OTFS signal-based high-precision target parameter estimation method for communication and induction integrated system

Abstract

The invention discloses an OTFS signal-based high-precision target parameter estimation method for a common-inductance integrated system, which is based on a recursive optimization super-resolution estimation algorithm, combines a grid refining method and a recursive optimization strategy, and adopts a two-stage estimation framework. In the first stage, super-resolution estimation of time delay and Doppler frequency shift is realized through a grid refinement technology, and the estimation precision is improved; and in the second stage, a multi-path interference recursive optimization mechanism is combined, so that the influence of a multi-path effect on estimation precision is effectively reduced, and the reliability and robustness of estimation are further improved. Under the condition of processing fractional time delay and Doppler frequency shift, the method provided by the invention obviously improves the estimation precision in a complex channel environment, and can effectively solve the problem of parameter estimation under fractional time delay and Doppler effect.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar signal processing, and in particular relates to a high-precision target parameter estimation method for a synaesthesia system based on OTFS signals. Background Art

[0002] With the development of intelligent transportation, autonomous driving, low-orbit satellites, and integrated air-ground networks, communication and perception fusion (i.e., synaesthesia) systems are becoming a key development direction for future wireless systems. By integrating communication and environmental perception functions into the same hardware platform and signal structure, synaesthesia systems can significantly improve resource utilization, system integration, and perception coverage. Orthogonal Time-Frequency-Space (OTFS) modulation, due to its robustness in highly dynamic and complex multipath environments, is widely recognized as one of the key waveform technologies supporting synaesthesia systems. Especially when faced with challenges such as high-speed moving targets, complex background scattering, and multipath interference, OTFS demonstrates excellent perception performance and anti-interference capabilities thanks to its natural adaptability to Doppler shift and delay spread.

[0003] However, the performance advantage of OTFS in the synaesthesia system depends heavily on the accuracy of target parameter estimation, especially the accurate extraction of physical parameters such as the target's Doppler shift, propagation delay, and echo path gain in dynamic scenarios. These parameters directly affect the accuracy of target positioning, velocity measurement, and tracking, and are the basis for achieving high-performance synaesthesia fusion. Because the actual target parameters are continuous values, traditional discrete grid-based channel estimation algorithms are prone to quantization errors when faced with fractional delay and fractional Doppler shift, and cannot meet the requirements of high-precision parameter estimation. Especially in environments with complex multipath effects, the superposition of multiple scattering paths further increases the difficulty of resolving parameter estimation, becoming a key bottleneck restricting the performance improvement of the synaesthesia system.

[0004] Therefore, how to achieve high-precision estimation of target parameters in high-speed dynamic and multipath interference environments has become one of the core challenges in promoting the implementation of the OTFS synaesthesia system. There is an urgent need to develop parameter estimation algorithms with super-resolution capabilities and strong robustness. Summary of the Invention

[0005] The object of the present invention is to provide a high-precision target parameter estimation method for a synaesthesia system based on OTFS signals.

[0006] The technical solution to achieve the purpose of the present invention is: a high-precision target parameter estimation method for a synaesthesia system based on OTFS signals, comprising the following steps:

[0007] Step 1: Establish the input and output relationship model of OTFS signal in DD domain;

[0008] Step 2: Calculate a rough estimate of the channel parameters using the received pilot signal matrix;

[0009] Step 3: Based on the rough estimate, the grid search range is gradually narrowed so that each round of search can be performed in a smaller area to obtain a finer estimate.

[0010] Step 4: After each path is estimated, perform mutual optimization and interference suppression between paths, and then go to step 2.

[0011] Step 5: After the path parameter estimation is completed, the distance and speed of the target are calculated based on the time delay and Doppler information.

[0012] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the steps of the above method are implemented when the processor executes the program.

[0013] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.

[0014] A computer program product comprises a computer program, which implements the steps of the above method when executed by a processor.

