Time measurement error calibration method and device and electronic equipment

Abstract

The invention relates to the technical field of network positioning, and provides a time measurement error calibration method and device and electronic equipment, and the method comprises the steps: receiving ranging signals transmitted by other nodes in a positioning network; extracting a pre-selected channel parameter from the ranging signal; the channel parameters are input into a pre-established time measurement error model to obtain time measurement errors, the time measurement error model associates probability distribution of the time measurement errors with the channel parameters based on a Gaussian process regression method, and parameters of the time measurement error model are obtained based on sample data training; and completing calibration according to the time measurement error to obtain a time measurement result. According to the method, the distance measurement precision of the wireless network positioning system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of network positioning, and in particular to a timing error calibration method, device and electronic equipment. Background Art

[0002] Real-time, high-precision positioning technology is an important foundation for the Internet of Things, smart factories, autonomous driving, search and rescue, and other fields. At present, the commonly used global satellite navigation system (GNSS) can achieve meter-level positioning accuracy in open environments. However, satellite signals are blocked by buildings such as buildings, bridges, and tunnels. Therefore, the positioning accuracy and stability of satellite navigation technology in urban environments are poor, and it cannot be used in indoor and underground scenarios. In scenarios where satellite navigation is limited, network positioning technology based on wireless signals is an effective alternative with broad application prospects.

[0003] Wireless network positioning systems usually measure the sending and receiving time of wireless signals between various network nodes to obtain the flight time of signals between nodes. Therefore, given the propagation speed of the signal, the distance between the nodes can be determined, and then the relative position of the nodes can be calculated. Therefore, the ultra-wideband ranging and positioning accuracy of the wireless network positioning system depends on the measurement accuracy of the signal sending and receiving time. Among various wireless measurement technologies, ultra-wideband technology can achieve higher timing accuracy by virtue of its large bandwidth advantage, and has therefore been widely used in the field of wireless network positioning.

[0004] However, in practice, the system is still affected by timing errors from various sources. Specifically, the system's timing errors include the deviation of the device's built-in clock, as well as errors caused by factors such as non-line-of-sight effects, multipath effects, and antenna delays during wireless signal propagation. In order to improve the system's ranging and positioning accuracy, it is necessary to eliminate the errors through a calibration algorithm. Summary of the invention

[0005] The present invention provides a timing error calibration method, device and electronic equipment, which are used to eliminate the influence of the timing error in a wireless network positioning system, and improve the ranging and positioning accuracy of the wireless network positioning system.

[0006] The present invention provides a timing error calibration method, comprising: receiving ranging signals sent by other nodes in a positioning network; extracting pre-selected channel parameters from the ranging signals; inputting the channel parameters into a pre-established timing error model to obtain a timing error, wherein the timing error model associates the probability distribution of the timing error with the channel parameters based on a Gaussian process regression method, and the parameters of the timing error model are obtained by training based on sample data; and completing calibration according to the timing error to obtain a timing result.

[0007] A timing error calibration method provided by the present invention is characterized in that the timing error includes a first error introduced by a wireless propagation process and a second error generated by the node itself, and the second error is a constant; the training strategy of the timing error model includes an offline training strategy, and the training steps of the offline training strategy include: based on a Gaussian process regression method, the probability distribution of the first error is associated with a channel parameter to obtain a part of the timing error model related to the first error; according to the part of the timing error model related to the first error and the second error, a variational lower bound of a ranging error likelihood function is constructed; based on the sample data, by maximizing the variational lower bound of the ranging error likelihood function, the parameters of the part of the first error and the second error are determined to obtain the timing error model.

[0008] A timing error calibration method provided by the present invention is characterized in that the step of acquiring sample data includes: placing the node to be calibrated in the positioning network at a fixed position in a preset area; acquiring the ranging results of each node to be calibrated in the positioning network through an ultra-wideband device equipped with the node to be calibrated in the positioning network, and recording the channel parameters associated with the acquired ranging results as sample channel parameters; calibrating the accurate position of the node to be calibrated in the positioning network through motion capture technology; calculating the accurate distance between each node to be calibrated in the positioning network according to the calibrated accurate position; and determining the sample ranging error based on the accurate distance between each node to be calibrated in the positioning network and the ranging result.

[0009] A timing error calibration method provided by the present invention is characterized in that the training strategy of the timing error model also includes an online training strategy, and the online training strategy is used to determine the second error of the newly added node in the positioning network. The training steps of the online training strategy include: establishing a likelihood function of the ranging result of the newly added node with respect to the node position of the newly added node and the second error; placing the newly added node at a fixed position in a preset area; obtaining the ranging result of the newly added node through an ultra-wideband device equipped with the newly added node, and recording the channel parameters associated with the obtained ranging result; and estimating the node position of the newly added node and the value of the second error by maximizing the established likelihood function according to the ranging result of the newly added node.

[0010] A timing error calibration method provided by the present invention is characterized in that the channel parameters include receiving power and a first path signal-to-noise ratio.

[0011] A timing error calibration method provided by the present invention is characterized in that the timing error model is stored locally in the node.

[0012] A timing error calibration method provided by the present invention is characterized in that the method further includes: sending the timing result to other nodes in the positioning network based on a ranging protocol; obtaining the timing result sent by other nodes in the positioning network, and calculating the positioning result through a pre-set positioning algorithm.

[0013] The present invention also provides a timing error calibration device, comprising: a receiving module, configured to receive ranging signals sent by other nodes in a positioning network; an extraction module, configured to extract channel parameters from the ranging signals; an input module, configured to input the channel parameters into a pre-established timing error model to obtain a timing error, wherein the timing error model associates the probability distribution of the timing error with the channel parameters based on a Gaussian process regression method, and the parameters of the timing error model are obtained by training based on sample data; and a calibration module, configured to complete calibration according to the timing error to obtain a timing result.

[0014] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any of the timing error calibration methods described above is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the timing error calibration method described in any one of the above methods is implemented.

[0016] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the timing error calibration method described above is implemented.

[0017] The timing error calibration method, device and electronic equipment provided by the present invention extract pre-selected channel parameters from the ranging signal, then input the channel parameters into a pre-established timing error model to obtain the timing error, and finally complete the calibration according to the timing error to obtain the timing result, thereby improving the ranging accuracy of the wireless network positioning system. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 This is one of the flow charts of the timing error calibration method provided by the present invention.

[0020] Figure 2It is a flow chart of the offline training strategy of the timing error model provided by the present invention.

[0021] Figure 3 It is a flow chart of the online training strategy of the timing error model provided by the present invention.

[0022] Figure 4 It is a schematic diagram of the distributed operation flow of the timing error calibration method provided by the present invention.

[0023] Figure 5 It is a schematic diagram of the relationship between the probability distribution of timing error and the received power given by the Gaussian process regression model provided by the present invention.

[0024] Figure 6 It is a schematic diagram of the relationship between the probability distribution of the timing error given by the Gaussian process regression model provided by the present invention and the first path signal-to-noise ratio.

[0025] Figure 7 It is a structural schematic diagram of the timing error calibration device provided by the present invention.

[0026] Figure 8 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] Combine the following Figure 1-Figure 8 The present invention describes a timing error calibration method, device and electronic equipment.

[0029] Figure 1 FIG. 1 is a flow chart of a timing error calibration method provided by an embodiment of the present invention. Figure 1 As shown, the method includes: Step 101: Receive ranging signals sent by other nodes in the positioning network.

[0030] In this embodiment, the nodes in the positioning network can receive ranging signals sent by other nodes in the positioning network. The nodes in the positioning network can adopt various wireless measurement technologies, such as ultra-wideband technology, to receive ranging signals sent by other nodes in the positioning network. Among various wireless measurement technologies, ultra-wideband technology can achieve higher timing accuracy by virtue of its large bandwidth advantage.

[0031] Step 102: extract pre-selected channel parameters from the ranging signal.

[0032] In this embodiment, the nodes in the positioning network can extract pre-selected channel parameters from the ranging signal. The channel parameters may include received power, first path signal-to-noise ratio, multipath components, arrival angle and other parameters that may affect the timing error.

