Search

CN-122027402-A - OTFS channel estimation method based on multi-superimposed pilot frequency and stripe energy detection

CN122027402ACN 122027402 ACN122027402 ACN 122027402ACN-122027402-A

Abstract

The invention belongs to the technical field of wireless communication channel estimation, and particularly relates to a channel estimation method based on orthogonal time-frequency space (OTFS) modulation of multi-superimposed pilot frequency stripe energy detection. The invention carries out discretization modeling on the time delay-Doppler (DD) domain channel and carries out optimization estimation by combining independent observation of multiple superimposed pilots, thereby obtaining higher estimation precision under the condition of fractional Doppler leakage and superimposed data interference. The method comprises the specific steps that a sending end establishes a truncated equivalent tap model based on a fractional Doppler channel, a plurality of mutually orthogonal superimposed pilot symbols are placed at different positions of a DD domain, a receiving end carries out threshold detection on leakage strip energy in the Doppler direction, and a plurality of pilot independent observations of a super-threshold part are extracted for joint estimation to obtain a refined channel estimation result. The invention can effectively capture the fraction Doppler leakage energy which is easy to lose in the traditional single-point detection, reduce the detection omission risk and improve the channel estimation performance under the condition of fraction Doppler while meeting the actual system requirements.

Inventors

  • HUANG JIE
  • XU CHONGBIN

Assignees

  • 复旦大学

Dates

Publication Date
20260512
Application Date
20260309

Claims (8)

