A method of estimating doppler shift in an OTFS system
Patent Information
- Application Number
- CN202311513096.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-11-14
- Publication Date
- 2026-08-21
- Estimated Expiration
- 2043-11-14
AI Technical Summary
通常情况下M远大于N,多普勒域分辨率较大,导致每条径的多普勒频移υi通常不是分辨率的整数倍,引入多普勒扩展,降低了信道的稀疏性,如图2所示
[0028] (1) The present invention is not affected by the resolution of the Doppler domain.
Smart Images

Figure CN117596115B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of Doppler estimation, and more specifically to a method for estimating Doppler frequency shift in an OTFS system. Background Technology
[0002] Most mainstream wireless communication system standards currently use Orthogonal Frequency Division Multiplexing (OFDM) as the basic waveform, which can significantly reduce the implementation complexity of transceivers in broadband channels, but its performance is poor in high-speed mobility. Ultra-high mobility manifests in the time domain as rapidly time-varying channel response, which puts significant pressure on the real-time performance and accuracy of channel estimation, as well as pilot overhead control. Furthermore, ultra-high mobility exhibits a strong Doppler effect in the frequency domain, leading to severe inter-carrier interference in OFDM systems, which is difficult to handle using traditional transceiver architectures. Against this backdrop, in recent years, R. Hadani et al. proposed a novel modulation technique called "Orthogonal Time-Frequency Modulation" (OTFS). The main idea of OTFS is to introduce a delay-Doppler domain, transforming the rapidly time-varying time-domain channel into a slowly varying and sparse delay-Doppler domain channel, such as... Figure 1 As shown. During OTFS modulation, each modulation symbol undergoes a frame of time-frequency resource expansion to obtain the channel's full diversity gain.
[0003] The OTFS signal can be represented by the time-domain channel as follows:
[0004] r(t)=∫∫h(τ,υ)s(t-τ)e j2πυ(t-τ) drdυ+ω(t)
[0005] Where s(t) is the time-domain signal of the OTFS, and ω(t) is the white noise in the channel. h(τ,υ) is the channel response in the time-delay-Doppler domain, which can be expressed as:
[0006]
[0007] Where P is the number of multipaths, τ i For the time delay of each path, υ i For the Doppler frequency shift of each path, h i For each path, the value is a complex number.
[0008] Assuming an OTFS frame has N symbols, and each symbol has M modulated data, the time delay domain resolution is T / M, where T is the duration of one symbol, and the Doppler domain resolution is Δf / N, where Δf is the subcarrier spacing. Typically, M is much larger than N, resulting in a large Doppler domain resolution and a Doppler frequency shift υ for each path. i Typically not an integer multiple of the resolution, Doppler spread is introduced to reduce the sparsity of the channel, such as... Figure 2 As shown.
[0009] There are many channel estimation methods for OTFS, and most of them require estimation of Doppler frequency shift. Due to the large resolution of the Doppler domain, its Doppler spread leads to inaccurate estimation of Doppler frequency shift. Summary of the Invention
[0010] To address the shortcomings of existing technologies, this invention proposes a method for estimating Doppler frequency shift in an OTFS system. This method utilizes a large-point-count FFT to improve Doppler estimation accuracy, unaffected by Doppler domain resolution. Accurate Doppler estimation can improve channel estimation accuracy, thereby enhancing system performance.
[0011] The technical solution adopted in this invention is as follows:
[0012] A method for estimating Doppler frequency shift in an OTFS system, the specific steps of which are as follows:
[0013] (1) Detect the delay of multipath propagation using a threshold method;
[0014] (2) Estimate the Doppler delay for each multipath path until all delay estimates are completed.
[0015] Furthermore, the specific processing procedure for step (2) is as follows:
[0016] (201) Extract the channel estimation result for a certain time delay and obtain the time-domain channel response through IDFT;
[0017] (202) Perform 2 on the time-domain channel response P Point FFT yields H HD Where p is a set value;
[0018] (203) Searching for H HD The maximum power value is indexed. If the maximum power value exceeds the preset threshold, a Doppler path exists. The index corresponds to the Doppler frequency shift size, and step (204) is executed. Otherwise, no Doppler path exists, and step (201) is returned to estimate the Doppler frequency shift of the next time delay.