[0015] Compared with the prior art, the present invention has the following significant advantages: (1) The accuracy of the target parameters estimated by the present invention is much higher than that of the traditional method; (2) The present invention solves the problem of large error in fractional delay Doppler parameter estimation in the traditional parameter estimation method; (3) The parameter estimation results of the present invention are less different from the preset results and are more accurate.

[0016] The present invention is further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a pilot scheme diagram of the algorithm of the present invention.

[0018] Figure 2 It is the pseudo code of the refinement estimation algorithm in step 3 of the algorithm of the present invention.

[0019] Figure 3 It is the algorithm flow chart in step 4 of the algorithm of the present invention.

[0020] Figure 4 2 is a comparison chart of channel estimation performance according to an embodiment of the present invention.

[0021] Figure 5 2 is a comparison chart of communication bit error rates according to an embodiment of the present invention.

[0022] Figure 6 2 is a diagram comparing distance estimation errors according to an embodiment of the present invention.

[0023] Figure 7 2 is a speed estimation error comparison diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The present invention proposes a ROSR target parameter estimation algorithm. This algorithm fully combines the grid refinement method with the recursive optimization strategy, and by designing a two-stage estimation framework, it effectively improves the accuracy of channel parameter estimation in complex environments. First, in the first stage, the algorithm uses grid refinement technology to perform high-resolution estimation of delay and Doppler frequency shift, thereby significantly improving the resolution ability of channel parameters. Subsequently, in the second stage, the algorithm further introduces a recursive optimization mechanism to reduce the impact of multipath interference through iterative optimization, thereby enhancing the robustness of channel estimation. Thanks to the design of this method, even in the case of fractional-order delay and Doppler frequency shift, it is still possible to accurately estimate the channel parameters.

[0025] like Figure 1 As shown in Figure 1, the Recursive Optimization-based Super-Resolution (ROSR) estimation algorithm uses a full-frame pilot scheme, which divides the OTFS frame into pilot frames and data frames. In this scheme, the first frame is entirely used for pilot transmission, with all symbols being pilot symbols, while subsequent frames are used for data transmission. A major advantage of this design is that there is no interference between pilot symbols and data symbols, allowing for more accurate channel estimation. By receiving pilot symbols, the system can construct an observation matrix for channel estimation and, combined with a specific channel model, infer the channel state.

[0026] The specific steps of the high-precision target parameter estimation method of the synaesthesia system based on OTFS signals of the present invention are described in detail below. The method includes:

[0027] Step 1: Express the input and output relationship of the OTFS signal in the DD domain as:

[0028]

[0029] in, is the channel response matrix in the DD domain, M is the number of subcarriers, N is the number of symbols, represents the receiver noise, and Represent the output and input signals of DD domain respectively. i represents the channel delay, v i represents the channel Doppler shift, h i represents the channel gain coefficient of the i-th path, Used to describe the delay and Doppler effect of each path.

[0030] Step 2: Perform a rough estimation of the channel parameters. First, transform the expression in step 1 into the following form:

[0031]

[0032] in, It is the product of the channel response matrix and the input signal, which represents the influence of the delay and Doppler frequency shift of the i-th path on the input signal. is a set matrix containing L paths, represents the channel gain of L paths.

[0033] Step 2-1: Convert the channel estimation problem into an optimization problem, namely:

[0034]

[0035] The goal of this formulation is to minimize ||y-Γ(τ,v)h|| 2 .in, is the channel parameter value to be estimated, y is the pilot symbol received by the receiver, Γ(τ,v) is the set matrix of delay and Doppler parameters of all paths, and h is the channel gain matrix.

[0036] Step 2-2: Since the optimization problem in step 2-1 has a three-dimensional search space (h, τ, v), the computational complexity is high. Therefore, to simplify the computational process, we first estimate (τ, v) and then estimate h. Given (τ, v), the objective function can be viewed as a quadratic least squares problem with respect to h. To minimize this quadratic form, we take the derivative with respect to h and set it to zero, i.e.:

[0037]

[0038] Among them, Γ H (τ,v) represents the conjugate matrix of Γ(τ,v).

