier to obtain the transmission mode switching judgment result.

[0049] Specifically, the signal strength and link quality indicators of the dual-mode Bluetooth chip are sampled. The signal strength is measured by the received signal strength indicator (RSSI), and the link quality indicator reflects the stability and transmission quality of the signal. A bandpass filter is used to filter the signal noise, effectively removing low-frequency and high-frequency noise and retaining the valid part of the signal. The passband frequency of the bandpass filter is set to and , then the filtered signal is expressed as:

[0050] ;

[0051] in, represents the filtered signal, is the original signal, is the frequency response of the bandpass filter, is the frequency variable. Collect the power consumption parameters of the chip. The power consumption includes standby current and working current. The standby current is the current when the chip is in a low power state, and the working current is the actual current of the chip during data transmission. Through the nano-ampere LDO circuit, these currents are accurately detected, the current value is output, and the power consumption parameters are obtained through the current detection circuit. Assume that the standby current is , the working current is , then the power consumption parameter vector is expressed as:

[0052] ;

[0053] in, is the power consumption, is the current (standby current or operating current), is the operating voltage of the chip. The transmission rate is collected and statistics are made by monitoring the throughput and transmission delay of data packets. The transmission rate is determined by the throughput (the amount of data transmitted per unit time) and the delay (the time required for data to be sent and received). Let the throughput be , the delay is , the transmission rate parameter vector is expressed as:

[0054] ;

[0055] Through the sliding window method, the data packet throughput and transmission delay within a certain time range are counted to obtain the time series data of the transmission rate. The signal strength parameter vector, power consumption parameter vector and transmission rate parameter vector are input into the deep time convolution network. These data vectors are processed by batch normalization to standardize each dimension of data so that the network can better train and infer. The formula for batch normalization is:

[0056] ;

[0057] in, is the original data, is the mean of the data, is the standard deviation of the data, is the standardized data. The standardized feature matrix is ​​input into the first time convolution layer of the deep time convolution network. This convolution layer uses a 1×3 convolution kernel, a step size of 1, a padding of 1, and a ReLU activation function for nonlinear transformation. The definition of the ReLU activation function is:

[0058] ;

[0059] Through convolution operation and ReLU activation function, the network can extract preliminary time series features from the original data. The convolution process is described by the following formula:

[0060] ;

[0061] in, is the convolution result, For input data, is the convolution kernel, is the size of the convolution kernel. The first feature map is input into the dilated convolution layer. The characteristic of the dilated convolution layer is that the dilation rate of the convolution kernel is greater than 1, thereby increasing the receptive field without increasing the amount of calculation. Suppose the dilation rate is , the convolution kernel size is still 1×3, and after the dilated convolution, a larger range of temporal feature maps are obtained. The formula for the dilated convolution layer is:

[0062] ;

[0063] in, is the expansion rate, is the convolution kernel, is the convolution kernel size. Dilated convolution can capture longer-distance temporal dependencies, thereby enriching temporal features. The temporal feature map processed by the dilated convolution layer is input into the multi-head attention layer. The multi-head attention layer contains multiple attention heads, each with a dimension of 64. Each attention head independently calculates the weight of the input data and sums them up to obtain weighted features. The multi-head attention mechanism can focus on different parts of the data at the same time, helping the model capture more potential rules. The attention weighted feature is expressed as:

[0064] ;

[0065] in, , and are query, key, and value matrices respectively, is the dimension of query and key, and the softmax function is used to normalize the calculation weights. The attention weighted features are input into the fully connected layer for feature dimensionality reduction. Through the fully connected layer, the network can map high-dimensional features to a lower-dimensional space and extract the most meaningful features. The softmax classifier is used to classify the reduced features and calculate the probability values ​​of the classic Bluetooth mode and the low-power Bluetooth mode. The calculation formula of the Softmax function is:

[0066] ;

[0067] in, Yes Category The score, Yes Category The probability of Traverse all categories. Through the probability value output by the softmax classifier, the probabilities of the classic Bluetooth mode and the low-power Bluetooth mode are obtained, and then the transmission mode switching decision is made to obtain the transmission mode switching decision result.

[0068] In a specific embodiment, the process of executing step S2 may specifically include the following steps:

[0069] The transmission mode switching determination result is divided into a classic Bluetooth mode parameter matrix and a low-power Bluetooth mode parameter matrix, and features are combined to obtain a hybrid mode feature vector;

[0070] The mixed mode feature vector is segmented according to the time window to obtain the data packet statistical matrix;

[0071] The data packet statistics matrix is ​​classified according to the transmission status, which includes normal transmission, packet loss retransmission and transmission timeout, and the state counter is used to accumulate statistics to obtain the transmission state counting result;

[0072] According to the transmission status counting result, the ratio of the number of packet losses in the unit time window to the total number of transmitted data packets is calculated to obtain the packet loss rate;

[0073] The sending timestamp and receiving timestamp of each data packet in the data packet statistical matrix are recorded, and the transmission delay of a single data packet is calculated by the timestamp difference to obtain the transmission delay sequence, and the transmission delay sequence is input into the median filter to filter outliers to obtain the transmission delay;

[0074] The signal transmission power in the mixed mode feature vector is sampled and combined with the standby current value of the nanoampere LDO circuit to obtain a power consumption sampling sequence. The product of the power consumption sampling sequence and the operating voltage is accumulated and converted through a power consumption classification mapping table to obtain real-time power consumption.

[0075] Generate transmission status evaluation data based on packet loss rate, transmission delay and real-time power consumption.