[0019] (204) Reconstruct the time-domain channel response corresponding to the Doppler frequency shift, eliminate the reconstructed time-domain channel response in the original time-domain channel response, and return to step (202).
[0020] Furthermore, the specific process of step (204) is as follows:
[0021] Let I be the index corresponding to the maximum power. max The corresponding Doppler frequency shift is:
[0022] doppler=I max ·N / 2 P
[0023] In the formula, N is the number of symbols in a frame of the OTFS system;
[0024] Reconstruct the time-domain channel response h corresponding to the Doppler frequency shift tmp for:
[0025]
[0026] Subtract the reconstructed time-domain channel response from the original time-domain channel response to obtain a new time-domain channel response, and then return to step (202).
[0027] The advantages of this invention compared to the prior art are as follows:
[0028] (1) The present invention is not affected by the resolution of the Doppler domain.
[0029] (2) This invention can estimate Doppler frequency shift with high accuracy. Attached Figure Description
[0030] Figure 1 This is a schematic diagram of the time-delay Doppler domain sparse channel response of the present invention.
[0031] Figure 2 The delay-Doppler domain channel response of this invention is (M=256, N=14, single path, delay=0, Doppler frequency shift=0.5*Doppler resolution).
[0032] Figure 3 This is a schematic diagram of the placement of time-delay Doppler domain pilots and data according to the present invention.
[0033] Figure 4 This is a flowchart of the Doppler estimation process of the present invention.
[0034] Figure 5 This is a schematic diagram of the intermediate process of Doppler estimation in this invention. Detailed Implementation
[0035] 0TFS systems typically insert pilot signals in the time-delay Doppler domain for channel estimation. To prevent Doppler spread from interfering with the useful data, a blank signal is added between the pilot signals and the useful data as an interval. See [link to specific placement method] for details. Figure 3 The blank signal occupies the entire Doppler domain, and the delay domain occupies a length equal to the maximum channel delay. After the receiver recovers the delay-delayed Doppler domain data, detecting the peak positions within the blank signal allows the determination of the number of multipaths, the delay, and the Doppler frequency shift. If the true Doppler frequency shift of a certain path is not an integer multiple of the Doppler resolution, the peak value of that path will expand into multiple paths, making it impossible to obtain the true Doppler frequency shift and increasing the complexity of channel estimation (the complexity of channel estimation increases with the number of paths).
[0036] Assuming that the initial channel estimate in the time-delay Doppler domain has already been obtained using pilot signals, this invention proposes a method for estimating the Doppler frequency shift in an OTFS system, the flowchart of which is shown below. Figure 4 The specific process is as follows:
[0037] Implementation Example 1
[0038] Assuming an OTFS system with M = 256 and N = 14, the initial channel response matrix H in the time-delay Doppler domain has been obtained based on the pilot signals. DD The dimension is N*L, where L is the maximum channel delay.
[0039] Step 1: Detect multipath delay
[0040] Accumulate H by column DD The energy is obtained from the channel response in the time delay domain and H. SUM :
[0041]
[0042] Detecting H values greater than a threshold SUM (m), whose set of m values is m k k = 0, 1, ..., K-1 represents the multipath delay to be detected, and K is the total number of delays exceeding the threshold. The threshold can be obtained by simulating the actual channel or by calculating H. SUM The mean of (m) is multiplied by a fixed empirical constant (usually 4 to 10).
[0043] Step 2: Estimate the delay m of all multipath paths k Doppler frequency shift, k = 0, 1, ... K-1
[0044] (201) The retrieval delay is m k Channel response H DD (n, m) k ), n = 0, 1, ..., N-1;
[0045] (202) Obtain the time-domain channel response through N-point IDFT;
[0046] (203) After 2 P Point FFT yields H HD The larger P is, the higher the accuracy of the estimated Doppler;
[0047] (204) Find the maximum power max(|H HD | 2 ), thus obtaining its index as I. max ;
[0048] (205) If the maximum power is less than the threshold, the current time delay Doppler estimation is complete, and the next time delay estimation is performed. Return to step (201) until all time delay estimations are completed. Otherwise, reconstruct the time-domain channel response h corresponding to the Doppler frequency shift. tmp Subtract the reconstructed time-domain channel response from the original time-domain channel response, and then proceed to step (203);
[0049] The threshold can be obtained by simulating the actual channel, or by calculating the maximum amplitude of the first iteration of each delay multiplied by a fixed empirical constant (usually 0.2 to 0.7).
