CN113395222B - Channel prediction throughput optimization method based on non-uniform pilot frequency - Google Patents
Channel prediction throughput optimization method based on non-uniform pilot frequency Download PDFInfo
- Publication number
- CN113395222B CN113395222B CN202110648859.3A CN202110648859A CN113395222B CN 113395222 B CN113395222 B CN 113395222B CN 202110648859 A CN202110648859 A CN 202110648859A CN 113395222 B CN113395222 B CN 113395222B
- Authority
- CN
- China
- Prior art keywords
- matrix
- uniform
- frequency
- pilot
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/0202—Channel estimation
- H04L25/0224—Channel estimation using sounding signals
- H04L25/0226—Channel estimation using sounding signals sounding signals per se
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/0202—Channel estimation
- H04L25/0204—Channel estimation of multiple channels
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/0202—Channel estimation
- H04L25/024—Channel estimation channel estimation algorithms
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
- H04L43/0876—Network utilisation, e.g. volume of load or congestion level
- H04L43/0888—Throughput
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Power Engineering (AREA)
- Environmental & Geological Engineering (AREA)
- Radio Transmission System (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
A non-uniform pilot frequency-based channel prediction throughput optimization method is characterized in that a non-uniform time domain or frequency domain pilot frequency idea is adopted, multiple channel prediction algorithms are combined, non-uniform pilot frequency interval processing is achieved, higher model parameter resolution is obtained, channel prediction is conducted, and data transmission performance during communication is improved, namely throughput loss is reduced.
Description
Technical Field
The invention relates to the technology in the field of communication, in particular to a channel prediction throughput optimization method based on time domain and frequency domain non-uniform pilot frequency.
Background
In a large-scale multiple-input multiple-output (MIMO) wireless communication system, obtaining accurate Channel State Information (CSI) is crucial to ensuring the performance of a wireless link. CSI may represent the combined effects of path loss, scattering, diffraction, fading, shadowing, etc., as a signal propagates from a transmitter to its corresponding receiver. However, as the number of antennas at the base station increases, there will be processing or computational delays that cause the CSI to be inaccurate. The mismatch of CSI in learning and in beamforming is referred to as "channel aging", which can cause significant degradation in the throughput performance of the MIMO system, and channel prediction is considered to be an effective technique to mitigate this performance degradation.
The mainstream method of channel prediction is to predict the CSI at the pilot symbol at the future time first, and then to complete the CSI of the middle non-pilot symbol by an interpolation method. For the CSI prediction technology of non-pilot symbols, a relatively classical low-pass or band-pass interpolation algorithm exists. The conventional low-pass interpolation algorithm can well complement the CSI of the non-pilot symbols at a low speed, but at a high speed, due to the limitation of nyquist sampling theorem, the pilot spacing cannot satisfy the condition of twice the maximum doppler frequency, and thus an aliased doppler spectrum cannot be recovered. The band-pass interpolation algorithm is more suitable for high-speed scenes, and the main energy is concentrated in a certain bandwidth under the high-speed condition, so that the main energy can be approximately regarded as a band-pass signal to carry out band-pass difference.
However, in an actual channel, it is difficult for a signal to perfectly conform to the precondition of a low-pass or band-pass signal, and although a doppler spectrum is slowly changed in the moving process of a user, if the doppler spectrum is observed for a long time, both the cut-off frequency and the center frequency of the signal may be changed, thereby affecting the performance of low-pass or band-pass interpolation. Besides, in order to determine the cut-off frequency parameter in the low-pass interpolation algorithm and the cut-off frequency and the center frequency parameter in the band-pass algorithm, a priori knowledge of the doppler spectrum is required, however, in actual operation, obtaining a priori information of the doppler spectrum means a larger overhead, and the amplitude information of the spectrum is unknown.
Therefore, for channel prediction in a high-speed scene, both low-pass and band-pass interpolation algorithms have defects, and a non-band-pass Doppler spectrum cannot be processed, so that the Doppler spectrum sensing problem under the more general condition of compressed sensing processing is introduced, and SRS prediction and non-SRS interpolation in a low-speed scene are combined into a whole. On the premise of the same pilot frequency overhead, a new channel prediction idea is to obtain more accurate channel prediction performance and higher throughput by a design idea of time domain frequency domain non-uniform pilot frequency and utilizing a compressed sensing technology which is mostly used for image processing.