[0076] Specifically, the transmission mode switching judgment result is divided into a classic Bluetooth mode parameter matrix and a low-power Bluetooth mode parameter matrix. In these matrices, the parameters of the classic Bluetooth mode involve higher transmission rates and larger power consumption, while the low-power Bluetooth mode focuses more on low power consumption and long standby capability. By combining the parameters of the two modes, a hybrid mode feature vector is obtained, which includes various parameters in the classic Bluetooth mode and the low-power Bluetooth mode. The hybrid mode feature vector is segmented according to time windows to better track and analyze the changes of various parameters during the transmission process. Each time window represents a fixed time period, during which all transmission feature data are collected and counted to form a data packet statistical matrix. The rows and columns of the matrix represent different time windows and different data packet features, such as throughput, delay, packet loss, etc. In this way, the transmission of data packets in different time periods is accurately analyzed to identify potential performance bottlenecks. The data packet statistical matrix is ​​classified according to the transmission status. The transmission status is divided into three types: normal transmission, packet loss retransmission, and transmission timeout. In the actual transmission process, packet loss may lead to retransmission, while transmission timeout means that the data packet has not been successfully transmitted within the predetermined time. These states are accumulated and counted through the state counter to obtain the transmission state counting result. The state counter accumulates the number of occurrences of each transmission state and evaluates the system performance based on the statistical results. Based on the transmission state counting results, the packet loss rate is calculated. The loss rate refers to the ratio of the number of lost packets per unit time to the total number of transmitted packets. The calculation formula for the packet loss rate is:

[0077]

[0078] in, is the packet loss rate, is the number of packet losses, is the total number of transmitted data packets. For the transmission delay of the data packet, the transmission delay of each data packet is calculated by recording the sending timestamp and receiving timestamp of each data packet. The transmission delay sequence is calculated by the following formula:

[0079] ;

[0080] in, For the The transmission delay of a data packet is To receive the timestamp, is the sending timestamp. If the transmission delay of some data packets is abnormal, they are processed through a median filter to remove these outliers. The median filter sorts the data sequence and takes the middle value to remove outliers. The transmission delay after filtering is expressed as:

[0081] ;

[0082] The signal transmission power in the mixed-mode feature vector is sampled to evaluate the power consumption. The signal transmission power is sampled together with the standby current of the nano-ampere LDO circuit, and the power consumption is calculated in combination with the operating voltage. The power consumption sampling sequence is expressed as:

[0083] ;

[0084] in, is the power consumption sampling value, is the standby current, is the working current, is the operating voltage. Through these sampled data, the power consumption changes in each time window are calculated in real time, and converted based on the power consumption classification mapping table to obtain the final real-time power consumption. Based on the three key parameters of packet loss rate, transmission delay and real-time power consumption, the transmission status evaluation data is generated. These data reflect the performance of the current transmission process and provide a decision basis for subsequent transmission mode switching and power regulation. For example, if the loss rate is too high or the transmission delay is too long, it is necessary to switch to low-power Bluetooth mode to reduce power consumption, or optimize the transmission strategy to improve efficiency.

[0085] In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0086] According to the loss rate threshold and the delay threshold, the packet loss rate and the transmission delay in the transmission status evaluation data are subjected to threshold analysis to obtain a power regulation reference matrix;

[0087] The power regulation reference matrix is ​​divided into a high power transmission state and a low power transmission state according to the transmission power level, and a power classification mapping vector is obtained;

[0088] Based on the power classification mapping vector, current detection is performed through a nano-ampere LDO circuit, the operating voltage of the nano-ampere LDO circuit is 1.0V, and the standby current threshold of the nano-ampere LDO circuit is 2μV, so as to obtain a standby power consumption control sequence;

[0089] Performing timed sampling detection on the standby power consumption control sequence to obtain a power consumption data sequence, and inputting the power consumption data sequence into a high-frequency clock circuit to adjust the operating frequency to obtain a clock control parameter;

[0090] Perform linear regression analysis on the power consumption data sequence according to the clock control parameters to obtain a power consumption prediction model;

[0091] The result output by the power consumption prediction model is converted into voltage through a DC / DC circuit to obtain a voltage control sequence, and the voltage control sequence is combined with a power classification mapping vector to obtain dynamic power control parameters by searching a preset dynamic power control table.

[0092] Specifically, according to the loss rate threshold and the delay threshold, the packet loss rate and the transmission delay in the transmission status evaluation data are subjected to threshold analysis. ) and transmission delay ( ) is an important parameter that affects power consumption. The loss rate reflects the probability of system data packet loss, while the delay affects the real-time performance and transmission efficiency of the data. If the loss rate exceeds a certain set threshold, or the delay exceeds a predetermined maximum value, it is determined that the current transmission state requires power adjustment. Threshold analysis based on loss rate and delay is expressed by the following formula:

[0093] ;

[0094] in, is the packet loss rate, is the number of packets lost, is the total number of packets transmitted. If , then enter the power adjustment stage. After obtaining the power adjustment reference matrix, it is divided into high-power transmission state and low-power transmission state according to the transmission power level. When the loss rate is high or the delay is large, the system tends to enter the high-power state to improve the transmission quality; on the contrary, when the system is in a low loss rate and low delay state, the system will switch to a low-power state to reduce power consumption. The division of these states is expressed by the following formula:

[0095] ;

[0096] Based on the state division, a power classification mapping vector is obtained. This vector contains the power settings under different transmission states. For example, in the high power state, the system's transmission power is large, while in the low power state, the transmission power will be reduced accordingly. The power classification mapping vector ( ) is expressed by the following formula:

[0097] ;

[0098] in, Indicates the transmit power in high power state, Indicates the transmit power in low power state. Current detection is performed through a nanoampere LDO circuit. The LDO circuit is capable of operating in a very low voltage and current range and is suitable for power management. The operating voltage of the LDO circuit is set to 1.0V and the standby current threshold is 2μV. The current detection process is expressed by the following formula:

[0099] ;

[0100] in, For time The current sampling value at the moment, is the operating voltage of the LDO circuit, is the load resistance. When the system is in standby mode, the current change is very small, usually around 2μV. Based on this current sampling sequence, the standby power consumption control sequence is obtained, which can dynamically adjust the standby power consumption and ensure that the system can still maintain basic functions in low power mode. The standby power consumption control sequence is detected by timing sampling to obtain the power consumption data sequence. The power consumption data sequence is calculated by the following formula:

