Low-power-consumption wake-up signal receiving method based on improved Reed-Muller coding

Abstract

The invention discloses a low-power-consumption wake-up signal receiving method based on improved Reed-Muller coding, relates to the technical field of low-power-consumption communication, and aims at solving the problem that the low-power-consumption wake-up signal receiving performance is insufficient due to the fact that the minimum Hamming distance of code words is limited in Reed-Muller coding in the existing 3GPP (3rd Generation Partnership Project) standard. According to the method, firstly, Reed-Muller coding in 3GPP is optimized: a simulated annealing algorithm is introduced, the optimal code word combination of coding is searched by using the global optimization capability of the simulated annealing algorithm, the minimum Hamming distance of each code word after coding is significantly increased, the anti-interference capability of low-power-consumption wake-up signal transmission is enhanced, and the detection error rate (MDR) during detection is reduced from the coding level; at a receiving end, a threshold value judgment receiving mode is used for further improving the detection performance, the receiving reliability of the low-power-consumption wake-up signal is effectively improved, and the method is suitable for an LP-WUS (low-power-consumption wake-up signal) system, especially suitable for scenes with complex noise interference and strict requirements for the false alarm rate and the equipment power consumption, such as the Internet of Things and the Internet of Vehicles. And the method has relatively high practicability and popularization value.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and in particular to a low-power wake-up signal (LP-WUS) reception method based on improved Reed-Muller (RM) coding. Specifically, by improving the Reed-Muller (RM) coding in the existing standard, it adapts to communication scenarios with low power consumption and high signal reception reliability requirements. It can be applied to low-power communication scenarios with complex interference environments such as the Internet of Things and the Internet of Vehicles, thereby improving the overall reception performance of the LP-WUS system. Background Technology

[0002] In recent years, the demand for low-power communication technologies in various Internet of Things (IoT) and vehicle-to-everything (V2X) applications has continued to grow. As a key component of the low-power communication architecture, the stability and reliability of signal reception in the LP-WUS system directly affect the overall system's operating efficiency. Currently, relevant technical specifications have clearly stipulated that the wake-up signal of the LP-WUS system can adopt a specific RM encoding scheme. This scheme generates the encoded signal through operations on information bits and a base sequence matrix, and adjusts the code length using rate matching to adapt to the system's transmission requirements.

[0003] As the requirements for signal reception performance in low-power communication scenarios continue to increase, the existing RM coding and supporting reception schemes in the standard are gradually showing technical limitations: on the one hand, under different information bit configurations, the base sequence optimization of the existing RM coding is not sufficient, resulting in limited anti-interference capability of the encoded signal, making it difficult to cope with complex communication environments; on the other hand, the signal detection scheme at the receiving end is not fully adapted to the common interference characteristics of LP-WUS systems, making it difficult for the system to meet the actual application requirements in terms of key performance indicators such as false alarm rate and false detection rate.

[0004] Furthermore, as low-power communication is increasingly integrated with other terrestrial communication systems, LP-WUS signals need to be transmitted in complex channel environments where multiple signals coexist. The shortcomings of existing coding and receiving technologies further affect the stable operation of the system. Therefore, it is urgent to optimize and improve the existing RM coding and receiving schemes to solve the receiving performance bottleneck of the LP-WUS system. Summary of the Invention

[0005] This invention provides a low-power wake-up signal receiving method based on improved Reed-Muller (RM) coding to solve the problems of limited Hamming distance and difficulty in controlling the false alarm rate at the receiver in the existing 3GPP standard RM coding.

[0006] The method provided by this invention has the following specific steps:

[0007] Step S1: Optimize the RM encoding basis sequence matrix according to the number of input information bits K. When K=3, use the traversal method to select the optimal permutation. When K≥4, use the simulated annealing algorithm to find the optimal permutation. The goal is to increase the minimum Hamming distance of the codewords and satisfy the constraints.

[0008] Step S2: Perform rate matching on the optimized RM encoded output codewords, and adjust the output codeword length according to the rules to adapt to transmission requirements.

[0009] Step S3: The receiver performs threshold detection based on noise, calculates the detection value, constructs the threshold vector, determines the weighting parameters and decision threshold, and controls the false alarm rate and false detection rate.

[0010] The advantages of this invention are:

[0011] 1. Significantly improved Hamming distance: By optimizing the base sequence matrix through traversal method (K=3) and simulated annealing algorithm (K≥4), the minimum Hamming distance of more codewords after encoding meets the constraint requirement: the length of even codewords is ≥ half the length of the encoded bitwords, and the length of odd codewords is ≥ (length of the encoded bitwords - 1) / 2.

[0012] 2. High-dimensional optimization, high efficiency and feasibility: The simulated annealing algorithm, through the mechanism of "global optimization + probability acceptance of inferior solutions", can find the optimal basis sequence permutation when K=4 and K=5 within 1e8 iterations, avoiding the traversal problem of 15! and 31! permutations, thus balancing accuracy and efficiency;

[0013] 3. Accurate noise adaptation detection: The receiver constructs a threshold vector based on the noise signal, and the weighting parameter 'a' is directly related to the target false alarm rate. In IoT and vehicle networking scenarios with complex noise interference, the false alarm rate can be stably controlled within the target range, while avoiding the problem of insufficient adaptation of non-target signal thresholds to noise misjudgment. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart of the overall process for optimizing the RM encoding base sequence matrix in this invention.

[0016] Figure 2 Flowchart of the simulated annealing algorithm optimization when K≥4

[0017] Figure 3 This is a flowchart of the noise-based threshold detection process for the signal receiver of the present invention. Detailed Implementation

[0018] To make the above-mentioned objectives and advantages of this application more apparent and understandable, this application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] Figure 1 This is a flowchart illustrating the overall process of RM-encoded base sequence matrix optimization in this invention. The specific steps are as follows:

[0020] Step S11: Input the number of information bits K, and enter the K value judgment stage (use the traversal method when K=3, and use the simulated annealing algorithm when K≥4);

[0021] Step S12: If K = 3, first generate all 3-bit combinations of 0 and 1 (total 2... 3 =8 types (including all-zero combinations), and an initial basis sequence matrix is ​​constructed based on these 8 combinations; since the actual encoding needs to exclude all-zero combinations, the matrix is ​​based on 2. 3 -1 = 7 non-zero combinations generate 7! = 5040 base sequence permutations; according to the RM encoding rules, the initial codeword matrix C is calculated using the formula C = (S × G) mod 2 (S is the information bit matrix based on 8 01 combinations, and G is the base sequence matrix of the current permutation); for each code length of the encoded 20 to 32 bits, it is determined whether the minimum Hamming distance corresponding to the permutation satisfies the constraints (even code length ≥ half the code length, odd code length ≥ (code length - 1) / 2);

[0022] Step S13: Select the permutation with the largest number of code lengths that satisfy the constraints, and determine it as the optimized base sequence matrix G;

[0023] Step S14: If K≥4 (K=4 or K=5), first generate all 0 and 1 combinations of Kbit (2 when K=4). 4 =16 types, K=5 when 2 5 =32 kinds), construct the information bit matrix S (taking non-zero combinations, 15 kinds when K=4, 31 kinds when K=5) and the initial generation matrix G (4×15 when K=4, 5×31 when K=5), and input the algorithm parameters (T_initial=1e7, Cool_rate=0.99, etc.);

[0024] Step S15: Generate the initial codeword matrix C, initialize the temperature to 1e7 and random column arrangement, calculate the initial fitness (the number of codeword lengths that satisfy the constraints) and initialize the global optimal solution;

[0025] Step S16: Enter the main loop for optimization, generate neighborhood solutions by swapping column indices, calculate new fitness, accept solutions according to the Metropolis criterion and update the global optimal solution, and at the same time cool down according to "current temperature = current temperature × 0.99";

[0026] Step S17: When the termination condition (temperature ≤ 1e-10 or iteration count ≥ 1e8) is met, the optimal arrangement of the base sequence matrix columns is output, and the base sequence arrangement that satisfies the constraint conditions (even code length ≥ half the code length, odd code length ≥ (code length - 1) / 2) the most is obtained.

[0027] Figure 2 The flowchart below shows the optimization process of the simulated annealing algorithm when K≥4. This flowchart illustrates the core process of optimizing the base sequence matrix using the simulated annealing algorithm when K≥4, and the steps are as follows:

[0028] Step S21: Input the parameters required for the algorithm, including: the information bit matrix S constructed based on all 0 and 1 combinations of Kbit (15×4 when K=4, containing 15 kinds of 4-bit non-zero combinations; 31×5 when K=5, containing 31 kinds of 5-bit non-zero combinations), the initial generation matrix G (4×15 when K=4, 5×31 when K=5), the target code length range m_range=20:31, and the algorithm control parameters (initial temperature 1e7, cooling rate 0.99, etc.);

[0029] Step S22: According to the RM encoding rules, generate the initial codeword matrix C (15×15 when K=4, 31×31 when K=5) using the formula C=(S×G)mod2;

[0030] Step S23: Initialize the current temperature to 1e7, and generate a random permutation of the columns of the base sequence matrix current_perm (when K=4, it is a random order of 1 to 15, and when K=5, it is a random order of 1 to 31);

[0031] Step S24: Call the fitness calculation function: adjust the column order of C according to current_perm to obtain C_perm, construct the cyclic expansion matrix C_ext (15×31 when K=4) covering code lengths 20 to 31, calculate the cumulative weight matrix W, count the number of code lengths that satisfy the Hamming distance constraint, and obtain current_fitness;

[0032] Step S25: Initialize the global optimal solution by setting best_perm = current_perm and best_fitness = current_fitness;

[0033] Step S26: Determine if the termination condition is met (temperature ≤ 1e-10 or iteration count ≥ 1e8). If not, randomly swap two columns in current_perm to generate a new permutation new_perm, and calculate new_fitness and fitness change Δf. If Δf > 0, directly accept new_perm and update current_perm and current_fitness. If Δf ≤ 0, calculate the acceptance probability P = e^(Δf / T). If P is greater than a random number in the interval [0,1], then accept new_perm.

[0034] Step S27: If current_fitness > best_fitness, update best_perm and best_fitness synchronously;

[0035] Step S28: Cool down according to "current temperature = current temperature × 0.99", increment the iteration count by 1, and return to step 6;

[0036] Step S29: When the termination condition is met, adjust the column order of the initial generation matrix G according to best_perm, and output the optimized basis sequence matrix G2 (K=4) or G3 (K=5).

[0037] Figure 3 This is a flowchart illustrating the noise-based threshold detection process of the signal receiver according to the present invention. The flowchart shows the process by which the receiver constructs a threshold using a noise signal and performs signal detection. The steps are as follows:

[0038] Step S31: Receive external pure noise signal;

[0039] Step S32: If M ≥ 1000 consecutive receptions are received to ensure statistical validity, perform a vector inner product operation (with the local target wake-up signal) on each signal to obtain the detection value;

[0040] Step S33: Summarize the detection values ​​of all noise signals and construct the threshold vector D;

[0041] Step S34: Based on the system target false alarm rate FAR_target (e.g., FAR≤0.01), and combined with the Gaussian distribution characteristics of the noise detection values, calculate the weighting parameter a (a∈[0,1]) using signal detection theory;

[0042] Step S35: Sort the threshold vector D in ascending order, calculate the index value idx = round(length(D) × a), and take the element at position idx in the sorted vector as the decision threshold Th;

[0043] Step S36: If the received signal is the signal to be detected, calculate the detection value v_rec, which is the vector inner product of the signal and the local target wake-up signal;

[0044] Step S37: Compare v_rec with Th. If v_rec > Th, it is determined to be a target wake-up signal, and the system is triggered to wake up. If v_rec ≤ Th, it is determined to be noise, and the system remains in a low-power state.

Claims

1. A low-power wake-up signal receiving method based on improved Reed-Muller coding, characterized in that, It includes the following steps: S1: For the Reed-Muller (RM) coding rules specified in the 3GPP TS 38.212 standard for Low Power Wake-up Signal (LP-WUS) systems, the base sequence matrix of the RM coding is optimized according to the different number of input information bits K to increase the minimum Hamming distance of the encoded codeword. The coding method is as follows: Assuming the original information bit vector is S (1 x M, where M is the number of information bits), the coding base sequence matrix G is selected, and the output codeword C = mod(S x G) T ,2). When K=3, this is achieved by traversing all the base sequence permutations corresponding to the 3-bit information bits (since 3 bits of information correspond to 2). 3 -1 = 7 non-zero combinations, so there are a total of 7! combinations. The encoding base sequence matrix G1, which maximizes the number of combinations satisfying the condition "minimum Hamming distance ≥ half the encoded bit length (≥ (encoded bit length - 1) / 2 when the bit length is odd)," is selected from those with a bit length of 20-32. G1 is: When K = 4, the simulated annealing algorithm is used to optimize the column permutation of the base sequence corresponding to the 15×4 information bit matrix, and the encoded base sequence matrix G2 is obtained. The simulated annealing algorithm includes initialization, main loop optimization, fitness calculation, and cooling termination steps. In the initialization stage, a random column permutation is generated and the initial fitness is calculated. In the main loop stage, a neighborhood solution is generated by exchanging column positions, and the solution is accepted and the optimal solution is updated according to the Metropolis criterion. In the fitness calculation stage, the fitness value is calculated by rearranging the columns of the codeword matrix, creating a cyclic extension matrix, and judging whether the minimum Hamming distance meets the constraints; when K = 5, the same simulated annealing algorithm as when K = 4 is used to obtain the encoded base sequence matrix G3. The G2 and G3 are as follows: S2: Perform rate matching on the optimized RM encoded output codeword. The rate matching rule is: Let the output codeword be f, the output codeword length be k, and the input codeword be d, satisfying f k =d k mod N k = 0, 1, 2, ..., M-1, where M is the length of the output codeword; S3: At the receiving end, a threshold-based detection algorithm is used, which specifically includes: S31: Detection value calculation: The received low-power wake-up signal (row vector) is subjected to a vector inner product operation with the target wake-up signal (column vector) stored locally to obtain a single detection value; S32: Threshold vector construction: Multiple of the detection values (the detection values correspond to pure noise signals) are collected to form a threshold vector D; S33: According to the target false alarm rate required by the system, the threshold weighting parameter a is calculated through signal detection theory, where a ∈ [0, 1]; S34: The threshold vector D is sorted from small to large, and the index value is "length(D)×a" (length(D) is the number of elements of the threshold vector D), and the element in D corresponding to this index is taken as the initial decision threshold.

2. The method according to claim 1, characterized in that, The parameter settings of the simulated annealing algorithm in step S1 are as follows: the initial temperature T_initial = 1e7, the cooling rate Cool_rate = 0.99, the minimum temperature T_min = 1e-10, the maximum number of iterations max_iterations = 1e8, and the target code length range m_range = 20:31; in the fitness calculation step, the cyclic extension matrix is obtained by splicing the rearranged codeword matrix C_perm multiple times and supplementing the 0th column of C_perm, and the cumulative weight is calculated by adding up the rows of the cyclic extension matrix.

3. The method according to claim 1, characterized in that, When K = 3 in step S1, the correspondence between the encoded bit length and the minimum Hamming distance based on matrix G2 is as follows:

4. The method according to claim 1, characterized in that, When K = 4 in step S1, the correspondence between the encoded bit length and the minimum Hamming distance based on matrix G2 is as follows:

5. The method according to claim 1, characterized in that, When K = 5 in step S1, the correspondence between the encoded bit length and the minimum Hamming distance based on matrix G3 is as follows:

6. The method according to claim 1, characterized in that, The determination of the threshold weighting parameter A (0 < A < 1) and the calculation process of the theoretical false alarm rate (FAR) in step S3 are as follows: Decision threshold selection and theoretical FAR calculation: Take the element corresponding to the index "round(length(D)×A)" (round is the rounding function, length(D) is the number of elements in the threshold vector D) in the sorted threshold vector D_weighted as the decision threshold; at this time, the theoretical false alarm rate FAR of the system satisfies the formula: FAR=(1-A) / 2^K, where K is the number of input information bits, A∈(0,1).

Images