OTFS Dynamic Threshold Channel Estimation Method Based on Dual Pilot Path Consistency
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUILIN UNIV OF ELECTRONIC TECH
- Filing Date
- 2026-03-16
- Publication Date
- 2026-06-02
Smart Images
Figure CN122137706A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of wireless communication technology, specifically relating to a channel estimation method for an orthogonal time-frequency-space (OTFS) modulation system, and particularly a dynamic threshold channel estimation method based on dual pilot consistency detection. Background Technology
[0002] Orthogonal time-frequency modulation (OTFS) technology is one of the core candidate technologies for future sixth-generation mobile communications. Compared with traditional orthogonal frequency division multiplexing (OFDM) technology, OTFS modulates information symbols in the delay-Doppler domain, allowing each symbol to experience the time-varying characteristics of the entire channel, thus exhibiting superior anti-Doppler performance and lower bit error rate in high-speed mobile scenarios.
[0003] However, the advantage of OTFS technology relies on the receiver's accurate perception of time-varying channels. Currently available OTFS channel estimation methods each have their own characteristics but also corresponding limitations. For example, pilot-based methods estimate channel parameters by detecting the amplitude around the pilot, offering the advantage of simple implementation but facing the inherent challenge of threshold selection: setting the threshold too high will cause missed detections of true paths, while setting it too low will lead to false alarms due to noise. Deep learning-based estimation methods can utilize neural networks to learn complex channel characteristics, but they rely on large amounts of training data and have limited generalization ability. Compressed sensing-based estimation methods, while able to achieve high estimation accuracy by utilizing the sparsity of the channel in the time-delay-Doppler domain, have high computational complexity, making them difficult to meet the requirements of real-time systems.
[0004] It should be noted that even though existing methods can adjust the threshold based on parameters such as channel noise characteristics, they often fail to fully consider the subtle features of the channel path. This makes it difficult to accurately distinguish between the real path and noise interference, ultimately limiting the overall performance of channel estimation. Therefore, there is an urgent need for a novel channel estimation method that can adaptively optimize the threshold and fully utilize the characteristics of the OTFS channel. Summary of the Invention
[0005] The purpose of this invention is to address the problems of excessive complexity, difficulty in threshold selection, and lack of adaptive optimization capabilities in existing OTFS channel estimation methods. This invention provides an OTFS dynamic threshold channel estimation method based on dual pilot path consistency, which is applicable to OTFS channel scenarios with integer delay and Doppler characteristics.
[0006] The present invention adopts the following technical solution:
[0007] Compared to traditional methods, this invention employs a dual-pilot design at the transmitting end and dynamically adjusts the detection threshold at the receiving end by comparing the path detection results of the two pilots, thereby identifying the true path. The remaining procedures are consistent with conventional methods. The following section will focus on describing this dual-pilot consistency dynamic threshold channel detection method, with the specific steps as follows:
[0008] (I) Design and modulation of dual pilot frames: In the delay-Doppler domain transmit grid, the first pilot symbol is inserted at the first position. Insert the second pilot symbol at the second position. And satisfy Ensure the power of the two pilot symbols is consistent. The time delay interval between the two pilots must be no less than the maximum time delay spread of the system to avoid mutual interference. Perform OTFS modulation on the DD domain grid to generate the time-domain transmit signal.
[0009] (II) Received Signal Processing and Initial Parameter Estimation: The receiver performs OTFS demodulation on the received time-domain signal to obtain the DD domain receive matrix. Based on the extended search window, channel estimation matrices are constructed for the two pilot signals respectively, and then... The remaining symbols are used to calculate the noise power. An initial threshold coefficient is set, and significant path energy points are detected from the estimation matrix to preliminarily estimate the time delay index, Doppler index, and path gain of each path.
[0010] (III) Dynamic Threshold Adjustment for Dual Pilots: Dynamic threshold optimization is performed based on the initial detection results of two pilots. The core of this optimization lies in distinguishing the detection differences presented by "determinism of the true path" and "randomness of noise." Under initial high threshold conditions, due to instantaneous fluctuations and estimation biases in the gain of the two pilots on the same true path, the detected path set (i.e., the set of delay and Doppler index) may exhibit two typical states: First, the path set is completely consistent, indicating that the current threshold can only capture the strongest common path; second, it exhibits an inclusion relationship, specifically, the path set of one pilot is completely contained within the path set of the other pilot. This reflects that the current threshold setting is too high, causing some true paths to be captured only on the pilot with better channel conditions. The system gradually reduces the threshold so that the gain of both pilots on the true path can stably exceed the detection threshold, thereby achieving the integrity and consistency of the path set. When the threshold continues to decrease to the point of introducing significant random noise, the randomness of the noise causes inconsistencies in the detection results of the two pilots and no inclusion relationship; at this point, iteration is immediately stopped. This process achieves rapid location of the optimal threshold through a coarse-to-fine search strategy, thereby effectively improving the system's estimation performance.
[0011] (iv) Determining the final true channel estimate: After the stopping condition is met, the intersection of the paths detected by the two pilots at the optimal threshold is taken as the true channel path. For each path in the intersection, its final complex channel gain is obtained by merging the independent estimates of the two pilots. This mechanism makes full use of the spatial diversity gain of the dual pilots, effectively suppresses the random error caused by threshold mismatch in single pilot detection, and finally outputs highly reliable channel parameter estimation results, including time delay, Doppler, and path gain. Attached Figure Description
[0012] Figure 1 It is a system flowchart.
[0013] Figure 2 This is a flowchart of the dual-pilot path consistency dynamic threshold channel detection process.
[0014] Figure 3 This is a schematic diagram of channel estimation results where the threshold is too high, leading to missed detections.
[0015] Figure 4 This is a schematic diagram of channel estimation results causing false alarms due to an excessively low threshold.
[0016] Figure 5 This is a schematic diagram of the channel estimation results of the method of the present invention. Detailed Implementation
[0017] The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the invention, and should not be construed as limiting the invention.
[0018] Part 1: Transmission and Reception of Dual Pilot Signals
[0019] S1, design the N and M parameter grid for the delay-Doppler domain. Design the pilot positions within the delay-Doppler domain transmit grid. A first pilot symbol is inserted at the first position, and a second pilot symbol of equal power is inserted at the second position. Both pilot symbols have a power of [missing information]. .
[0020] At the same time, the time delay index spacing between the two pilots must be greater than the system's maximum time delay protection value. This is to avoid the two pilots interfering with each other during channel transmission.
[0021] S2 performs ISFFT and IFFT transformations on the conjugate pilot signal to convert it to the time domain for transmission. After passing through a multipath time-varying time-domain channel, the received signal undergoes SFFT and FFT transformations to convert it to the time-delay-Doppler domain, yielding the received signal matrix. .
[0022] Part Two: Initial Channel Estimation of Pilot Response
[0023] According to the received signal matrix With the initial information of the two pilots, the channel detection matrix corresponding to the two pilots can be extracted. The initial channel parameters can be estimated by using the initialized threshold, and the estimated values of the delay index, Doppler index and path gain of the two sets of multipath channels can be obtained.
[0024] S1, based on the system's maximum delay and maximum Doppler frequency shift The search range for delay and Doppler index is specified and expanded (to prevent missed detections in actual channel paths). The expanded delay search range is... The Doppler search range is Then the detection matrix , Size is .in," ", indicates rounding up.
[0025] S2 will receive the signal matrix Remove the two detection matrices , Calculate the average power of the remaining symbols, which is the average noise power. At this point, the initial threshold is determined. (in ).
[0026] S3, with , central position ,exist , Perform a cyclic scan within. For , Each relative position within Calculate its in Corresponding actual loop position ,in , ( , (For the corresponding pilot positions), respectively inside The value is stored accordingly in , of inside.
[0027] S4, at this point, the complete detection matrix is obtained. , For power values exceeding the threshold described in step S2 unit That is, satisfying the relation Based on its position relative to the center of the matrix Determine the relative delay index of the path. and relative Doppler index And based on the received value of the unit and the known pilot symbol value Estimate the channel path gain of this path. .
[0028] At this point, the delay indices of the two multipath channels are obtained. Doppler Index and path gain The estimated value.
[0029] Part 3: Consistency-Based Dynamic Threshold Adjustment
[0030] S1, Under the current threshold, analyze the path set in the initial detection results of the two pilots. and The consistency relationship.
[0031] S2, if (The paths are exactly the same) or If the path contains elements of another path, the current threshold is deemed too high, leading to a missed detection. For example... Figure 3 As shown, under this high threshold, both the existing single-pilot method and this scheme only detect the main path, resulting in the loss of multiple weak paths because they did not reach the detection threshold, which illustrates the problem of missed detection caused by fixing a high threshold.
[0032] At this time, according to Update coefficient ( (where S is the step size), lower the threshold, and proceed to the next iteration, that is, repeat step S1.
[0033] S3, if If there is no inclusion relationship, then the current threshold is determined to have fallen to the critical point where noise interference exists, and the iteration stops. At this time, the coefficient... Find the optimal value and record the threshold value at this point. This critical state implies a threshold. It begins to reach a small number of noise peaks with the highest power. For example... Figure 4 As shown, at this threshold, for traditional single-pilot detection, in addition to the real path, multiple noise peaks are misjudged as paths, resulting in some false components in the estimated channel that do not exist in the original channel, indicating that a fixed low threshold can cause false alarms.
[0034] Since the power of the noise peak is necessarily higher than its average power Therefore, there is The threshold As the final detection threshold, it can reliably distinguish the vast majority of real paths from noise. Assuming the channel noise follows a Gaussian model, at this threshold, the probability of misclassifying a noisy path as a real path during single-pilot detection is... The probability of this happening independently for both pilots is... Furthermore, due to the constraint of dual-pilot path consistency, the noise path that is misjudged must have the same time delay and Doppler index on both pilots. In reality, the system's false alarm probability is much lower than... .
[0035] Part Four: Determining the Final Channel Estimation Results
[0036] S1, obtaining the optimal coefficients The two sets of paths corresponding to the two pilot signals. The intersection of the two sets. As the final set of true channel paths.