[0101] ;

[0102] in, For time Power consumption data at the moment, is the current sampling value at that time, is the operating voltage. The power consumption data sequence is input into the high-frequency clock circuit to adjust the operating frequency of the system. The operating frequency is adjusted using the following formula:

[0103] ;

[0104] in, For time The working frequency at the moment, is the maximum frequency, is the maximum power consumption. Based on the real-time power consumption data, the system will dynamically adjust the clock frequency to reduce power consumption and maintain transmission performance. After obtaining the power consumption data sequence and the clock control parameters after frequency adjustment, a power consumption prediction model is constructed through linear regression analysis. The linear regression model is expressed by the following formula:

[0105] ;

[0106] in, For time The predicted power consumption at the moment, is the regression coefficient, is the error term. Through linear regression, the power consumption under given clock frequency and delay conditions is predicted. The output of the power consumption prediction model will be converted into voltage through the DC / DC circuit to obtain a voltage control sequence. The DC / DC circuit can adjust the output voltage according to the input power consumption to meet the power requirements of the system. The voltage conversion formula is expressed as:

[0107] ;

[0108] in, is the output voltage, is the input voltage, To predict power consumption, The maximum power consumption is obtained by combining the voltage control sequence with the power classification mapping vector, adjusting the power and voltage according to different working modes (high power or low power) to obtain the optimal transmission performance and power consumption performance. By looking up the preset dynamic power control table, the final dynamic power control parameters are obtained. These control parameters are used to adjust the power of the Bluetooth chip in real time to maintain the best performance in different transmission environments while minimizing power consumption.

[0109] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0110] The dynamic power control parameters are classified according to the classic Bluetooth mode and the low-power Bluetooth mode to obtain the radio frequency sensitivity parameter matrix;

[0111] The radio frequency sensitivity parameter matrix is ​​input into the received signal strength indicator, and the radio frequency signal is sampled and analyzed to obtain a signal strength distribution sequence;

[0112] Automatic gain control is performed on the signal strength distribution sequence to obtain a gain control parameter, and segment encoding is performed on the data packet according to the gain control parameter, the data packet is divided into a frame header, a control field and a data field, and different compression ratios are set for different fields to obtain a segment compression strategy;

[0113] The segmented compression strategy is applied to the data packet encoding process, the control field is compressed 2:1 by the displacement compression algorithm, and the data field is compressed 4:1 by the run-length encoding algorithm to obtain the compressed encoded data stream;

[0114] Reorganize the frame format of the compressed coded data stream, encapsulate the compressed control field and data field according to the Bluetooth protocol specification, add the synchronization word and CRC check code, and obtain the protocol frame sequence;

[0115] The protocol frame sequence is mapped and associated with the radio frequency sensitivity parameter matrix to generate a mapping table including the transmission power level, the receiving sensitivity level and the data packet format, and obtain a transmission parameter configuration table;

[0116] Prioritize the parameters in the transmission parameter configuration table to obtain dual-mode transmission protocol configuration parameters.

[0117] Specifically, the dynamic power control parameters are classified into classic Bluetooth mode and low-power Bluetooth mode. Different RF sensitivity parameters are set for classic Bluetooth mode and low-power Bluetooth mode according to different transmission requirements and environments. The classic Bluetooth mode requires stronger signal sensitivity and higher transmission power to ensure stable transmission over a longer distance, while the low-power Bluetooth mode requires lower transmission power and sensitivity to reduce power consumption. After the classification is completed, the RF sensitivity parameter matrix ( ), which is expressed by the following formula:

[0118] ;

[0119] in, Indicates the radio frequency sensitivity parameter in classic Bluetooth mode. Represents the RF sensitivity parameter in low-power Bluetooth mode. The RF sensitivity parameter matrix is ​​input into the received signal strength indicator (RSSI) for sampling analysis. RSSI is used to measure the strength of the received RF signal. Through this process, the strength distribution sequence of the RF signal is obtained ( ), that is, the change of signal strength at different time points. The signal strength distribution sequence is expressed by the following formula:

[0120] ;

[0121] in, Indicates at time The signal strength received at all times. Automatic gain control is performed on the signal strength distribution sequence, and the system gain is automatically adjusted according to the received signal strength to maintain signal stability. Gain control parameters ( ) is used to adjust the gain of the system so that the received signal strength always remains within an ideal range. The calculation formula for gain control is expressed as:

[0122] ;

[0123] in, Indicates at time The gain control parameter at time, is the received signal strength, is the target signal strength. Based on the gain control parameter, the data packet is segmented and encoded, and the data packet is divided into a frame header, a control field, and a data field, and different compression ratios are set according to the characteristics of different fields. The frame header contains synchronization information and control information, the control field contains scheduling information, and the data field contains the actual transmission data. The compression ratio of the control field is 2:1, while the compression ratio of the data field is 4:1. Through the compression strategy, the size of the data packet is effectively reduced, the bandwidth required for transmission is reduced, and the transmission efficiency is optimized. The process is expressed by the following formula:

[0124] ;

[0125] in, Indicates the compressed data field. Indicates the first parts, Indicates the compression ratio of each field (2:1 for the control field and 4:1 for the data field). The frame format of the compressed and encoded data stream is reorganized, and the compressed control field and data field are encapsulated according to the Bluetooth protocol specification. The encapsulation process includes adding a synchronization word and a CRC checksum to the front of the data packet to ensure the integrity and accuracy of the data. The synchronization word is used to identify the beginning of the data packet, and the CRC checksum is used to detect errors in the data packet during transmission. The process is represented by the following formula:

[0126] ;

[0127] in, Indicates the final protocol frame, SYNC is the synchronization word, and Respectively represent the compressed control field and data field, and CRC is the CRC check code. After the protocol frame sequence is generated, it is mapped and associated with the RF sensitivity according to the RF sensitivity parameter matrix to obtain a mapping table containing the transmit power level, the receive sensitivity level and the data packet format. Mapping table ( ) is expressed by the following formula:

[0128] ;

[0129] in, Indicates the transmit power level. Indicates the receiving sensitivity level. Indicates the data packet format. Prioritize each parameter in the transmission parameter configuration table to develop the optimal transmission control strategy based on transmission requirements and power consumption requirements. Priority sorting is expressed by the following formula:

[0130] ;

[0131] in, are the weights of transmit power, receive sensitivity, and packet format, reflecting their importance in the overall transmission efficiency. In this way, the system dynamically adjusts each parameter according to the priority of the transmission task to obtain the most suitable dual-mode transmission protocol configuration parameters ( .