  1. 1. The OTFS channel estimation method based on multi-superposition pilot frequency and stripe energy detection is characterized in that the method covers fractional Doppler leakage energy by constructing a truncated equivalent tap model, improves anti-interference capacity by using multi-pilot frequency scattered superposition, combines pilot frequency observation values, and obtains better estimation performance, and the method comprises the following specific steps: step (1), aiming at energy leakage caused by fractional Doppler, constructing a DD domain truncated equivalent tap model: Under fractional Doppler, the leakage nuclear energy is highly concentrated near the main lobe and decays along with the offset in a side lobe, and the preset cut-off radius To cover Doppler relative shift And will Is incorporated into the equivalent interference term so that each physical path can be equivalently represented as The discrete equivalent taps are used for converting the continuous fraction Doppler parameter estimation problem into a finite dimension block sparse discrete estimation problem with structural support in a DD domain; wherein due to the truncated radius To make the equivalent maximum Doppler spread from Becomes as follows In order to ensure that the observation areas of any two pilot symbols corresponding to each pilot frequency are not overlapped after the DD domain channel is passed, an observation area isolation rule is set for constraint, namely, the distance between adjacent pilot frequencies is not less than the maximum delay spread or not less than twice of the equivalent maximum Doppler spread, thereby determining the equivalent observation area boundary corresponding to each pilot frequency symbol at a receiving end so as to obtain mutually independent pilot frequency observation for joint estimation; step (2), designing a multi-superposition pilot frequency frame structure: Embedding a plurality of pilot symbols into different positions of a transmission frame in a delay-Doppler grid according to the equivalent observation area boundary determined in the step (1) according to the observation area isolation rule, and superposing the pilot symbols and the data symbols on the same grid point to construct a DD domain transmission matrix Wherein For the doppler dimension index to be used, Index for time delay dimension; Step (3), OTFS signaling: for the DD domain transmission matrix in step (2) Sequentially executing inverse octyl Fourier transform and Hessenberg transform, converting into a time domain continuous signal and transmitting the time domain continuous signal to a wireless channel; step (4), OTFS signal reception: Receiving the time domain signal, and sequentially performing Wiggner transformation and octyl Fourier transformation to obtain DD domain receiving matrix ; Step (5), partition scanning and band energy threshold detection: For each pilot position Wherein , The number of pilot frequencies is abbreviated as Establishing a zone scanning window based on maximum delay and Doppler spread Calculating the energy of the band of each candidate point in the window in the Doppler direction Statistics of the stripe energy And a preset threshold value Comparing, and extracting candidate points exceeding a threshold value to form a candidate set; Step (6), multi-pilot joint channel estimation: phase compensating observation points within the candidate set to eliminate delay-Doppler phase rotation, utilizing the And carrying out joint estimation on independent observation values of the pilot symbols at the corresponding observation points respectively based on a least square criterion to obtain an initial estimation value of the equivalent tap coefficient.
  2. 2. The OTFS channel estimation method according to claim 1, wherein the constructing the DD domain truncated equivalent tap model in step (1) is performed by Input-output relationship equivalent; ; Wherein, the The symbol matrix is transmitted for the DD domain, Is an equivalent noise term; As the number of the multi-paths, Is the first Complex gain of the strip path; And Respectively the first Integer doppler and delay index of the strip path, As the number of fractional doppler indices, And Respectively represent pairs of And The operation of the module taking is circulated, Is an equivalent noise term; The Doppler leakage kernel has the expression: ; the input-output relationship is then converted into a truncated equivalent tap model, the expression of which is as follows: ; Wherein the equivalent tap coefficient Is defined as Is a deterministic phase factor; ; I.e. equivalent noise term The method comprises the steps of adding noise and residual fraction Doppler interference which is not captured by a truncated model, wherein the expression of the residual fraction Doppler interference is as follows: 。
  3. 3. The OTFS channel estimation method of claim 2, wherein the superimposed pilot symbol placement rule in step 2 comprises selecting on a delay-Doppler grid Pilot frequency point [ ] ) Wherein The following conditions are satisfied between any two pilots to avoid mutual interference in the DD domain: (a) The cyclic distance of the Doppler domain is equal to or greater than twice the equivalent maximum Doppler spread: ; (b) The cyclic distance of the delay domain is greater than or equal to the maximum delay spread ; Wherein, the For the maximum doppler spread to be the maximum doppler spread, For the maximum delay spread to be the largest, Is the equivalent leak cutoff radius.
  4. 4. The OTFS channel estimation method according to claim 3, wherein the OTFS signal is sent in step 3, specifically: at the transmitting end, the DD domain superposition transmitting matrix is firstly transformed (ISFFT) by inverse octyl Fourier transform Mapped to Time-Frequency (TF) domain: ; Wherein, the And Representing the time and frequency indexes of the time-frequency grid respectively, and then converting the time-frequency domain symbol by utilizing the Hessenberg transformation Modulated into a continuous time-domain transmission signal : ; Wherein, the In order to transmit the pulse-shaping waveform, In order to be a symbol period, Is the subcarrier spacing.
  5. 5. The OTFS channel estimation method according to claim 4, characterized in that the OTFS signal is received in step 4, in particular: The receiving end executes the inverse transformation of the above process, namely, sequentially demodulates through the Wiggner transformation, and re-projects the signals from the time-frequency domain back to the DD domain in cooperation with the octyl Fourier transformation, and finally recovers to obtain the receiving matrix for the subsequent channel estimation and detection 。
  6. 6. The OTFS channel estimation method according to claim 5, wherein the partition scan window in step 5 is: wherein ; Is the delay-Doppler shift relative to pilot position, the strip energy statistic The calculation formula of (2) is as follows: ; by length in Doppler direction All the energy in the band is accumulated to compensate for the energy spread caused by the fractional doppler.
  7. 7. The OTFS channel estimation method according to claim 6, wherein the preset threshold in step 5 The method meets the following conditions: , wherein, Representing the noise energy of the transmission channel, Represents the average value of all data symbol energies in a transmitted data frame of one DD domain of the transmitting end, Is a scaling factor determined based on a preset false alarm probability.
  8. 8. The method for OTFS channel estimation according to claim 7, wherein the phase compensation of the received signal containing pilot component in step (6) is performed by a compensation term based on the position of the pilot at the transmitting end Determining relative displacement between the pilot frequency and the corresponding scanning point, wherein the scanning point is the offset point of the pilot frequency on the time delay-Doppler grid, and the relative displacement comprises Doppler offset Offset from time delay I.e. , And (2) and The compensation term is used to eliminate the delay-Doppler phase rotation generated by OTFS conversion to obtain the recovered received signal The joint estimation takes the form of a minimum linear mean square error: ; Wherein, the Is the first The conjugate of the individual superimposed pilot symbols, For a priori statistical power of the channel tap gains, Is the equivalent total interference noise power including additive noise and superimposed data interference.