[0033] Step 103: Input the channel parameters into a pre-established timing error model to obtain the timing error.

[0034] In this embodiment, the timing error model associates the probability distribution of the timing error with the channel parameters based on the Gaussian process regression method. Gaussian process regression (GPR) is a probability-based non-parametric machine learning method used to solve regression problems. It maps the function to a prior distribution on the Gaussian process, and then updates the prior according to the observed data to obtain the posterior distribution. The parameters of the timing error model are obtained based on sample data training. The sample data may include sample channel parameters and sample timing errors corresponding to the sample channel parameters, and may also include sample channel parameters and sample ranging results or sample timing results corresponding to the sample channel parameters. The sample ranging error or sample timing error can be obtained through the sample ranging results or sample timing results.

[0035] Step 104: complete calibration according to the timing error to obtain the timing result.

[0036] Specifically, the timing error output by the trained timing error model and the actual arrival time of the local received ranging signal can be used to predict the timing error. In addition, in order to further improve the ranging accuracy, the clock error can also be considered to determine the final timing result.

[0037] The timing error calibration method provided by the present invention extracts pre-selected channel parameters from the ranging signal, then inputs the channel parameters into a pre-established timing error model to obtain the timing error, and finally completes the calibration based on the timing error to obtain the timing result, thereby improving the ranging accuracy of the wireless network positioning system.

[0038] In some optional implementations, the timing error includes a first error introduced by a wireless propagation process and a second error generated by the node itself, and the second error is a constant; the training strategy of the timing error model includes an offline training strategy, and the training steps of the offline training strategy include: based on a Gaussian process regression method, associating the probability distribution of the first error with a channel parameter to obtain a part of the timing error model related to the first error; constructing a variational lower bound of a ranging error likelihood function based on the part of the first error and the second error in the timing error model; based on sample data, determining the parameters of the part of the first error and the second error by maximizing the variational lower bound of the ranging error likelihood function to obtain a timing error model.

[0039] In statistics and machine learning, maximizing a variational lower bound on the likelihood function of the range error is a method for optimizing model parameters, especially when dealing with complex probabilistic models. This method allows estimating model parameters in a more stable and computationally efficient way. The likelihood function is a probabilistic estimate of the model parameters based on the observed data. A variational lower bound is a simplified lower bound on the likelihood function that is usually easier to optimize. A variational lower bound approximates the original likelihood function by introducing auxiliary variables (variational parameters).

[0040] In this implementation, the error is further subdivided into the error introduced by the wireless propagation process that is only related to the channel parameters and is independent of the nodes, i.e., the first error, and the error constant that is independent of the channel parameters and represents the difference introduced by different nodes in the production process, i.e., the second error, so that the constructed model is more accurate. And the parameters of the first error part in the timing error model can be reused on other nodes, further improving the efficiency of the system.

[0041] In some optional implementations, the step of acquiring sample data includes: placing the node to be calibrated in the positioning network at a fixed position in a preset area; acquiring the ranging results of each node to be calibrated in the positioning network through an ultra-wideband device equipped with the node to be calibrated in the positioning network, and recording the channel parameters associated with the acquired ranging results as sample channel parameters; calibrating the exact position of the node to be calibrated in the positioning network through motion capture technology; calculating the exact distance between each node to be calibrated in the positioning network according to the calibrated exact position; and determining the sample ranging error based on the exact distance between each node to be calibrated in the positioning network and the ranging results.

[0042] In this implementation, the node to be calibrated is the node whose timing error model parameters are to be determined. The preset area may include the coverage area of ​​the positioning network. If multiple sets of data are required, the position of the node to be calibrated may be changed, and the above steps of obtaining sample data may be repeated.