[0132] In a specific embodiment, the process of executing step S5 may specifically include the following steps:

[0133] Perform timing sampling based on the dual-mode transmission protocol configuration parameters to obtain a timing sampling matrix, and calculate the time interval between the sending time and the receiving time of the data packet in the timing sampling matrix to obtain a time interval statistical sequence;

[0134] The time interval statistical sequence is input into the high-speed signal sampling unit, and the signal strength is sampled in real time to obtain a signal strength sampling sequence;

[0135] Perform differential operation on the signal strength sampling sequence, and obtain the signal strength change rate matrix based on the signal strength difference between two adjacent sampling points divided by the sampling time interval;

[0136] The signal strength change rate matrix is ​​grouped and counted according to different sensitivity levels to obtain the signal change characteristic value, and the time interval statistical sequence and the signal change characteristic value are combined by parameters to obtain the error detection sequence;

[0137] The error detection sequence is input into the CRC check unit, and a cyclic redundancy check is performed on each data packet to obtain the link error rate parameter. Feature mapping is performed based on the time interval statistical sequence, the signal change characteristic value and the link error rate parameter to obtain the dual-mode state feature sequence.

[0138] Specifically, when performing timing sampling, the dual-mode transmission protocol configuration parameters are used to determine the sampling time and relevant information of the data packet. The configuration parameters include transmission status, transmission mode, power control, sensitivity threshold, etc. Based on these configuration parameters, the sending time and receiving time of the data packet are collected, and the efficiency and quality of the transmission are evaluated by calculating the time interval. For each data packet, its sending time ( ) and receiving time ( ), and then calculate the time interval between data packets ( ), which is the difference between the receiving time and the sending time, and the formula is as follows:

[0139] ;

[0140] The obtained time intervals constitute the time interval statistical series ( ), reflecting the transmission delay characteristics of data packets at different time points, helping to analyze delay fluctuations during transmission. The time interval statistical sequence can effectively reveal changes in network status and help the system determine whether there are problems such as packet loss and excessive delay. The time interval statistical sequence is input into the high-speed signal sampling unit for real-time signal sampling. The signal sampling unit monitors the received signal strength in real time and generates a signal strength sampling sequence ). Real-time sampling of signal strength is an important indicator for evaluating signal quality and link stability. The signal strength sampling sequence is used to determine whether the signal is stable and whether there is attenuation. The calculation formula for the signal strength sampling sequence is:

[0141] ;

[0142] in, Indicated in The received signal strength at the moment (RSSI). The signal strength sampling sequence is differentiated to calculate the rate of change of signal strength over time, thereby evaluating the fluctuation of signal quality. The signal strength change rate is calculated by dividing the difference of signal strength between two adjacent sampling points by the time interval ( ). The signal intensity change rate is expressed as:

[0143] ;

[0144] in, Indicates The signal strength change rate at time, and Respectively represent the signal strength of two adjacent sampling moments, is the time interval between two adjacent samples. The signal strength change rate matrix ( ), which reflects the fluctuation of signal strength in the time dimension. If the change rate is too large, it means that the signal quality is unstable, which will affect the data transmission quality. The signal strength change rate matrix is ​​grouped and counted according to different sensitivity levels. Different sensitivity levels correspond to different signal quality requirements. In the low sensitivity mode, it is more sensitive to signal changes, so the signal change rate will be higher. Through group statistics, the signal change characteristic value ( ). The calculation formula of the signal change characteristic value is:

[0145] ;

[0146] in, is the signal change characteristic value, is the first values, Represents the number of samples. The time interval statistical sequence and the signal change characteristic value are combined by parameters to obtain the error detection sequence ( ). The transmission quality of the data packet is evaluated based on the time interval and signal changes, and then it is determined whether there are problems such as bit errors or packet loss. The calculation formula for the error detection sequence is:

[0147] ;

[0148] in, Indicates The error detection result at the moment, It is a composite function based on the time interval and the characteristic value of the signal change, which represents a comprehensive evaluation of the transmission quality of each data packet. The error detection sequence is input into the cyclic redundancy check (CRC) unit for detection to confirm the transmission integrity of the data packet. The CRC check determines whether the data is erroneous by calculating a check value for each data packet. If the check result does not match, it means that the data packet has an error during transmission. For each data packet, the CRC check obtains the link error rate ( ). The calculation formula of link bit error rate is:

[0149] ;

[0150] in, represents the link bit error rate, Number of errors is the number of packets with errors, and Total number of packets is the total number of packets. Based on the time interval statistical sequence, signal change characteristic value and link bit error rate parameter, feature mapping is performed to obtain the dual-mode state feature sequence. This feature sequence can help the system dynamically adjust and optimize the efficiency and stability of data transmission according to the current transmission state. The calculation formula of the dual-mode state feature sequence is:

[0151] ;

[0152] in, Represents the dual-mode state feature sequence, including multi-dimensional features such as time interval statistical sequence, signal change feature value and link bit error rate.