Disclosure of Invention
Aiming at the problem that the prior art can not perform more accurate channel prediction at high speed, the invention provides a channel prediction throughput optimization method based on non-uniform pilot frequency, which realizes non-uniform pilot frequency interval processing by adopting the idea of non-uniform time domain or frequency domain pilot frequency and utilizing the combination of various channel prediction algorithms, obtains higher model parameter resolution and performs channel prediction, and realizes the improvement of data transmission performance during communication, namely the reduction of throughput loss.
The invention is realized by the following technical scheme:
the invention relates to a channel prediction throughput optimization method based on non-uniform pilot frequency, which specifically comprises the following steps:
step 1: inputting y to be an original signal with the size of Nx 1 in the time domain, and obtaining a non-uniform compressed observation signal S ═ y of y, wherein: the size of S is M multiplied by 1, phi is a measuring matrix with the size of M multiplied by N, and the non-uniform pilot frequency S is obtained by setting the measuring matrix.
The non-uniform pilot frequency refers to: the time or frequency pilot intervals are arrays of varying sizes.
The array is not limited to have a certain change rule or random distribution.
Preferably, a space is provided between the minimum value and the maximum value in the array.
Step 2: since y is usually not sparse in the time domain and sparse in the doppler domain, let ψ be the transform matrix, K sparse x ψ y, convert the observation matrix to S Tx, where T Φ ψ H Is a sensing matrix of size M × N.
And step 3: performing signal recovery on the compressed observation signal S by using an Orthogonal Matching Pursuit (OMP) algorithm based on greedy pursuit with low complexity in compressed sensing to obtain a sparse parameter x in the step 2, namely an angle parameter omega, and converting the angle omega into a real frequency
And 4, step 4: after the signal frequency is recovered by the OMP method, the signal amplitude is recovered by a Regularized Least Square (RLS) method,wherein W is a matrix of samples, where,
and 5: the frequency parameter obtained in the step 3And the amplitude parameter obtained in step 4Model of incoming channel In (3), signal recovery is performed.
Technical effects
The invention integrally improves the channel prediction performance under the high-speed condition of occupying the same pilot frequency resource and under the same communication condition in a high-mobility scene by using a channel prediction method based on non-uniform pilot frequency setting.
Drawings
FIG. 1 is a diagram of uniform pilots and non-uniform pilots;
FIG. 2 is a diagram of time domain and frequency domain uniform pilot, time domain non-uniform pilot, frequency domain non-uniform pilot, and time domain and frequency domain non-uniform pilot;
in the figure: the horizontal axis is the time domain and the vertical axis is the frequency domain;
FIG. 3 is a flow chart of the present invention;
FIG. 4 is a diagram illustrating channel prediction throughput for non-uniform pilot and uniform pilot in the time domain;
in the figure: the average throughput loss of the compared burg algorithm is 36.62%, and the average throughput loss of the time-domain non-uniform pilot is 23.75%;
FIG. 5 is a diagram of channel predicted throughput for non-uniform pilots and uniform pilots in time and frequency domains;
in the figure: the average throughput loss of the compared burg algorithm is 42.07%, and the average throughput loss of the time domain frequency domain non-uniform pilot is 37.48%.
Detailed Description
The implementation scenario of this embodiment is an MIMO channel in a high-speed scenario: the method is characterized in that a transmitting end is a rectangular antenna array, a receiving end is a uniform linear array multi-antenna system, a channel prediction algorithm, specifically a combination of an Orthogonal Matching Pursuit (OMP) algorithm and a Regularized Least Square (RLS) algorithm, is adopted by adopting non-uniform time domain or frequency domain pilot frequency in a physical sense, higher model parameter resolution is obtained, and reduction of throughput loss during data transmission during communication is realized.
As shown in fig. 3, this embodiment specifically includes the following steps:
step 1, inputting y to an original signal with the time domain size of nx1, and obtaining a compressed observation signal S of y ═ phiy, wherein: s is M multiplied by 1, phi is a measuring matrix with the size of M multiplied by N, and the non-uniform compression observation signal S, namely the non-uniform pilot frequency, is obtained through setting the measuring matrix.