[0039] From this we get:

[0040]

[0041] Step 2-3, order:

[0042] Ω=y H Γ(τ,v)(Γ H (τ,v)Γ(τ,v)) -1 Γ H (τ,v)y

[0043] Observing the expression of Ω, we can see that it is only related to (τ, v) and has nothing to do with the channel gain h. Therefore, the minimization problem in step 2-1 can be converted into a maximization problem of solving the cost function Ω, that is:

[0044]

[0045] After estimating the time delay and Doppler frequency shift through the above formula, Substitute the result in step 2-2 to obtain the channel gain

[0046]

[0047] Step 3: In the channel estimation process, refinement is a key step to improve the accuracy of delay and Doppler shift. This process is based on the parameters obtained in step 2. Specifically, refinement gradually narrows the range of the grid search so that each round of search can be performed in a smaller area, thereby gradually approaching the optimal solution. The pseudo code of the refinement algorithm is as follows: Figure 2 This strategy of gradually narrowing the search space helps improve the accuracy of the estimation and is the key to improving the accuracy of delay and Doppler parameter estimation.

[0048] Step 3-1. To achieve this goal, a refined search strategy was designed to improve estimation accuracy by finely controlling the search step size at each iteration. The core of this strategy is to dynamically adjust the search range based on the current estimation results at each iteration, thereby gradually approaching the actual delay and Doppler values. During the initialization phase, several key parameters are first set to ensure the convergence and accuracy of the algorithm, thereby ensuring that excessive deviation or convergence instability does not occur during the refined search process. The search step size for delay and Doppler shift at iteration t can be expressed as:

[0049]

[0050] Where M is the number of subcarriers, N is the number of symbols, Δf is the subcarrier spacing, T is the symbol duration, t is the number of iterations, α is the step-size scaling factor for the delay dimension, and β is the step-size scaling factor for the Doppler dimension. As can be seen from the above formula, the search range in the delay and Doppler dimensions is gradually narrowed with each iteration, and the step size is gradually reduced, ensuring that each optimization is performed within a local area, thereby achieving more accurate parameter estimation. This gradual narrowing of the search space helps improve the overall performance of the algorithm and ensures high channel estimation accuracy in complex multipath environments.

[0051] Step 3-2: After determining the search step size, the next step is to construct a two-dimensional grid search space. The core goal of this step is to form a grid structure that can gradually approach the optimal parameters by reasonably dividing the search range of the two dimensions of delay and Doppler shift. The two-dimensional grid not only needs to cover all possible values of delay and Doppler shift, but also needs to ensure the rationality of the grid density, so as to balance the computational complexity and estimation accuracy. The search space Φ at the tth iteration (t) It can be expressed as:

[0052]

[0053] Where, is the result of the tth iteration, and the search range of the delay dimension is Ψ τ and the Doppler search range Ψ v They are:

[0054] Ψ τ ={-α,...,0,…,α}

[0055] Ψ v ={-β,…,0,…,β}

[0056] Step 3-3: After constructing the search space, the algorithm will calculate each search point (τ i ,v i )∈Φ (t) The value of the cost function Ω at . By evaluating all search points, the algorithm can determine the maximum value of the cost function within this range and select the optimal combination of delay and Doppler shift as the new estimate. In this process, the maximum value of the cost function usually corresponds to the optimal estimate of the channel parameters. Therefore, by maximizing the cost function, the algorithm can gradually approach the ideal solution. After completing the current iteration, it is necessary to determine whether the convergence conditions are met. The specific convergence conditions usually include the following two cases:

[0057]