[0153] In a specific embodiment, the process of executing step S6 may specifically include the following steps:

[0154] The dual-mode state feature sequence is input into the policy network layer of the two-layer deep reinforcement learning. The policy network layer contains the classic Bluetooth policy generator and the low-power Bluetooth value evaluator, which use the Actor network structure and the Critic network structure respectively to obtain the dual-mode transmission action strategy.

[0155] The dual-mode transmission action strategy is input into the target network layer of the two-layer deep reinforcement learning for processing. The target network layer includes a dual-mode target strategy network and a dual-mode target value network, which are responsible for evaluating the target state under different Bluetooth modes and obtaining the dual-mode target Q value.

[0156] The dual-mode state feature sequence and the dual-mode target Q value are input into the experience playback buffer, and the state transition samples of the classic Bluetooth mode and the low-power Bluetooth mode are recorded respectively through the hierarchical storage structure to obtain the dual-mode training sample set;

[0157] The timing difference error is calculated based on the dual-mode training sample set, and the transmission power, link quality and bit error rate are used as evaluation indicators of the reward function. The network parameters in the two modes are updated respectively to obtain the dual-mode optimization parameters.

[0158] Perform soft update on the dual-mode optimization parameters. Set different update rates based on the transmission characteristics of different Bluetooth modes. The update rate for the classic Bluetooth mode is 0.02, and the update rate for the low-power Bluetooth mode is 0.01. Then, the updated dual-mode target parameters are obtained.

[0159] Apply the updated dual-mode target parameters to the current transmission state, dynamically select the Bluetooth working mode according to the data packet transmission priority and power consumption requirements, and obtain an adaptive transmission control strategy;

[0160] The dynamic power control parameters, sensitivity threshold and compression coding parameters of the dual-mode Bluetooth chip are generated based on the adaptive transmission control strategy, and the strategy is converted into the target transmission control parameters through the parameter mapping table.

[0161] Specifically, the dual-mode state feature sequence is input into the policy network layer of the two-layer deep reinforcement learning system so as to perform intelligent optimization according to the current transmission state. In this process, the policy network layer contains a classic Bluetooth policy generator and a low-power Bluetooth value evaluator, in which the classic Bluetooth mode and the low-power Bluetooth mode adopt the Actor network structure and the Critic network structure respectively. These structures can independently calculate the transmission strategies and values ​​in different modes and generate corresponding dual-mode transmission action strategies. The dual-mode state feature sequence contains multiple transmission state parameters, such as time interval, signal strength change, link bit error rate, etc. After the feature sequence is input into the policy network layer of the two-layer deep reinforcement learning, the policy network layer uses the Actor-Critic architecture to make intelligent decisions. In this architecture, the Actor network is used to generate action strategies under a given state, while the Critic network evaluates the quality of the strategy and calculates the value under the strategy. The Actor network is based on the state feature sequence. ) Output an action strategy ( ), that is, to choose the classic Bluetooth mode or the low-power Bluetooth mode. The Critic network calculates the value of the strategy based on the combination of state and action, and obtains the dual-mode transmission action strategy ( ). The dual-mode transmission action strategy is input into the target network layer of the dual-layer deep reinforcement learning for processing. The target network layer contains a dual-mode target strategy network and a dual-mode target value network, which evaluate the target state in the classic Bluetooth mode and the low-power Bluetooth mode respectively to obtain the dual-mode target Q value. The Q value represents the long-term return expectation under a specific state and action combination, and is a core concept in reinforcement learning. The calculation formula for the dual-mode target Q value is:

[0162] ;

[0163] in, Indicates status and actions The Q value, Yes Status and actions Instant rewards, is the discount factor, which determines the weight of future rewards, From the next state The Q value of the best action is selected. The dual-mode state feature sequence and the dual-mode target Q value are input into the experience replay buffer, and the state transition samples of the classic Bluetooth mode and the low-power Bluetooth mode are recorded respectively through the hierarchical storage structure. The experience replay buffer is an important component in deep reinforcement learning, which is used to store past state transition samples. These samples are randomly selected for training the network. At this stage, the dual-mode training sample set ( ) is constructed by storing the quadruple of state, action, reward, and next state in classic Bluetooth and low-power Bluetooth modes. The training sample set is represented as:

[0164] ;

[0165] in, Indicates time status, is the action performed, is the reward obtained after performing the action, is the state at the next moment. Based on the dual-mode training sample set, the temporal difference error is calculated. The temporal difference error is used to measure the update degree of the Q value, and the formula is as follows:

[0166] ;

[0167] in, is the timing difference error, It's a reward. is the discount factor, is the Q value of the current state and action, is the maximum Q value of the next state. According to the timing difference error, the network parameters are updated. In order to evaluate the effect of dual-mode transmission, the transmission power, link quality and bit error rate are used as evaluation indicators of the reward function. The system takes these indicators into consideration and forms a reward function based on these characteristics ( ). The formula of the reward function is:

[0168] Link Quality Error Rate

[0169] in, are weight coefficients, which respectively represent the influence of transmission power, link quality and bit error rate on the total reward. By calculating the reward function, the network parameters are updated according to the above-mentioned timing difference error and reward value, and the dual-mode optimization parameters ( ). These optimization parameters can adjust the transmission strategies in classic Bluetooth and low-power Bluetooth modes, allowing the system to cope with different working conditions more effectively. For classic Bluetooth mode and low-power Bluetooth mode, the network update rate is different to adapt to their respective transmission characteristics. The update rate of classic Bluetooth mode is 0.02, and the update rate of low-power Bluetooth mode is 0.01. This means that the classic Bluetooth mode will update its strategy more frequently, while the low-power Bluetooth mode will update at a slower rate to avoid power consumption caused by too frequent adjustments. Through soft update, the updated dual-mode target parameters ( ). The updated parameters are applied to the current transmission state. According to the transmission priority and power consumption requirements of the data packet, the system dynamically selects the Bluetooth working mode to implement an adaptive transmission control strategy. Transmission priority and power consumption requirements will affect the final transmission mode selection. The classic Bluetooth mode is selected for applications with high bandwidth requirements, while the low-power Bluetooth mode is used for applications with low bandwidth and energy-saving requirements. Based on the adaptive transmission control strategy, the dynamic power control parameters, sensitivity threshold and compression coding parameters of the dual-mode Bluetooth chip are generated. These parameters are converted through a parameter mapping table to ultimately form the target transmission control parameters. The target transmission control parameters include the transmit power ( )、Receiver Sensitivity( ), data packet compression rate ( ), etc. These parameters are used to adjust the working mode of the chip, optimize transmission performance and reduce power consumption. According to the transmission status and working conditions, the power level, compression strategy and sensitivity threshold are dynamically adjusted to achieve the best transmission effect and energy efficiency in different application scenarios.