[0043] In some optional implementations, the channel parameters include received power and a first path signal-to-noise ratio. In this case, the construction process of the timing error model may be as follows: When the UWB device equipped by the node receives a ranging signal, it will collect a set of channel impulse response sequences. Assume that the channel impulse response sequence contains sampling values, of which the The sampling value is , then the received power of the signal It can be expressed as: ,

[0044] The first path signal-to-noise ratio It can be expressed as: ,

[0045] in, and is a parameter determined by the UWB hardware, is the characteristic amplitude parameter given by the rising edge detection algorithm one, It is the standard deviation of the received signal noise given by the rising edge detection algorithm. Since the two channel parameters of received power and first path signal-to-noise ratio can better reflect the impact of channel factors on ultra-wideband timing error, the present invention models the timing error based on these two channel parameters.

[0046] On this basis, consider a The positioning network of nodes. Assuming that the node At the moment Received from node The received power of the signal and the first path signal-to-noise ratio are recorded as , then the node Measurement of reception time It can be modeled as: ,

[0047] in, Characterizing clock errors, efficient calibration can be achieved through existing ranging protocols, so the method proposed in this invention assumes The calibration has been completed and this error is no longer considered. represents the channel error, which is the error modeled and calibrated by the proposed method.

[0048] Furthermore, the channel error can be modeled as: ,

[0049] in, , Represents the error introduced by the wireless propagation process, which is only related to the channel parameters but not to the nodes; represents the error constant introduced by the differences between different nodes in the production process and has nothing to do with the channel parameters. Here, we define .

[0050] Furthermore, we can This breaks down into: ,

[0051] in, represent system deviation and random noise respectively.

[0052] On this basis, assuming that the noise satisfies a zero-mean Gaussian distribution with a variance of It is also related to the channel parameters, namely, Then, based on the Gaussian process regression method, we can determine specific form.

[0053] First, define an implicit function , so that it satisfies Then, the variational Gaussian process regression model was used to analyze Modeling.

[0054] Specifically, the introduction of the induction point With induced variables , and are Introducing variational distribution .

[0055] In addition, define the kernel function: ,

[0056] In the above definition, the vector The dimension and consistent, and are the model parameters to be determined. For simplicity, we define .

[0057] On this basis, assuming that the input channel parameters ,but The probability distribution of can be modeled as: ,

[0058] in, , is defined as follows: ,

[0059] In the above formula, ,vector With the matrix satisfy .

[0060] Here, we define To express So far, the present invention associates the probability distribution of the timing error with the channel parameters and completes the modeling of the timing error.

[0061] The values ​​of the parameters in the timing error model can be determined based on the measured data of the ultra-wideband positioning system. The timing error model can use two training strategies: offline training and online training. If the parameters of the timing error model are unknown, an offline training strategy is required to obtain the optimal estimate of the complete parameters of the error model after the accurate position of the network nodes is calibrated. If the timing error model is The parameters included (the parameters of the timing error model for the first error, i.e. ) is known, and the error constant associated with the network node (second error) is unknown, an online training strategy can be used to determine the The best estimate of . The parameter values ​​of can be applied to all nodes, and the online training strategy can efficiently extend the error model obtained by the offline training strategy to newly added uncalibrated nodes, thereby greatly reducing the cost and workload of calibrating the node position.

[0062] Taking the training of the calibration algorithm under the SM-NR ranging protocol as an example, for a In a fully connected positioning network of nodes, the SM-NR protocol can obtain the ranging results between any two nodes in a ranging cycle by having each node broadcast the ranging signal in turn, while eliminating clock-type timing errors in the measurement process.

[0063] In the SM-NR protocol, nodes The distance measurement error between It can be approximated as: ,

[0064] in, is the propagation speed of wireless signals, and Represents nodes The measurement error of the signal sending time. Mainly caused by the antenna delay at the transmitter, it has nothing to do with the channel parameters, so it can be set , and Incorporated as a constant among.

[0065] For the offline training strategy, since the true value of the timing error is difficult to measure through experimental methods, the present invention calculates the true value of the ranging error between each node pair based on the true position of the network node, and estimates the optimal parameters of the timing error model by maximizing the variational lower bound of the ranging error likelihood function. For simplicity, considering the measurement results within a ranging cycle, the set of true values ​​of the ranging error is defined as ,but The variational lower bound of the log-likelihood function It can be written as: ,

[0066] in, , , The definition of Consistent.

[0067] Hence, see Figure 2 , Figure 2 It is a flow chart of the offline training strategy of the timing error model provided by the present invention. Figure 2 First, the variational lower bound of the ranging error likelihood function is constructed from the timing error model. The timing error model can include the part about the first error and the second error. The implementation process of the offline training strategy can be described as follows: First, the network node to be calibrated is placed at a certain position, several cycles of ultra-wideband measurement are run, and the exact position of the network node is calibrated using the motion capture system to calculate the true value of the ranging error; then, in order to collect as many sets of ranging error data as possible under different channel parameters to improve the generalization performance of the error model, it is necessary to change the position of the node, repeat the ultra-wideband measurement and node position calibration. After collecting enough data, it can be maximized. The optimal error model parameters are obtained by: ,

[0068] Thereby completing the task of offline training.

[0069] In some optional implementations, the training strategy of the timing error model also includes an online training strategy, which is used to determine the second error of the newly added node in the positioning network, and the training steps of the online training strategy include: establishing a likelihood function of the distance measurement result of the newly added node about the node position of the newly added node and the second error; placing the newly added node at a fixed position in a preset area; obtaining the distance measurement result of the newly added node through the ultra-wideband device equipped with the newly added node, and recording the channel parameters associated with the obtained distance measurement result; according to the distance measurement result of the newly added node, by maximizing the established likelihood function, estimating the node position of the newly added node and the value of the second error. In this implementation, the online training strategy can determine the parameters of the error model without calibrating the exact position of the node, which reduces the workload of node calibration.

[0070] See also Figure 3 , Figure 3 It is a flow chart of the online training strategy of the timing error model provided by the present invention. The online training strategy is applicable to the case where the offline training strategy is extended to newly added network nodes: assuming that the optimal error model parameters have been obtained for a batch of network nodes through the offline calibration strategy, and there are another batch of newly added nodes to be calibrated, the online training strategy can be used to complete the calibration of the newly added nodes.

[0071] In this case, it can be assumed that the parameters Known Unknown; In addition, assume that the number of network nodes is ,node The location is , , the positions of all nodes are fixed and unknown. The goal of the online training strategy is to obtain the error model parameters based on the UWB measurement results and the given error model parameters. , estimate the model parameters With node position The optimal value of .

[0072] First, set the node The uncalibrated ultra-wideband ranging result is , and define the set of ranging results as ,but The likelihood function can be written as: ,

[0073] in,

[0074] ,

[0075] and All can be obtained through the given Therefore, the error model parameters and node positions can be obtained by maximizing the log-likelihood: .

[0076] Specifically, for the above non-convex optimization problem, we can first ignore In the case of multidimensional scaling (MDS) method, we give The initial estimate of ; then using the gradient descent method, estimate The optimal value of .

[0077] In some optional implementations, the timing error model is stored locally at the node, so that model parameters are configured for each node instead of each node pair, thereby reducing the computational and storage complexity of the calibration method.

[0078] In some optional implementations, the method further includes: sending the timing results to other nodes in the positioning network based on the ranging protocol; obtaining the timing results sent by other nodes in the positioning network, and calculating the positioning results through a pre-set positioning algorithm. The ranging protocol can use the signal multiplexing network ranging (SM-NR) protocol, or the two-way ranging (TWR) or time difference measurement (TDoA) protocol. The positioning algorithm can use triangulation positioning, hyperbolic positioning, multilateral positioning and other methods.

[0079] Existing methods often adopt a centralized operation strategy, which requires the channel parameters collected by each network node to be aggregated to the central processor for unified calibration, which brings a large communication overhead. The distributed operation strategy adopted by this implementation avoids the additional communication overhead introduced by the existing centralized calibration method, and at the same time evenly distributes the calibration calculation burden to each network node, thereby improving the operation efficiency. Therefore, it can achieve efficient operation in a large-scale positioning network.

[0080] See also Figure 4 , Figure 4 This is a schematic diagram of the distributed operation process of the timing error calibration method provided by the present invention. Two network nodes i and j in the network positioning system are selected as examples. Each node in the network positioning system receives ranging signals from other nodes, and then calculates channel parameters and original timing results based on these signals, and then calibrates the timing results through the timing error model, and finally obtains the calibrated ranging and positioning results through clock synchronization, ranging and positioning algorithms. In addition, Figure 4 The operation process of the timing error model is also shown, including the channel impulse response waveform of the received signal, extracting parameters such as the received power and the first path signal-to-noise ratio, calculating the signal arrival time as the timing result, estimating the mean and variance of the timing error based on the Gaussian process regression model, and estimating the calibrated timing result and its variance by subtracting the original timing result from the timing error to obtain the calibrated timing result.

[0081] Specifically, after obtaining the timing error model and the optimal model parameters, the error model can be run in the network positioning system to calibrate the timing error and improve the ranging and positioning accuracy of the system. Specifically, first, before the network positioning system starts running, all network nodes save the timing error model and its parameters obtained in the first two steps locally. Subsequently, during the operation of the system, for any network node, whenever it receives a ranging signal, the node will measure the arrival time of the signal. , and extract parameters such as received power and first path signal-to-noise ratio from the channel impulse response of the signal , input it into the timing error model, and obtain the estimated value of the timing error, whose mean is , the variance is Based on this, the calibrated timing result can be obtained. The probability distribution that satisfies: .

[0082] After that, according to the design of the ranging protocol, when the node needs to send a ranging signal to other nodes, By sending it together with the ranging signal, the calibration result of the node can be shared with other nodes; when the ranging and positioning results need to be calculated, each node can obtain the calibrated ranging and positioning results based on the calibrated timing results. In the above method, when calculating the calibrated timing results, each node only needs to use the channel parameters collected by the node Therefore, the calculation work of timing error calibration can be completely performed locally on each node, without the need for a centralized processor to summarize the channel parameters collected by each node, thereby achieving distributed operation.

[0083] The embodiments of the present invention provide a distributed operation scheme of the calibration algorithm under the SM-NR ranging protocol. In the original SM-NR ranging protocol, for Node positioning network, each ranging cycle contains Second measurement: In this measurement, the node Broadcast a ranging signal to all other nodes and send the time of sending this signal With Node The arrival times of all previously received signals Send it to other nodes at the same time; after receiving the signal, other nodes measure the arrival time of this signal and save. In this measurement, node 1 broadcasts a ranging signal again, and then the ranging and positioning results can be calculated based on all the timing results. After adding the proposed calibration method, each node calculates the calibrated signal arrival time locally through the timing error model after receiving the signal. ; in the In this measurement, the node Sent to other nodes Replace with calibrated Here, the node You can choose to send only The average value of can significantly improve the ranging and positioning accuracy of the system without introducing additional communication overhead; you can also choose to send The mean and variance of the system are calculated, which further improves the positioning accuracy of the system while increasing a certain communication overhead. After the ranging cycle is completed, the ranging and positioning results can be calculated based on the calibrated timing results.

[0084] Figure 5 It is a schematic diagram of the relationship between the probability distribution of timing error and the received power given by the Gaussian process regression model provided by the present invention. Figure 5 The X-axis represents the received power in decibels (dBm), and the Y-axis represents the timing error in nanoseconds (ns). Figure 6 It is a schematic diagram of the relationship between the probability distribution of the timing error given by the Gaussian process regression model provided by the present invention and the first path signal-to-noise ratio. Figure 6 The X-axis represents the first-path signal-to-noise ratio in decibels (dB), and the Y-axis represents the timing error in nanoseconds (ns).

[0085] Figure 7 A schematic diagram of the structure of the timing error calibration device provided in the embodiment of the present application is shown in FIG. Figure 7 As shown, it specifically includes: a receiving module 701, configured to receive ranging signals sent by other nodes in the positioning network; an extraction module 702, configured to extract channel parameters from the ranging signals; an input module 703, configured to input the channel parameters into a pre-established timing error model to obtain a timing error, the timing error model associates the probability distribution of the timing error with the channel parameters based on a Gaussian process regression method, and the parameters of the timing error model are obtained by training based on sample data; and a calibration module 704, configured to complete calibration according to the timing error to obtain a timing result.

[0086] In some optional implementations, the timing error includes a first error introduced by a wireless propagation process and a second error generated by the node itself, and the second error is a constant; the training strategy of the timing error model includes an offline training strategy, and the training steps of the offline training strategy include: based on a Gaussian process regression method, associating the probability distribution of the first error with a channel parameter to obtain a part of the timing error model related to the first error; constructing a variational lower bound of a ranging error likelihood function based on the part of the first error and the second error in the timing error model; based on sample data, determining the parameters of the part of the first error and the second error by maximizing the variational lower bound of the ranging error likelihood function to obtain a timing error model.

[0087] In some optional implementations, the step of acquiring sample data includes: placing the node to be calibrated in the positioning network at a fixed position in a preset area; acquiring the ranging results of each node to be calibrated in the positioning network through an ultra-wideband device equipped with the node to be calibrated in the positioning network, and recording the channel parameters associated with the acquired ranging results as sample channel parameters; calibrating the exact position of the node to be calibrated in the positioning network through motion capture technology; calculating the exact distance between each node to be calibrated in the positioning network according to the calibrated exact position; and determining the sample ranging error based on the exact distance between each node to be calibrated in the positioning network and the ranging results.

[0088] In some optional implementations, the training strategy of the timing error model also includes an online training strategy, which is used to determine the second error of a newly added node in the positioning network, and the training steps of the online training strategy include: establishing a likelihood function of the ranging result of the newly added node with respect to the node position of the newly added node and the second error; placing the newly added node at a fixed position in a preset area; obtaining the ranging result of the newly added node through an ultra-wideband device equipped with the newly added node, and recording the channel parameters associated with the obtained ranging result; based on the ranging result of the newly added node, estimating the node position of the newly added node and the value of the second error by maximizing the established likelihood function.

[0089] In some optional implementations, the channel parameters include received power and a first path signal-to-noise ratio.

[0090] In some optional implementations, the timing error model is stored locally at the node.

[0091] In some optional implementations, the device also includes a positioning module, which is configured to: send the timing results to other nodes in the positioning network based on the ranging protocol; obtain the timing results sent by other nodes in the positioning network, and calculate the positioning results through a pre-set positioning algorithm.

[0092] In this embodiment, the specific processing of the receiving module 701, the extraction module 702, the input module 703 and the calibration module 704 of the timing error calibration device can be referred to. Figure 1 This corresponds to step 101, step 102, step 103 and step 104 in the embodiment.

[0093] The timing error calibration device provided by the present invention extracts pre-selected channel parameters from the ranging signal, then inputs the channel parameters into a pre-established timing error model to obtain the timing error, and finally completes the calibration according to the timing error to obtain the timing result, thereby improving the ranging accuracy of the wireless network positioning system.

[0094] Figure 8 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 8 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the timing error calibration method, which includes: receiving ranging signals sent by other nodes in the positioning network; extracting pre-selected channel parameters from the ranging signals; inputting the channel parameters into a pre-established timing error model to obtain the timing error, the timing error model associates the probability distribution of the timing error with the channel parameters based on the Gaussian process regression method, and the parameters of the timing error model are obtained by training based on sample data; and completing the calibration according to the timing error to obtain the timing result.

[0095] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of software function modules and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a 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, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0096] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the timing error calibration method provided by the above methods, and the method includes: receiving ranging signals sent by other nodes in the positioning network; extracting pre-selected channel parameters from the ranging signals; inputting the channel parameters into a pre-established timing error model to obtain a timing error, the timing error model associates the probability distribution of the timing error with the channel parameters based on a Gaussian process regression method, and the parameters of the timing error model are obtained based on training of sample data; and completing calibration according to the timing error to obtain a timing result.

[0097] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the timing error calibration method provided by the above-mentioned methods, the method comprising: receiving ranging signals sent by other nodes in the positioning network; extracting pre-selected channel parameters from the ranging signals; inputting the channel parameters into a pre-established timing error model to obtain a timing error, the timing error model associating the probability distribution of the timing error with the channel parameters based on a Gaussian process regression method, the parameters of the timing error model being obtained through training based on sample data; and completing calibration based on the timing error to obtain a timing result.

[0098] The device embodiments described above are merely illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative effort.

[0099] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. 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 timing error calibration method, characterized in that: include: Receive ranging signals sent by other nodes in the positioning network; extracting pre-selected channel parameters from the ranging signal; Inputting the channel parameters into a pre-established timing error model to obtain a timing error, wherein the timing error model associates a probability distribution of the timing error with the channel parameters based on a Gaussian process regression method, and the parameters of the timing error model are obtained by training based on sample data, wherein the sample data includes sample channel parameters and sample ranging errors corresponding to the sample channel parameters; Calibration is performed according to the timing error to obtain a timing result.

2. The timing error calibration method according to claim 1, characterized in that: The timing error includes a first error introduced by the wireless propagation process and a second error generated by the node itself, wherein the second error is a constant; the training strategy of the timing error model includes an offline training strategy, and the training steps of the offline training strategy include: Based on a Gaussian process regression method, the probability distribution of the first error is associated with a channel parameter to obtain a part of the timing error model related to the first error; Constructing a variational lower bound of a ranging error likelihood function according to a portion of the timing error model related to the first error and the second error; Based on the sample data, the parameters of the first error portion and the second error are determined by maximizing the variational lower bound of the ranging error likelihood function to obtain the timing error model.

3. The timing error calibration method according to claim 1, characterized in that: The step of obtaining the sample data comprises: Placing the node to be calibrated in the positioning network at a fixed position within a preset area; Obtaining the ranging results of each node to be calibrated in the positioning network through an ultra-wideband device equipped by the node to be calibrated in the positioning network, and recording the channel parameters associated with the obtained ranging results as sample channel parameters; Calibrate the exact position of the node to be calibrated in the positioning network by motion capture technology; Calculating the accurate distance between the nodes to be calibrated in the positioning network according to the calibrated accurate positions; A sample ranging error is determined based on the accurate distances between the nodes to be calibrated in the positioning network and the ranging result.

4. The timing error calibration method according to claim 2, characterized in that: The training strategy of the timing error model further includes an online training strategy, which is used to determine the second error of the newly added node in the positioning network. The training steps of the online training strategy include: Establishing a likelihood function of the distance measurement result of the newly added node with respect to the node position of the newly added node and the second error; Placing the newly added node at a fixed position within a preset area; Obtaining a ranging result of the newly added node through an ultra-wideband device equipped with the newly added node, and recording a channel parameter associated with the obtained ranging result; According to the distance measurement result of the newly added node, the node position of the newly added node and the value of the second error are estimated by maximizing the established likelihood function.

5. The timing error calibration method according to claim 1, characterized in that: The channel parameters include receiving power and a first path signal-to-noise ratio.

6. The timing error calibration method according to claim 1, characterized in that: The timing error model is stored locally in the node.

7. The timing error calibration method according to claim 1, characterized in that: The method further comprises: Sending the timing result to other nodes in the positioning network based on a ranging protocol; The timing results sent by other nodes in the positioning network are obtained, and the positioning result is calculated by using a preset positioning algorithm.

8. A timing error calibration device, characterized in that: include: A receiving module, configured to receive ranging signals sent by other nodes in the positioning network; An extraction module, configured to extract channel parameters from the ranging signal; An input module is configured to input the channel parameters into a pre-established timing error model to obtain a timing error, wherein the timing error model associates a probability distribution of the timing error with the channel parameters based on a Gaussian process regression method, and the parameters of the timing error model are obtained by training based on sample data, wherein the sample data includes sample channel parameters and sample ranging errors corresponding to the sample channel parameters; The calibration module is configured to complete calibration according to the timing error to obtain a timing result.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the timing error calibration method according to any one of claims 1 to 7 when executing the program.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the timing error calibration method according to any one of claims 1 to 7 is implemented.

Images