[0058] Where, ε τ and ε v are the convergence thresholds of delay and Doppler frequency shift set in the initialization phase, t max is the upper limit of the number of iterations. First, if the difference between the current estimate and the previous iteration result is less than the set threshold, it indicates that the algorithm is close to convergence and the estimation result is accurate enough. Second, if the number of iterations reaches the preset upper limit and the algorithm still does not converge, it can be considered that the algorithm has completed enough optimization processes and stops further iterations. If any of the above conditions are met, the refinement estimation process will terminate and the current iteration result will be output. As the estimated values of time delay and Doppler shift. If the algorithm has not yet met the convergence criteria, it proceeds to the next iteration. Each iteration re-evaluates the cost function in the updated search space and adjusts the parameter estimates based on the maximized cost function value. Through multiple iterations, the estimates of time delay and Doppler shift gradually converge to the optimal solution.

[0059] Step 4: After the current path is estimated, the algorithm verifies whether the maximum number of paths has been reached. If not, the algorithm enters the path optimization phase before moving on to the next path estimation. If so, the channel estimation result is output.

[0060] After each path is estimated, mutual optimization and interference suppression between paths are performed, which is crucial to improving the accuracy of the overall channel estimation. This optimization process not only relies on the estimation results of the current path, but also needs to combine the estimation information of the previous path to achieve global optimization, further reduce interference, and improve the accuracy of the estimation. The algorithm flow is as follows: Figure 3 shown.

[0061] Step 4-1. During the path estimation process, the residual vector is defined as an important indicator for evaluating the accuracy of the current path estimation. The residual vector reflects the difference between the current path estimation and the true channel model and can effectively quantify the estimation error. Therefore, the residual vector is not only used to evaluate the accuracy of the path estimation, but also serves as a basis for determining whether to further add paths or adjust the estimation accuracy. By calculating the residual of each path, it can be determined whether the predetermined accuracy requirements are met, thereby deciding whether to continue iterative optimization. Assume that the residual vector of the i-th path is ε i , which is defined as follows:

[0062]

[0063] The estimation of each path depends not only on its own parameters but also on the influence of other paths. In a multipath propagation environment, mutual interference between paths can lead to the accumulation of estimation errors, significantly affecting the accuracy of the overall channel estimation. To effectively address this issue, after each path is estimated, an optimization mechanism is required to reduce the interference between paths. This ensures that the estimates of each path complement each other while improving accuracy, thereby optimizing the overall channel estimation.

[0064] Step 4-2, if And i<L max , which means that the current residual change exceeds the set threshold And the maximum number of path estimates L has not been reached max , if the current newly estimated path cannot make the residual change lower than the set threshold This indicates that additional paths may have been missed and the channel model is not yet complete. This indicates that the currently estimated number of paths is insufficient to fully account for all path components in the received signal, and that additional unestimated paths may exist. Therefore, the algorithm continues to increase the number of paths. Before estimating the parameters of the next path, the algorithm first optimizes the parameters of each estimated path. Specifically, the algorithm fills the corresponding columns of the matrix Γ(τ, v) with the estimated values of other paths and performs coarse and refined estimates again. This ensures that the estimate of each path not only relies on its own parameters but also fully considers information from other paths. This ensures that the impact of other paths on the current path estimate is effectively taken into account during the optimization process, resulting in a more accurate estimate.

[0065] Step 4-3, if Or i=L max , the algorithm stops estimating. The former indicates that the residual change has met the required accuracy and no further optimization is required, thus terminating the estimation process. The latter indicates that the number of estimated paths has reached the algorithm's preset maximum number of paths, and the algorithm stops estimating to avoid the risk of excessive computational complexity or overfitting due to an excessive number of paths. Therefore, the algorithm's termination condition ensures both estimation accuracy and the rational use of computing resources, guaranteeing the effectiveness and efficiency of the estimation process.

[0066] Step 5: After the path parameter estimation is completed, the distance and speed of the target are calculated based on the time delay and Doppler information.

[0067] Example

[0068] The number of symbols N=32, the number of subcarriers M=32, the modulation mode is 4QAM, the carrier frequency is 4 GHz, the subcarrier spacing is 15 kHz, and the simulation environment is set to the Additive White Gaussian Noise (AWGN) channel.

[0069] Figure 4 The channel estimation performance of the LS, OMP, 2D-SOMP, and ROSR algorithms under different pilot signal-to-noise ratios (SNRs) under fractional delay-Doppler conditions is demonstrated. As can be seen from the figure, the ROSR parameter estimation algorithm demonstrates significant advantages in fractional delay-Doppler scenarios and in suppressing path interference.

[0070] Figure 5 The BER performance of the LS, OMP, 2D-SOMP, and ROSR algorithms in the OTFS system as a function of the SNR is demonstrated. The ROSR algorithm exhibits the best performance with the lowest BER under all SNR conditions, and the BER decreases rapidly with increasing SNR.

[0071] Figure 6 、 Figure 7 The performance of the ROSR algorithm in estimating target range and velocity parameters is compared with that of the MF algorithm, the FFT algorithm, the MUSIC algorithm, and the improved MF-F algorithm. The ROSR algorithm significantly outperforms the other compared algorithms in terms of noise immunity and estimation accuracy, demonstrating greater adaptability and stability. Its performance makes it well-suited for scenarios requiring high-precision target parameter perception, particularly in complex environments, providing reliable support for high-precision target perception tasks.

Claims

1. A high-precision target parameter estimation method for a synaesthesia system based on OTFS signals, characterized in that: The following steps are involved: Step 1: Establish the input and output relationship model of OTFS signal in DD domain; Step 2: Calculate a rough estimate of the channel parameters using the received pilot signal matrix; Step 3: Based on the rough estimate, the grid search range is gradually narrowed so that each round of search can be performed in a smaller area to obtain a finer estimate. Step 4: After each path is estimated, perform mutual optimization and interference suppression between paths, and then go to step 2. Step 5: After the path parameter estimation is completed, the distance and speed of the target are calculated based on the time delay and Doppler information.

2. The high-precision target parameter estimation method for the synaesthesia system based on OTFS signals according to claim 1 is characterized in that: The specific process of step 1 is: The input and output relationship of the OTFS signal in the DD domain is expressed as: in, is the channel response matrix in the DD domain, M is the number of subcarriers, N is the number of symbols, represents the noise of the receiver, and Represent the output and input signals of DD domain respectively; τ i represents the channel delay, v i represents the channel Doppler shift, h i represents the channel gain coefficient of the i-th path, Used to describe the delay and Doppler effect of each path.

3. The high-precision target parameter estimation method for the synaesthesia system based on OTFS signals according to claim 2 is characterized in that: The specific process of step 2 is: To perform a rough estimation of the channel parameters, first transform the expression in step 1 into the following form: in, It is the product of the channel response matrix and the input signal, which represents the influence of the delay and Doppler frequency shift of the i-th path on the input signal. is a set matrix containing L paths, represents the channel gain of L paths; Step 2-1: Convert the channel estimation problem into an optimization problem, namely: in, is the channel parameter value to be estimated. The goal of this formula is to minimize ‖y-Γ(τ,v)h‖‖ 2 ; Step 2-2: Since the optimization problem in step 2-1 has a three-dimensional search space (h, τ, v), in order to simplify the calculation process, first choose to estimate (τ, v), and then estimate h; given (τ, v), the objective function is regarded as a quadratic least squares problem with respect to h. In order to minimize this quadratic form, take the derivative of h and set the derivative to 0, that is: Among them, Γ H (τ,v) represents the conjugate matrix of Γ(τ,v); From this we get: Step 2-3, order: Ω=y H C(τ,v)(C H (τ,v)Γ(τ,v)) -1 C H (t,v)y Observing the expression of Ω, we can see that it is only related to (τ, v) and has nothing to do with the channel gain h. Therefore, the minimization problem in step 2-1 is transformed into a maximization problem of solving the cost function Ω, that is: After estimating the time delay and Doppler frequency shift through the above formula, Substitute the result in step 2-2 to obtain the channel gain 4. The high-precision target parameter estimation method for the synaesthesia system based on OTFS signals according to claim 3 is characterized in that: In step 3, the refinement estimation gradually narrows the scope of the grid search so that each round of search can be carried out in a smaller area, thereby gradually approaching the optimal solution. The specific process is as follows: Step 3-1. Design a refined search strategy to improve the estimation accuracy by controlling the search step size of each iteration; dynamically adjust the search range according to the current estimation results at each iteration, so as to gradually approach the actual delay and Doppler values; in the initialization stage, first set the key parameters to ensure the convergence and accuracy of the algorithm, thereby ensuring that there will be no offset or convergence instability during the refined search process; the delay search step size at the tth iteration and Doppler shift search step Expressed as: Where M is the number of subcarriers, N is the number of symbols, Δf is the subcarrier spacing, T is the symbol duration, t is the number of iterations, α is the delay dimension step size scaling factor, and β is the Doppler dimension step size scaling factor. Step 3-2: After determining the search step size, construct a two-dimensional grid search space; By dividing the search range into two dimensions of time delay and Doppler frequency shift, a grid structure is formed that can gradually approach the optimal parameters; the search space Φ at the tth iteration (t) Expressed as: Where, is the result of the tth iteration, and the search range of the delay dimension is Ψ τ and the Doppler search range Ψ v They are: P τ ={-a,...,0,...,a} P v ={-β,...,0,...,β} Step 3-3: After constructing the search space, the algorithm will calculate each search point (τ i ,v i )∈Φ (t) The value of the cost function Ω at the search point; by evaluating all search points, the algorithm can determine the maximum value of the cost function within the range and select the optimal combination of delay and Doppler shift as the new estimate. In this process, the maximum value of the cost function corresponds to the optimal estimate of the channel parameters. Therefore, by maximizing the cost function, the algorithm can gradually approach the ideal solution. After completing the current iteration step, it is determined whether the convergence condition is met. The convergence condition includes the following two cases: Where, ε τ and ε v are the convergence thresholds of delay and Doppler frequency shift set in the initialization phase, t max is the upper limit of the number of iterations; If the difference between the current estimate and the previous iteration result is less than the set threshold, it indicates that the algorithm is close to convergence; if the number of iterations reaches the preset upper limit and the algorithm still has not converged, it is considered that the algorithm has completed enough optimization processes and stops further iterations; if any of the above conditions are met, the refinement estimation process will terminate and the current iteration result will be output as estimates of time delay and Doppler shift; If the algorithm has not yet met the convergence conditions, it will continue to the next iteration; Each iteration re-evaluates the cost function in the updated search space and adjusts the parameter estimates based on the maximized cost function value; through multiple iterations, the estimates of delay and Doppler shift will gradually approach the optimal solution.

5. The high-precision target parameter estimation method for the synaesthesia system based on OTFS signals according to claim 4 is characterized in that: Step 4: After each path is estimated, mutual optimization and interference suppression between paths are performed. The specific process is as follows: Step 4-1: By calculating the residual of each path, determine whether the predetermined accuracy requirement is met, and then decide whether to continue iterative optimization; assuming that the residual vector of the i-th path is ε i , which is defined as follows: Step 4-2, if And i<L max , indicating that the current residual change exceeds the set threshold And the maximum number of path estimates L has not been reached max , if the current newly estimated path cannot make the residual change lower than the set threshold Continue to increase the number of paths; before estimating the parameters of the next path, the algorithm first optimizes the parameters of each estimated path; specifically, it fills the estimated values of other paths into the corresponding columns of the matrix Γ(τ,v) and performs coarse and fine estimation again; Step 4-3, if Or i=L max , the algorithm stops the estimation process.

6. 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, the steps of the method according to any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images