[0170] The above describes the data packet transmission control method of the low-power dual-mode Bluetooth chip in the embodiment of the present invention. The following describes the data packet transmission control device of the low-power dual-mode Bluetooth chip in the embodiment of the present invention. Figure 2 , an embodiment of a data packet transmission control device of a low-power dual-mode Bluetooth chip in an embodiment of the present invention includes:

[0171] The acquisition module is used to obtain the signal strength parameters, power consumption parameters, transmission rate parameters and environmental parameters of the dual-mode Bluetooth chip, and input them into the deep time convolution network for time series feature extraction to obtain the transmission mode switching judgment result;

[0172] A calculation module, used to calculate transmission status evaluation data according to the transmission mode switching determination result, the transmission status evaluation data including: packet loss rate, transmission delay and real-time power consumption;

[0173] The adjustment module is used to adjust the chip transmission power in stages based on the transmission status evaluation data, and to control the standby power consumption using a nano-ampere LDO circuit to obtain dynamic power control parameters;

[0174] The encoding module is used to set the radio frequency sensitivity threshold according to the dynamic power control parameter and compress and encode the data packet to obtain the dual-mode transmission protocol configuration parameters;

[0175] A sampling module is used to perform timing sampling based on the dual-mode transmission protocol configuration parameters, calculate the packet time interval, signal strength change rate and link bit error rate, and construct a dual-mode state feature sequence;

[0176] The optimization module is used to calculate the packet delivery success rate, end-to-end transmission delay and total system power consumption based on the dual-mode state feature sequence, and optimize the transmission control parameters through two-layer deep reinforcement learning to obtain the target transmission control parameters.

[0177] Through the cooperation of the above components, the deep time convolution network is used to extract time series features, and combined with the multi-head attention mechanism, the transmission status of the dual-mode Bluetooth chip is accurately perceived and the mode switching judgment is made, making the mode switching more accurate and efficient. The dynamic power control scheme based on the nano-ampere LDO circuit, combined with the efficient conversion of the DC / DC circuit, significantly reduces the power consumption of the chip, especially the current consumption in the standby state is reduced to the order of 2μV. The segmented compression coding strategy is adopted to optimize the compression ratio of the control field and the data field by 2:1 and 4:1, respectively, which reduces the protocol overhead and improves the data transmission efficiency. The adaptive optimization of transmission parameters is realized through the two-layer deep reinforcement learning framework, in which the classic Bluetooth mode and the low power mode adopt different update strategies respectively, making the parameter optimization more accurate. The anomaly detection mechanism based on CRC check is designed, combined with the comprehensive evaluation of the signal strength change rate and the link bit error rate, to enhance the anti-interference ability and transmission reliability of the system. By dynamically adjusting the RF sensitivity threshold, the intelligent adjustment of the receiving sensitivity is realized, and the signal reception quality is improved. The experience replay mechanism based on the hierarchical storage structure improves the training efficiency of deep reinforcement learning and accelerates the response speed of the system to environmental changes. By adopting a multi-dimensional reward function evaluation indicator and comprehensively considering the transmission power, link quality and bit error rate, the optimal balance between transmission performance and power consumption is achieved.

[0178] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0179] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0180] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0181] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0182] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0183] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0184] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A data packet transmission control method for a low-power dual-mode Bluetooth chip, characterized in that: The method comprises: The signal strength, link quality index, standby current, working current, data packet throughput and transmission delay of the dual-mode Bluetooth chip are obtained, and input into the deep time convolution network for timing feature extraction to obtain the transmission mode switching judgment result; Calculating transmission state evaluation data according to the transmission mode switching determination result, wherein the transmission state evaluation data includes: packet loss rate, transmission delay and real-time power consumption; Based on the transmission status evaluation data, the chip transmission power is adjusted in stages, and the standby power consumption is controlled by using a nano-ampere level LDO circuit to obtain a dynamic power control parameter; The radio frequency sensitivity threshold is set according to the dynamic power control parameter, and the data packet is compressed and encoded to obtain the dual-mode transmission protocol configuration parameter; specifically including: classifying the dynamic power control parameter according to the classic Bluetooth mode and the low-power Bluetooth mode to obtain the radio frequency sensitivity parameter matrix; inputting the radio frequency sensitivity parameter matrix into the received signal strength indicator, sampling and analyzing the radio frequency signal, and obtaining a signal strength distribution sequence; performing automatic gain control on the signal strength distribution sequence to obtain a gain control parameter, and segmenting the data packet according to the gain control parameter, dividing the data packet into a frame header, a control field and a data field, and setting different compression ratios for different fields to obtain a segmented compression strategy; The segmented compression strategy is applied to the data packet encoding process, the control field is compressed 2:1 by a displacement compression algorithm, and the data field is compressed 4:1 by a run-length encoding algorithm to obtain a compressed encoded data stream; the compressed encoded data stream is reorganized in frame format, the compressed control field and data field are encapsulated according to the Bluetooth protocol specification, and a synchronization word and a CRC check code are added to obtain a protocol frame sequence; the protocol frame sequence is mapped and associated with the radio frequency sensitivity parameter matrix to generate a mapping table including a transmission power level, a receiving sensitivity level and a data packet format to obtain a transmission parameter configuration table; the parameters in the transmission parameter configuration table are prioritized to obtain dual-mode transmission protocol configuration parameters; Performing timing sampling based on the dual-mode transmission protocol configuration parameters, calculating the packet time interval, signal strength change rate and link bit error rate, and constructing a dual-mode state feature sequence; The data packet delivery success rate, end-to-end transmission delay and total system power consumption are calculated based on the dual-mode state feature sequence, and the transmission control parameters are optimized through double-layer deep reinforcement learning to obtain the target transmission control parameters.