The sensing matrix T needs to meet the finite equidistant property (RIP) to recover x with higher probability, specifically: wherein: 0<δ<1。
Preferably, the sensing matrix is a random gaussian matrix, a Partial Fourier Matrix (PFM), or the like. In some high-speed scenes, due to the limitation of the density of pilot points, a random gaussian matrix (the gaussian matrix needs to obtain CSI information of all symbols) cannot be applied, so that a partial fourier matrix is often selected as a sensing matrix, and when in an application scene, signals are sparse in a doppler domain, so that a transformation matrix psi in the sensing matrix T is a DFT matrix of N × N, and by selecting the matrixRandomly selecting the K rows in ψ is achieved. As can be seen from the foregoing y pilots, i.e., the compressed observed signals are S ═ Φ y, the non-uniform pilot point distribution setting can be achieved by modifying and determining the values of the selection matrix Φ.
The selection matrix shows that the distribution of the pilot frequency is non-uniform in the practical application scene, and the distribution condition of the non-uniform pilot frequency can be set by modifying the selection matrix, specifically: for the time domain non-uniform pilot, the intervals between pilot points are not fixed but have different sizes, the sizes can be arranged in a certain order or randomly distributed, the displayed pilot points are non-uniformly distributed in time, at this time, the frequency is still uniformly distributed, i.e., the pilot intervals in frequency are fixed, the schematic diagram is shown in the upper right corner of fig. 2, and the performance verification result is shown in fig. 4.
In addition to the time domain non-uniform pilot frequency, similarly, non-uniform pilot frequency setting may also be performed in the frequency domain, which is a similar way to the time domain non-uniform pilot frequency setting, that is, the frequency domain is non-uniformly set on the basis of the time domain non-uniformity to implement the time-frequency domain non-uniformity, at this time, the time-frequency domain needs to be processed respectively, and the minimum interval of the time-frequency domain pilot frequency needs to be reduced, so that the resolution of parameters is improved in the time-frequency domain to improve the performance of channel estimation or channel prediction, a schematic diagram of which is shown in the lower right corner of fig. 2, and a performance verification result of which is shown in fig. 5.
In order to improve the method under the condition of the same overhead for fair comparison, the density of non-uniform pilot frequency in a time domain or a frequency domain is set to be the same as the density of equally-spaced pilot frequency in a protocol, only the minimum spacing of the pilot frequency is smaller, and in a physical sense, the smaller pilot frequency granularity can possibly distinguish larger Doppler frequency.
And 3, performing signal recovery on the compressed observation signal S by adopting an Orthogonal Matching Pursuit (OMP) algorithm which is low in complexity and based on greedy pursuit in compressed sensing, and specifically: selecting columns in the measurement matrix in a greedy iteration mode, enabling the selected columns in each iteration to be maximally related to the current redundant vector, subtracting related parts from the measurement vector to be residual errors, repeatedly iterating the residual errors, and forcibly stopping iteration until iteration times reach sparsity, so that the frequency position of the signal Doppler spectrum in a sparse domain can be recovered, a required angle parameter omega is obtained, and the angle omega is converted into a real frequency
In the OMP process, the number K of effective paths is smaller than the number M of pilot frequencies and smaller than the window size N, matching tracking is to select a column which is most matched with a signal x in a sensing matrix T, construct a sparse matrix, solve a signal residual error and then iterate.
Step 4, after restoring the signal frequency by using the OMP method, restoring the signal amplitude by using a Regularized Least Square (RLS) method, so that the problem of overfitting which is easy to exist when the number of samples of data is far larger than the dimensionality by using a least square method (LS) can be solved, and the method specifically comprises the following steps: a sample matrix is first constructed, the sample set of the matrix being represented by the matrix W, with a size of M x K, i.e.,then the amplitude is represented by RLS
Step 5, the frequency parameter obtained in the step 3 is usedAnd the amplitude parameter obtained in step 4Model of incoming channel Wherein Q is the number of time domain points, f s Is the sampling frequency.
The embodiment is specifically realized by using Cost2100 analog simulation data, and the Cost2100 channel model is a random channel model based on geometry and can show the characteristics of a multi-user MIMO channel along with time, frequency and space changes. The principle of the stochastic channel model is to model the stochastic nature of the radio channel by analyzing the geometric distribution of objects (or scatterers) in the environment that scatter radio waves. In the random channel model, the radio channel is formed by the superposition of propagation paths, which are called multipath components. The multipath components are generated due to interaction between radio waves and objects in the environment, each multipath component has corresponding time delay and angle information, and the set speed of original simulation data meets a high-speed scene of 90km/h or more.
The performance verification of the time domain non-uniform pilot frequency is carried out by utilizing the Cost2100 analog simulation data, the channel prediction of a compressed sensing method is carried out on 120km/h channel data under the time domain non-uniform pilot frequency, the distribution of the non-uniform pilot frequency is set to be a fixed non-uniform position, the obtained prediction part is compared with the throughput of the burg method, and a schematic diagram of the channel prediction throughput of the time domain non-uniform pilot frequency and the uniform pilot frequency is obtained.