Description

OTFS channel estimation method based on multi-superimposed pilot frequency and stripe energy detection Technical Field The invention belongs to the technical field of wireless communication, in particular to a channel estimation method of an OTFS modulation system, and particularly relates to an OTFS channel estimation method based on multi-superimposed pilot frequency and strip energy detection under a fractional Doppler scene. Background The OTFS modulation technology converts a fading channel which changes rapidly along with time into near quasi-static sparse representation of the DD domain by two-dimensional modulation in the DD domain, and has great potential in a high-speed mobile communication scene. Accurate Channel State Information (CSI) is a precondition for achieving highly reliable data detection and equalization. In existing OTFS channel estimation schemes, an Embedded Pilot (Embedded Pilot) typically needs to be configured with a guard interval around the Pilot to suppress interference between the Pilot and data, but this causes significant spectral efficiency loss. The superimposed pilot (Superimposed Pilot) improves the spectrum efficiency by overlapping the pilot and the data in energy, but also brings the problem of increased pilot detection difficulty. In particular in practical communication environments, the fractional doppler effect can cause significant leakage of energy along the doppler axis. The traditional estimation scheme based on single-point pulse or single-point energy detection is extremely easy to cause the problem of detection accuracy reduction when leakage energy exists and is interfered by superimposed data. Therefore, how to design a low-complexity channel estimation scheme capable of ensuring spectrum efficiency and stably coping with fractional doppler leakage is a technical problem of the current OTFS system trend engineering application. Disclosure of Invention The invention aims to provide an OTFS channel estimation method with low detection omission risk and high estimation precision based on multi-superposition pilot frequency and stripe energy detection, which aims to solve the technical problems of detection precision reduction caused by energy leakage, low frequency spectrum efficiency caused by embedded pilot frequency due to a guard interval and the like in a fractional Doppler scene in the existing OTFS channel estimation technology. The truncated equivalent tap model is constructed to cover the fractional Doppler leakage energy, the anti-interference capacity is improved by utilizing multi-pilot scattered superposition, and the pilot observation values are combined to obtain better estimation performance. The OTFS channel estimation method based on multi-superposition pilot frequency and stripe energy detection provided by the invention covers fractional Doppler leakage energy by constructing a truncated equivalent tap model, improves anti-interference capacity by using multi-pilot frequency scattered superposition, combines pilot frequency observation values, and obtains better estimation performance, and the method comprises the following specific steps: In the step (1), considering that in an actual communication environment, the fractional Doppler can cause severe leakage of energy along the Doppler axis, the step converts the continuous parameter estimation problem into the finite dimension block sparse discrete estimation problem based on structural support by discretizing and modeling a DD domain channel. Specifically, the preset cutoff radiusTo cover Doppler relative shiftAnd willThe tail leakage component of (c) incorporates an equivalent interference term. At this time, the response of each physical propagation path in the DD domain no longer appears as a single discrete point, but appears as an integer coordinateIs centered and has a length in the Doppler dimensionIs a sparse block of (1); Definition of the first embodiment Block sparse discrete support set corresponding to strip pathThe method comprises the following steps: ; the support set forms a Doppler direction length of Such that the DD domain channel matrix behaves asThe linear superposition of the sparse blocks exhibits significant structured sparse properties in that the non-zero elements are not randomly distributed, but appear in a clustered fashion along the Doppler axis, forming a particular banding pattern of energy blocks. Wherein due to the truncated radiusTo make the equivalent maximum Doppler spread fromBecomes as follows. In order to ensure that the observation areas of any two pilot symbols corresponding to each pilot frequency are not overlapped after the DD domain channel is passed, an observation area isolation rule is set for constraint, namely, the distance between adjacent pilot frequencies is not less than the maximum delay spread or not less than twice of the equivalent maximum Doppler spread, thereby determining the equivalent observation area boundary correspond