2. The data packet transmission control method of the low-power dual-mode Bluetooth chip according to claim 1, characterized in that: The signal strength, link quality index, standby current, working current, data packet throughput and transmission delay of the dual-mode Bluetooth chip are obtained, and input into the deep time convolution network for timing feature extraction to obtain the transmission mode switching determination result, including: The signal strength and link quality indicators of the dual-mode Bluetooth chip are sampled, and noise is filtered through a bandpass filter to obtain a signal strength parameter vector. The standby current and working current of the dual-mode Bluetooth chip are collected, and current detection is performed through a nano-ampere LDO circuit to obtain a power consumption parameter vector. The data packet throughput and transmission delay of the dual-mode Bluetooth chip are monitored, and data statistics are performed through a sliding window to obtain a transmission rate parameter vector. Inputting the signal strength parameter vector, the power consumption parameter vector and the transmission rate parameter vector into the input layer of a deep temporal convolutional network, and performing batch normalization processing to obtain a standardized feature matrix; Input the standardized feature matrix into the first time convolution layer of the deep time convolution network for processing, where the convolution kernel size of the first time convolution layer is 1×3, the step size is 1, the padding is 1, and the ReLU activation function is used to obtain a first feature map; Inputting the first feature map into the dilated convolution layer of the deep temporal convolutional network for processing, wherein the dilation rate of the dilated convolution layer is 2, the convolution kernel size is 1×3, and batch normalization and ReLU activation function are used to obtain a temporal feature map; The temporal feature map is processed through a multi-head attention layer, where the multi-head attention layer includes 8 attention heads, each with a dimension of 64, to obtain an attention weighted feature; The attention weighted features are input into the fully connected layer for feature dimension reduction, and the probability values ​​of the classic Bluetooth mode and the low-power Bluetooth mode are calculated through the softmax classifier to obtain the transmission mode switching judgment result.

3. The data packet transmission control method of the low-power dual-mode Bluetooth chip according to claim 2 is characterized in that: The calculating of transmission state evaluation data according to the transmission mode switching determination result, wherein the transmission state evaluation data includes: packet loss rate, transmission delay and real-time power consumption, includes: Dividing the transmission mode switching determination result into a classic Bluetooth mode parameter matrix and a low-power Bluetooth mode parameter matrix, and performing feature combination to obtain a mixed mode feature vector; Segmenting the mixed mode feature vector according to time windows to obtain a data packet statistical matrix; The data packet statistics matrix is ​​classified according to the transmission status, wherein the transmission status includes normal transmission, packet loss retransmission and transmission timeout, and the status counter is used to accumulate statistics to obtain the transmission status counting result; According to the transmission state counting result, the ratio of the number of packet losses in the unit time window to the total number of transmitted data packets is calculated to obtain the packet loss rate; Recording the sending timestamp and receiving timestamp of each data packet in the data packet statistical matrix, calculating the transmission delay of a single data packet by the timestamp difference to obtain a transmission delay sequence, and inputting the transmission delay sequence into a median filter to filter outliers to obtain a transmission delay; The signal transmission power in the mixed mode feature vector is sampled, and combined with the standby current value of the nanoampere LDO circuit, a power consumption sampling sequence is obtained, and the product of the power consumption sampling sequence and the operating voltage is accumulated and converted through a power consumption classification mapping table to obtain real-time power consumption; Transmission status evaluation data is generated based on the data packet loss rate, the transmission delay and the real-time power consumption.

4. The data packet transmission control method of the low-power dual-mode Bluetooth chip according to claim 3 is characterized in that: The chip transmit power is adjusted in stages based on the transmission status evaluation data, and standby power consumption is controlled by using a nanoampere-level LDO circuit to obtain dynamic power control parameters, including: According to the loss rate threshold and the delay threshold, a threshold analysis is performed on the packet loss rate and the transmission delay in the transmission state evaluation data to obtain a power regulation reference matrix; Dividing the power adjustment reference matrix into a high-power transmission state and a low-power transmission state according to the transmission power level, and obtaining a power classification mapping vector; Based on the power classification mapping vector, current detection is performed through a nanoampere-level LDO circuit, the operating voltage of the nanoampere-level LDO circuit is 1.0V, and the standby current threshold of the nanoampere-level LDO circuit is 2μV, so as to obtain a standby power consumption control sequence; Performing timed sampling detection on the standby power consumption control sequence to obtain a power consumption data sequence, and inputting the power consumption data sequence into a high-frequency clock circuit to adjust the operating frequency to obtain a clock control parameter; Performing linear regression analysis on the power consumption data sequence according to the clock control parameters to obtain a power consumption prediction model; The result output by the power consumption prediction model is converted into voltage through a DC / DC circuit to obtain a voltage control sequence, and the voltage control sequence is combined with the power classification mapping vector to obtain dynamic power control parameters by searching a preset dynamic power control table.