Similarly, the Cost2100 analog simulation data is used for verifying the time-frequency domain non-uniform pilot frequency performance, channel prediction of a compressed sensing method is carried out on 90km/h channel data under the time-frequency domain non-uniform pilot frequency, the distribution of the non-uniform pilot frequency is set to be a fixed non-uniform position, an obtained prediction part is compared with the throughput of the burg method, a schematic diagram of the channel prediction throughput of the frequency domain non-uniform pilot frequency and the uniform pilot frequency is obtained, it can be seen from the diagram that the non-uniform pilot frequency has a certain effect on improvement of the throughput performance, the average throughput loss of a compared burg algorithm is 42.07%, the average throughput loss of the frequency domain non-uniform pilot frequency is 37.48%, and the throughput loss performance is improved by about 4.5%.
The foregoing embodiments may be modified in several ways by those skilled in the art without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and not by the foregoing embodiments, and each implementation within its scope is limited by the present invention.
Claims (6)
1. A channel prediction throughput optimization method based on non-uniform pilot frequency is characterized by comprising the following steps:
step 1: inputting y to be an original signal with the size of Nx 1 in the time domain, and obtaining a non-uniform compressed observation signal S ═ y of y, wherein: the size of S is Mx 1, phi is a measurement matrix with the size of Mx N, and non-uniform pilot frequency S is obtained by setting the measurement matrix;
step 2: let ψ be the transformation matrix, K sparse x ψ y, and convert the observation matrix to S Tx, where T Φ ψ H Is a sensing matrix of size mxn;
and step 3: performing signal recovery on the compression observation signal S by adopting an orthogonal matching pursuit algorithm based on greedy pursuit to obtain a sparse parameter x in the step 2, namely an angle parameter omega, and converting the angle omega into a real frequency
And 4, step 4: after the signal frequency is restored by the OMP method, the signal amplitude is restored by the regularization least square method, wherein W is a matrix of samples, where,
2. The non-uniform pilot-based channel prediction throughput optimization method of claim 1, wherein the non-uniform pilot refers to: the time or frequency pilot intervals are arrays of different sizes, and an interval is arranged between the minimum value and the maximum value in the arrays.
4. The non-uniform pilot-based channel prediction throughput optimization method of claim 1 or 3, wherein a partial Fourier matrix is used as the sensing matrix, and when the signal is sparse in the Doppler domain, the transformation matrix psi in the sensing matrix T is an NxN DFT matrix, and the transformation matrix psi is selected by selecting the matrixRandomly selecting K rows in psi is achieved and the non-uniform pilot point distribution setting is achieved by modifying and determining the values of the selection matrix phi.
5. The non-uniform pilot-based channel prediction throughput optimization method of claim 2, wherein for the non-uniform pilot in the time domain, the intervals between pilot points are not fixed but are different in size, the sizes are arranged in sequence or randomly distributed, and the pilot points are displayed in a non-uniform distribution in time, while the frequencies are still uniformly distributed, i.e. the pilot intervals in frequency are fixed;
for the time-frequency domain non-uniform pilot frequency, namely, the frequency domain is non-uniformly set on the basis of the time domain non-uniformity, the non-uniformity of the time-frequency domain is realized.