[0037] S2, for For each path in the array, estimate its independent gain on the two pilots. and Merging This is to obtain a more accurate complex channel gain, and finally synthesize a complete channel impulse response estimate. Figure 5 As shown, at this threshold The dual-pilot dynamic threshold selection method achieves a good balance between avoiding missed detections and suppressing false alarms.
Claims
1. An OTFS dynamic threshold channel estimation method based on dual pilot path consistency, characterized in that, Includes the following steps: (I) Design and modulation of dual pilot frames: In the delay-Doppler domain transmit grid, the first pilot symbol is inserted at the first position. Insert the second pilot symbol at the second position. And satisfy Ensure the power of the two pilot symbols is consistent. The time delay interval between the two pilots must be no less than the maximum time delay spread of the system to avoid mutual interference. Perform OTFS modulation on the DD domain grid to generate the time-domain transmit signal. (II) Received Signal Processing and Initial Parameter Estimation: The receiver performs OTFS demodulation on the received time-domain signal to obtain the DD domain receive matrix. Based on the extended search window, channel estimation matrices are constructed for the two pilot signals respectively, and then... The remaining symbols are used to calculate the noise power. An initial threshold coefficient is set, and significant path energy points are detected from the estimation matrix to preliminarily estimate the time delay index, Doppler index, and path gain of each path. (III) Dynamic Threshold Adjustment for Dual Pilots: Dynamic threshold optimization is performed based on the initial detection results of two pilots. The core of this optimization lies in distinguishing the detection differences presented by "determinism of the true path" and "randomness of noise." Under initial high threshold conditions, due to instantaneous fluctuations and estimation biases in the gain of the two pilots on the same true path, the detected path set (i.e., the set of delay and Doppler index) may exhibit two typical states: First, the path set is completely consistent, indicating that the current threshold can only capture the strongest common path; second, it exhibits an inclusion relationship, specifically, the path set of one pilot is completely contained within the path set of the other pilot. This reflects that the current threshold setting is too high, causing some true paths to be captured only on the pilot with better channel conditions. The system gradually reduces the threshold so that the gain of both pilots on the true path can stably exceed the detection threshold, thereby achieving the integrity and consistency of the path set. When the threshold continues to decrease to the point of introducing significant random noise, the randomness of the noise causes inconsistencies in the detection results of the two pilots and no inclusion relationship; at this point, iteration is immediately stopped. This process achieves rapid location of the optimal threshold through a coarse-to-fine search strategy, thereby effectively improving the system's estimation performance. (iv) Determining the final true channel estimate: After the stopping condition is met, the intersection of the paths detected by the two pilots at the optimal threshold is taken as the true channel path. For each path in the intersection, its final complex channel gain is obtained by merging the independent estimates of the two pilots. This mechanism makes full use of the spatial diversity gain of the dual pilots, effectively suppresses the random error caused by threshold mismatch in single pilot detection, and finally outputs highly reliable channel parameter estimation results, including time delay, Doppler, and path gain.
2. The method according to claim 1, characterized in that, In step (a), before or after inserting the dual pilots, the following is also included: The two pilot symbols have the same symbol value throughout the entire delay-Doppler grid. Furthermore, the symbols within the area defined by the system-preset maximum delay index and maximum Doppler index, centered on the first and second positions of each pilot symbol, are set as blank protection symbols to ensure that the two pilots are not interfered with during channel transmission.
3. The method as described in claim 1, characterized in that, The formula for calculating the initial threshold in step (iii) is as follows: in, These are the initial weighting coefficients. The average power of the noise. For a single pilot symbol at the transmitting end The power, that is This formula sets an initial threshold through a linear combination of noise power and pilot transmit power, aiming to effectively distinguish between signal and noise. During design, it should be ensured that the pilot power is higher than the noise power; otherwise, the signal will be submerged in noise, causing the system to lose its effective detection capability.
4. The method as described in claim 3, characterized in that, The average power of the noise Obtained through the following methods: From the received signal matrix In the middle, the region covered by the two pilot detection matrices is excluded, and the average power of all remaining symbols is calculated.
5. The method as described in claim 1, characterized in that, The dynamic threshold adjustment in step (iii) specifically includes: S1, the time delay index and Doppler index detected by the two pilots are recorded as two sets of paths. S2, when the path sets are completely identical or contain each other, use a fixed step size. Reduce weighting coefficient . S3, gradually decrease The noise interference threshold is determined when the two sets of paths are neither completely identical nor contain each other. This threshold is then recorded as the final threshold.
6. The method as described in claim 5, characterized in that, The fixed step size The value range is 0.05 to 0.2, and the initial weighting coefficient... The value range is 0.3 to 0.
7. The above parameter range is an empirically optimized value determined based on the signal-to-noise ratio conditions of typical underwater acoustic and wireless channels, in order to balance convergence speed and detection accuracy.
7. The method as described in claim 1, characterized in that, The path gain merging in step (iv) uses a weighted average method: For each path in the intersection, its path gain estimates on the two pilots are combined using the following formula: in , This is the path gain estimate corresponding to the detection of two pilots based on the final threshold.