5. The data packet transmission control method of the low-power dual-mode Bluetooth chip according to claim 1, characterized in that: The performing of timing sampling based on the dual-mode transmission protocol configuration parameters, calculating the packet time interval, signal strength change rate and link bit error rate, and constructing a dual-mode state feature sequence includes: Performing timing sampling based on the dual-mode transmission protocol configuration parameters to obtain a timing sampling matrix, and performing time interval calculation on the data packet sending time and receiving time in the timing sampling matrix to obtain a time interval statistical sequence; The time interval statistical sequence is input into a high-speed signal sampling unit to sample the signal strength in real time to obtain a signal strength sampling sequence; Performing a differential operation on the signal strength sampling sequence, and obtaining a signal strength change rate matrix based on a signal strength difference between two adjacent sampling points divided by a sampling time interval; The signal strength change rate matrix is ​​grouped and counted according to different sensitivity levels to obtain a signal change characteristic value, and the time interval statistical sequence and the signal change characteristic value are parameter-combined to obtain an error detection sequence; The error detection sequence is input into the CRC check unit, a cyclic redundancy check is performed on each data packet to obtain a link error rate parameter, and feature mapping is performed based on the time interval statistical sequence, the signal change characteristic value and the link error rate parameter to obtain a dual-mode state feature sequence.

6. The data packet transmission control method of the low-power dual-mode Bluetooth chip according to claim 5, characterized in that: The method of calculating the data packet delivery success rate, end-to-end transmission delay and total system power consumption based on the dual-mode state feature sequence, and optimizing the transmission control parameters through double-layer deep reinforcement learning to obtain the target transmission control parameters includes: Inputting the dual-mode state feature sequence into the strategy network layer of the double-layer deep reinforcement learning, the strategy network layer includes a classic Bluetooth strategy generator and a low-power Bluetooth value evaluator, respectively using an Actor network structure and a Critic network structure, to obtain a dual-mode transmission action strategy; The dual-mode transmission action strategy is input into the target network layer of the dual-layer deep reinforcement learning for processing, wherein the target network layer includes a dual-mode target strategy network and a dual-mode target value network, which are responsible for evaluating the target state under different Bluetooth modes to obtain the dual-mode target Q value; Inputting the dual-mode state feature sequence and the dual-mode target Q value into the experience playback buffer, and recording the state transition samples of the classic Bluetooth mode and the low-power Bluetooth mode respectively through a hierarchical storage structure to obtain a dual-mode training sample set; Calculating the timing difference error based on the dual-mode training sample set, taking the transmission power, link quality and bit error rate as evaluation indicators of the reward function, respectively updating the network parameters in the two modes, and obtaining the dual-mode optimization parameters; The dual-mode optimization parameters are soft-updated, and different update rates are set based on the transmission characteristics of different Bluetooth modes, with the update rate of the classic Bluetooth mode being 0.02 and the update rate of the low-power Bluetooth mode being 0.01, to obtain updated dual-mode target parameters; Applying the updated dual-mode target parameters to the current transmission state, dynamically selecting the Bluetooth operating mode according to the data packet transmission priority and power consumption requirements, and obtaining an adaptive transmission control strategy; The dynamic power control parameters, sensitivity threshold and compression coding parameters of the dual-mode Bluetooth chip are generated based on the adaptive transmission control strategy, and the strategy is converted into target transmission control parameters through a parameter mapping table.

7. A data packet transmission control device for a low-power dual-mode Bluetooth chip, characterized in that: The method for controlling data packet transmission of a low-power dual-mode Bluetooth chip according to any one of claims 1 to 6 is used, wherein the data packet transmission control device of the low-power dual-mode Bluetooth chip comprises: The acquisition module is used to obtain the signal strength, link quality index, standby current, working current, data packet throughput and transmission delay of the dual-mode Bluetooth chip, and input it into the deep time convolution network for timing feature extraction to obtain the transmission mode switching judgment result; A calculation module, used to calculate transmission state evaluation data according to the transmission mode switching determination result, wherein the transmission state evaluation data includes: packet loss rate, transmission delay and real-time power consumption; An adjustment module, used to perform hierarchical adjustment on the chip transmission power based on the transmission status evaluation data, and to control the standby power consumption by using a nanoampere-level LDO circuit to obtain a dynamic power control parameter; The coding module is used to set the radio frequency sensitivity threshold according to the dynamic power control parameter, and compress and encode the data packet to obtain the dual-mode transmission protocol configuration parameter; specifically comprising: classifying the dynamic power control parameter according to the classic Bluetooth mode and the low-power Bluetooth mode to obtain the radio frequency sensitivity parameter matrix; inputting the radio frequency sensitivity parameter matrix into the received signal strength indicator, sampling and analyzing the radio frequency signal to obtain the signal strength distribution sequence; performing automatic gain control on the signal strength distribution sequence to obtain the gain control parameter, and segmenting the data packet according to the gain control parameter, dividing the data packet into a frame header, a control field and a data field, and setting different compression ratios for different fields to obtain the segmented compression compression strategy; applying the segmented compression strategy to the data packet encoding process, compressing the control field by 2:1 through the displacement compression algorithm, and compressing the data field by 4:1 through the run-length coding algorithm to obtain a compressed encoded data stream; reorganizing the compressed encoded data stream into a frame format, encapsulating the compressed control field and data field according to the Bluetooth protocol specification, adding a synchronization word and a CRC check code, and obtaining a protocol frame sequence; mapping and associating the protocol frame sequence with the RF sensitivity parameter matrix, generating a mapping table including the transmission power level, the receiving sensitivity level and the data packet format, and obtaining a transmission parameter configuration table; prioritizing the parameters in the transmission parameter configuration table to obtain dual-mode transmission protocol configuration parameters; A sampling module, used to perform timing sampling based on the dual-mode transmission protocol configuration parameters, calculate the packet time interval, signal strength change rate and link bit error rate, and construct a dual-mode state feature sequence; The optimization module is used to calculate the data packet delivery success rate, end-to-end transmission delay and total system power consumption based on the dual-mode state feature sequence, and optimize the transmission control parameters through double-layer deep reinforcement learning to obtain the target transmission control parameters.

8. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and is characterized in that when the processor executes the computer program, the data packet transmission control method of the low-power dual-mode Bluetooth chip described in any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the processor executes the data packet transmission control method for a low-power dual-mode Bluetooth chip according to any one of claims 1 to 6.

Images