[0050] Reconstruct the time-domain channel response h corresponding to the Doppler frequency shift tmp Specific steps:
[0051] Assume the index corresponding to the maximum amplitude is I. max Then the corresponding doppler should be
[0052] doppler=I max ·N / 2 P
[0053] Reconstructing the time-domain channel response is
[0054]
[0055] In the original time-domain channel response h t Subtract the reconstructed time-domain channel response h tmp The new time-domain channel response h is obtained. t1 :
[0056] h t1 =h t -h tmp
[0057] Intermediate results of a Doppler estimation process with a certain time delay are as follows: Figure 5 As shown:
[0058] Figure 5 In Figure (a), the channel response in the Doppler domain for a given time delay is shown. Two paths are pre-defined for this delay, with actual Doppler values of [0.22, 2.56], in units of Doppler domain resolution. The amplitudes of the two paths are [1, 0.5]. It can be seen that the channel response exhibits severe Doppler spread. If the Doppler is determined directly using threshold path finding, many paths will be found, and the Doppler estimation will be inaccurate, only estimating results that are integer multiples of the Doppler resolution.
[0059] (b) The figure shows H in the first iteration. HDRegarding the amplitude response, due to the increased number of FFT points (8192 points in the figure), the Doppler resolution also decreased. Based on the maximum value index, the Doppler of the first path can be calculated as 0.19, in Doppler domain resolution. The deviation from the true value is 0.03, but this is still much more accurate than direct detection in the time-delayed Doppler domain. Simultaneously, the threshold is calculated as: maximum value * fixed empirical constant (0.3) = 0.32.
[0060] (c) The figure shows H in the second iteration. HD The amplitude response shows that after the Doppler path detected in the first iteration is reconstructed and subtracted, only the Doppler value of the other path remains. Since the maximum value is greater than the previously calculated threshold of 0.32, the existence of the other path can be confirmed. Its Doppler value can be calculated to be 2.57 based on the maximum value index, with the unit being the Doppler domain resolution. Because the interference of the first path has been eliminated, the Doppler estimation deviation of the second path is smaller.
[0061] (d) The figure shows H in the third iteration. HD The amplitude response shows that the maximum value is already less than the threshold, so it is considered that there is no path, and the Doppler estimation of this time delay ends.
[0062] The Doppler offset error estimated through the above process is smaller and is not limited by the Doppler resolution of the OTFS system. Using this estimation result to correct the initial channel estimate can yield a more accurate channel estimate, thereby improving the overall system performance.
Claims
1. A method for estimating Doppler frequency shift in an OTFS system, characterized in that, The specific steps are as follows: (1) Detect the delay of multipath propagation by using a delay threshold; (2) Estimate the Doppler delay for each multipath path one by one until all delay estimates are completed. Specifically: (201) Take out the channel estimation result for a certain time delay and obtain the time-domain channel response through IDFT; (202) Perform 2 on the time-domain channel response P Point FFT, obtain Where p is a set value; (203) Search The maximum power value is indexed. If the maximum power value exceeds the preset power threshold, a Doppler path exists. The index corresponds to the Doppler frequency shift size, and step (204) is executed. Otherwise, no Doppler path exists, and the process returns to step (201) to estimate the Doppler frequency shift of the next time delay. (204) Reconstruct the time-domain channel response corresponding to the Doppler frequency shift, eliminate the reconstructed time-domain channel response in the original time-domain channel response, and return to step (202).
2. The method for estimating Doppler frequency shift in an OTFS system according to claim 1, characterized in that, The specific process of step (204) is as follows: Let the index corresponding to the maximum power be The corresponding Doppler frequency shift is: In the formula, N is the number of symbols in a frame of the OTFS system; Reconstruct the time-domain channel response corresponding to the Doppler frequency shift for: Subtract the reconstructed time-domain channel response from the original time-domain channel response to obtain a new time-domain channel response, and then return to step (202).
Citation Information
Patent Citations
Low-dimensional subspace OTFS channel estimation method
CN113852579A
Sensitivity and inductance integrated signal design and beam forming method based on time delay alignment modulation
CN114866124A