6. The non-uniform pilot based channel prediction throughput optimization method of claim 5, wherein for the non-uniform pilots in time-frequency domain, the time-frequency domain is processed respectively to reduce the minimum interval of the time-frequency domain pilots, thereby increasing the resolution of the parameters in the time-frequency domain.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110648859.3A CN113395222B (en) | 2021-06-10 | 2021-06-10 | Channel prediction throughput optimization method based on non-uniform pilot frequency |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110648859.3A CN113395222B (en) | 2021-06-10 | 2021-06-10 | Channel prediction throughput optimization method based on non-uniform pilot frequency |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113395222A CN113395222A (en) | 2021-09-14 |
CN113395222B true CN113395222B (en) | 2022-08-05 |
Family
ID=77620394
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110648859.3A Active CN113395222B (en) | 2021-06-10 | 2021-06-10 | Channel prediction throughput optimization method based on non-uniform pilot frequency |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113395222B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2478360A (en) * | 2010-03-05 | 2011-09-07 | Toshiba Res Europ Ltd | Optimal pilot symbol design/placement in spatial and delay correlated MIMO OFDM channels |
CN104780128A (en) * | 2015-04-14 | 2015-07-15 | 哈尔滨工程大学 | Sparse channel estimation and pilot optimization method for underwater sound OFDMA uplink communication |
CN111726308A (en) * | 2020-06-15 | 2020-09-29 | 哈尔滨工程大学 | Orthogonal matching pursuit channel estimation method based on frequency response pre-interpolation |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8824527B2 (en) * | 2011-11-15 | 2014-09-02 | Acorn Technologies, Inc. | OFDM receiver with time domain channel estimation |
CN105981320B (en) * | 2014-02-19 | 2019-06-21 | 华为技术有限公司 | A kind of channel state information feedback method, device, user equipment and base station |
CN103997397B (en) * | 2014-04-30 | 2019-10-01 | 北京邮电大学 | The non-homogeneous pilot design method in airspace |
CN112929301B (en) * | 2019-12-08 | 2022-05-31 | 中国科学院声学研究所 | OFDM underwater acoustic time-varying Doppler frequency offset estimation method based on non-uniform space-time carrier |
-
2021
- 2021-06-10 CN CN202110648859.3A patent/CN113395222B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2478360A (en) * | 2010-03-05 | 2011-09-07 | Toshiba Res Europ Ltd | Optimal pilot symbol design/placement in spatial and delay correlated MIMO OFDM channels |
CN104780128A (en) * | 2015-04-14 | 2015-07-15 | 哈尔滨工程大学 | Sparse channel estimation and pilot optimization method for underwater sound OFDMA uplink communication |
CN111726308A (en) * | 2020-06-15 | 2020-09-29 | 哈尔滨工程大学 | Orthogonal matching pursuit channel estimation method based on frequency response pre-interpolation |
Non-Patent Citations (1)
Title |
---|
面向5G***基于压缩感知理论的信道状态获取研究;徐晶;《中国优秀硕士学位论文全文数据库 信息科技辑》;20150815;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN113395222A (en) | 2021-09-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Li et al. | Deep residual learning meets OFDM channel estimation | |
CN113507426B (en) | OTFS modulation-based joint channel estimation and signal detection method and device | |
CN104052691B (en) | MIMO-OFDM system channel estimation method based on compressed sensing | |
CN106559367B (en) | MIMO-OFDM system millimeter waves channel estimation methods based on low-rank tensor resolution | |
Wei et al. | An AMP-based network with deep residual learning for mmWave beamspace channel estimation | |
CN108881076B (en) | MIMO-FBMC/OQAM system channel estimation method based on compressed sensing | |
CN111030952B (en) | Beam space channel estimation method and system of millimeter wave system | |
CN109104252B (en) | Testing device and testing method for multi-input multi-output mode system | |
CN109005133B (en) | Double-sparse multi-path channel model and channel estimation method based on model | |
CN104869086B (en) | MIMO ofdm communication system down channels method of estimation, device based on two dimensional compaction perception | |
CN108259397B (en) | Large-scale MIMO system channel estimation method based on adaptive regularization subspace tracking compressed sensing algorithm | |
CN110380994B (en) | Fast Bayesian matching pursuit marine sparse channel estimation method | |
Rodríguez-Fernández et al. | A frequency-domain approach to wideband channel estimation in millimeter wave systems | |
CN108199990B (en) | non-Gaussian noise 3D-MIMO channel estimation method | |
Rodríguez-Fernández et al. | A compressive sensing-maximum likelihood approach for off-grid wideband channel estimation at mmWave | |
CN113595944A (en) | Channel estimation method for millimeter wave MIMO hybrid precoding system | |
Uwaechia et al. | Compressed channel estimation for massive MIMO-OFDM systems over doubly selective channels | |
CN1150689C (en) | Iterative projection with initialisation using network user identifier | |
CN115606157A (en) | Channel estimation for antenna arrays | |
US10608686B1 (en) | Circuit and method for enabling channel denoising in a wireless communication apparatus | |
Chen et al. | Channel estimation for massive MIMO-OTFS system in asymmetrical architecture | |
CN113395222B (en) | Channel prediction throughput optimization method based on non-uniform pilot frequency | |
Zhang et al. | An optimization for channel estimation based on compressed channel sensing | |
CN114978818B (en) | Adaptive channel estimation method and system based on compressed sensing | |
CN106534028B (en) | OFDM channel tracking method based on compressed sensing |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |