CN110890931B - Uplink time-varying channel prediction method based on improved Prony method - Google Patents

Abstract

The invention relates to an uplink time-varying channel prediction method based on an improved Prony method, which is applied to a TDD system and comprises the following steps: s1, calculating an initial Prony coefficient; s2, predicting CSI at the next moment according to the Prony coefficient; s3, decoding the data symbols sent by the user according to the predicted CSI; s4, estimating CSI by using the decoded data symbols; s5, updating the Prony coefficient by using the estimated CSI; s6, predicting CSI of the next moment; and S7, decoding the next data symbol sent by the user and repeating the steps. The uplink time-varying channel prediction method based on the improved Prony method can accurately predict and estimate the CSI of the channel at different moments, and improves the efficiency of channel prediction, thereby improving the overall performance of the system.

Description

Uplink time-varying channel prediction method based on improved Prony method

Technical Field

The invention relates to the technical field of wireless communication, in particular to an uplink time-varying channel prediction method based on an improved Prony method.

Background

In a massive MIMO system, due to massive antennas at the base station, channels between different users and the base station are progressively orthogonal, so that the data transmission rate and energy efficiency of the system can be greatly improved by using only a simple signal processing technique. However, this advantage is obtained based on the assumption that the base station can accurately estimate the CSI, and therefore, the CSI acquisition is very important for the system.

In the fast time-varying channel, the CSI at the current time and the CSI at the next time are different, and the CSI at each time needs to be estimated. Whereas the conventional channel estimation method can estimate only CSI in a fixed state. Therefore, for time-varying channels, the conventional approach is no longer applicable. Based on the Prony method, the user only needs to send a small amount of pilot signals, and the base station processes the pilot signals to obtain the CSI estimation value of the corresponding moment so as to predict and estimate the CSI of the later moment. However, in the channel estimation based on the conventional Prony method, the number of required pilot signals must be larger than the number of paths of the channel. Therefore, there is a need for an improvement in the Prony method to improve the efficiency of channel prediction.

Disclosure of Invention

The invention aims to provide an uplink time-varying channel prediction method based on an improved Prony method aiming at the defects of the prior art.

In order to achieve the purpose, the invention adopts the following technical scheme:

an uplink time-varying channel prediction method based on an improved Prony method is applied to a TDD (time division duplex) system, and comprises the following steps:

s1, calculating an initial Prony coefficient;

s2, predicting CSI (channel state information) at the next moment according to the Prony coefficient;

s3, decoding the data symbols sent by the user according to the predicted CSI;

s4, estimating CSI by using the decoded data symbols;

s5, updating the Prony coefficient by using the estimated CSI;

s6, predicting CSI of the next moment;

and S7, decoding the next data symbol sent by the user and repeating the steps.

Further, the base station in the TDD system has M_bRoot antenna, number of clusters L for a single user of a single antenna_cEach cluster has a main path.

Further, in S1, the method for calculating the initial Prony coefficient according to the CSI possessed by the base station includes the following steps:

s11, setting the actual CSI as:

wherein matrix A is of dimension M_b×L_cFormed by the steering vectors and the initial phases of the paths, M_bIndicates the number of antennas, L, of the base station_cRepresenting the total number of paths, v, of the signal transmission_jDenotes a doppler frequency component of the j-th path, (j ═ 0,1, …, L_c)

Let the sampling time interval be Δ t, and denote h (k) as the channel at the kth sampling time;

h(k)＝h(kΔt)；

s12, the base station sends N according to the user_pObtaining the first N pilot signals_pEstimation of CSI for each time instant

Order to

S13, establishing an equation set according to the improved Prony method:

s14, pair H₀SVD decomposition (singular value decomposition), H₀＝U₀D₀V₀ ^HWherein, H₀Is front N_p-1 matrix U of CSI₀Is M_b×(N_p-1) unit arrays, D₀Is (N)_p-1)×(N_p-1) diagonal matrix with elements on the diagonal matrix arranged in descending order, V₀Is (N)_p-1)×(N_p-1) setting a number e greater than 0 for the unit matrix, and removing singular values less than e for increasing the stability of the equation; suppose there is N_qIf the singular value is greater than the element of E, then take U₀And V₀Front N of_qThe columns form a new matrix

Get D₀Front N of_qThe diagonal elements form a new diagonal matrix

S15, solving the equation by using a least square method to obtain an initial Prony coefficient:

further, the CSI predicted at the next time in S2 is specifically as follows:

further, the decoding the data symbols sent by the user in S3 includes the following steps:

the user sends data symbols s to the base station_mThe first transmitted signal is denoted s₁Where m is 1, the signal received by the base station is y_mRepresents:

where ρ is the signal-to-noise ratio, w_mUsing prediction of Gaussian noise for standard complex normal distribution based on signals received by the base station

Decoding signals transmitted by users, for signals transmitted by users

And (4) showing.

Further, the CSI estimation in S4 is specifically as follows:

wherein

Represents to the Nth_pThe CSI for + m time instances is the first estimated value,

will be provided with

Projected onto a matrix U₀The space thus generated isTo the final estimated CSI, as follows:

further, updating the Prony coefficient by using the estimated CSI in S5 specifically includes the following steps:

s51, construction matrix

Each time an estimate is obtained

It is estimated by the top N_p-1 pieces of

Put into a matrix with the dimension of (M +1) M_b×N_pWhen m is 1, the matrix is as follows:

s52, pair

Performing SVD decomposition in the same process

Obtaining a matrix by SVD

And updating the Prony coefficient vector, wherein the calculation process of the updated Prony coefficient vector is as follows:

further, in S6, the CSI predicted at the next time is specifically as follows:

further, S7 is specifically as follows:

making m equal to m +1, obtaining Nth step according to the above steps_pPredicted value of CSI at + m +1 times

By adopting the technical scheme of the invention, the invention has the beneficial effects that: compared with the prior art, the traditional channel estimation method cannot be suitable for the time-varying channel, the uplink time-varying channel prediction method based on the improved Prony method can accurately predict and estimate the CSI of the channel at different moments, and the method improves the efficiency of channel prediction, thereby improving the overall performance of the system.

Drawings

Fig. 1 is a simulation diagram of a system bit error rate of an uplink time varying channel prediction method based on an improved Prony method provided by the invention.

Detailed Description

Specific embodiments of the present invention will be further described with reference to the accompanying drawings.

An uplink time-varying channel prediction method based on an improved Prony method is applied to a TDD system, and comprises the following steps:

s1, calculating an initial Prony coefficient;

s2, predicting CSI (channel state information) at the next moment according to the Prony coefficient;

s3, decoding the data symbols sent by the user according to the predicted CSI;

s4, estimating CSI by using the decoded data symbols;

s5, updating the Prony coefficient by using the estimated CSI;

s6, predicting CSI of the next moment;

and S7, decoding the next data symbol sent by the user and repeating the steps.

The base station in the TDD system has M_bRoot antenna, number of clusters L for a single user of a single antenna_cEach cluster has a main path.

In S1, the method for calculating the initial Prony coefficient according to the CSI possessed by the base station includes the following steps:

s11, setting the actual CSI as:

wherein matrix A is of dimension M_b×L_cFormed by the steering vectors and the initial phases of the paths, M_bIndicates the number of antennas, L, of the base station_cRepresenting the total number of paths for signal transmission. v. of_jDenotes a doppler frequency component of the j-th path, (j ═ 0,1, …, L_c)，

Let the sampling time interval be Δ t, and denote h (k) as the channel at the kth sampling time;

h(k)＝h(kΔt)；

s12, the base station sends N according to the user_pObtaining the first N pilot signals_pEstimation of CSI for each time instant

Order to

S13, establishing an equation set according to the improved Prony method:

s14, pair H₀SVD decomposition (singular value decomposition), H₀＝U₀D₀V₀ ^HWherein H is₀Is front N_p1 matrix of CSI, U₀Is M_b×(N_p-1) unit arrays, D₀Is (N)_p-1)×(N_p-1) diagonal matrix with elements on the diagonal matrix arranged in descending order, V₀Is (N)_p-1)×(N_p-1) setting a number e greater than 0 for the unit matrix, and removing singular values less than e for increasing the stability of the equation; suppose there is N_qIf the singular value is greater than the element of E, then take U₀And V₀Front N of_qThe columns form a new matrix

Get D₀Front N of_qThe diagonal elements form a new diagonal matrix

S15, solving the equation by using a least square method to obtain an initial Prony coefficient:

specifically, the CSI predicted at the next time in S2 is as follows:

the decoding of the data symbols sent by the user in S3 includes the following steps:

the user sends data symbols s to the base station_mThe first transmitted signal is denoted s₁Where m is 1, the signal received by the base station is y_mRepresents:

where ρ is the signal-to-noise ratio, w_mUsing prediction of Gaussian noise for standard complex normal distribution based on signals received by the base station

Decoding user stationTransmitted signals, for signals transmitted by users

And (4) showing.

The CSI estimation in S4 is specifically as follows:

wherein

Represents to the Nth_pThe CSI at + m time instants is the first estimated value.

Will be provided with

Projected onto a matrix U₀The generated space, i.e. the final estimated CSI, is as follows:

in S5, updating the Prony coefficient by using the estimated CSI, specifically including the following steps:

s51, construction matrix

Each time an estimate is obtained

It is estimated by the top N_p-1 pieces of

Put into a matrix with the dimension of (M +1) M_b×N_pWhen m is 1, the matrix is as follows:

s52, pair

Performing SVD decomposition in the same process

Obtaining a matrix by SVD

And updating the Prony coefficient vector, wherein the calculation process of the updated Prony coefficient vector is as follows:

specifically, the CSI predicted at the next time in S6 is as follows:

s7 is specifically as follows:

making m equal to m +1, obtaining Nth step according to the above steps_pPredicted value of CSI at + m +1 times

The present invention assumes that the base station in the system has 32 antennas, and the antennas are distributed in 8 × 4 queues. For a single user with a single antenna, 19 clusters exist between the user and the base station, and the main path of each cluster is 1. The number of pilot signals transmitted by the users is 9, 15 and 20 respectively. The model of the channel adopts the standard proposed by 3GPP TR 36.873V 12.7.0(2017-12), and the parameters are shown in table 1:

TABLE 1

In the channel model, the time delay distribution scale factor r_τ3, the delay spread DS is 10m, and the shadow fading standard deviation ζ is 3 dB. Scale factor of AOA, AOD

ZOA, ZOD scale factor

Angle spread ASA 9m, ASD 10m, ZSA 10m, ZSD 10m, cluster ASA c _ASA22 deg. tuft

Cluster

Magnitude of offset angle alpha_mSelected from tables 7.3-3 in the standard. The size of the E is 10/rho, and a data symbol s sent by a user_mThe symbols are decoded using the maximum likelihood method for the elements in the standard 16-QAM.

FIG. 1 is a simulation chart of the bit error rate of the system under the above example conditions, where "N" is_p9 is a graph of the bit error rate versus the signal-to-noise ratio when the number of pilot signals transmitted by the user is 9, where "N" is_p15 is a plot of bit error rate versus signal-to-noise ratio for a user transmitting 15 pilot signals, where "N" is_p20 "is a graph of the bit error rate versus the signal-to-noise ratio when the number of pilot signals transmitted by the user is 20. As can be seen from fig. 1, in the preferred embodiment of the present invention, when the snr is higher, the number of pilot signals transmitted by the user is larger, the error rate is lower, and the performance of the system is better.

It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims (1)

1. An uplink time-varying channel prediction method based on an improved Prony method is applied to a TDD system, and is characterized in that the uplink time-varying channel prediction method comprises the following steps:

s1, calculating an initial Prony coefficient;

s2, predicting CSI at the next moment according to the Prony coefficient;

s3, decoding the data symbols sent by the user according to the predicted CSI;

s4, estimating CSI by using the decoded data symbols;

s5, updating the Prony coefficient by using the estimated CSI;

s6, predicting CSI of the next moment;

s7, decoding the next data symbol sent by the user and repeating the steps;

the base station in the TDD system has M_bRoot antenna, number of clusters L for a single user of a single antenna_cEach cluster has a main path; in S1, the method for calculating the initial Prony coefficient according to the CSI possessed by the base station includes the following steps:

s11, setting the actual CSI as:

wherein matrix A is of dimension M_b×L_cFormed by the steering vectors and the initial phases of the paths, M_bIndicates the number of antennas, L, of the base station_cRepresenting the total number of paths, v, of the signal transmission_jDoppler frequency component representing the jth path, j being 0,1, …, L_c；

Let the sampling time interval be Δ t, and denote h (k) as the channel at the kth sampling time;

h(k)＝h(kΔt)；

s12, the base station sends N according to the user_pObtaining the first N pilot signals_pEstimation of CSI for each time instant

Order to

S13, establishing an equation set according to the improved Prony method:

s14, pair H₀Carrying out SVD decomposition of H₀＝U₀D₀V₀ ^H，H₀Is front N_p-1 matrix of CSI) where U₀Is M_b×(N_p-1) unit arrays, D₀Is (N)_p-1)×(N_p-1) diagonal matrix with elements on the diagonal matrix arranged in descending order, V₀Is (N)_p-1)×(N_p-1) setting a number e greater than 0 for the unit matrix, and removing singular values less than e for increasing the stability of the equation; suppose there is N_qIf the singular value is greater than the element of E, then take U₀And V₀Front N of_qThe columns form a new matrix

Get D₀Front N of_qThe diagonal elements form a new diagonal matrix

S15, solving the equation by using a least square method to obtain an initial Prony coefficient:

specifically, the CSI predicted at the next time in S2 is as follows:

the decoding of the data symbols sent by the user in S3 includes the following steps:

the user sends data symbols s to the base station_mThe first transmitted signal is denoted s₁Where m is 1, the signal received by the base station is y_mRepresents:

where ρ is the signal-to-noise ratio, w_mUsing prediction of Gaussian noise for standard complex normal distribution based on signals received by the base station

Decoding signals transmitted by users, for signals transmitted by users

Represents;

the CSI estimation in S4 is specifically as follows:

wherein

Represents to the Nth_pThe CSI for + m time instances is the first estimated value,

will be provided with

Projected onto a matrix U₀The generated space, i.e. the final estimated CSI, is as follows:

in S5, updating the Prony coefficient by using the estimated CSI, specifically including the following steps:

s51, construction matrix

Each time an estimate is obtained

It is estimated by the top N_p-1 pieces of

Put into a matrix with the dimension of (M +1) M_b×N_pWhen m is 1, the matrix is as follows:

s52, pair

Performing SVD decomposition in the same process

Obtaining a matrix by SVD

And updating the Prony coefficient vector, wherein the calculation process of the updated Prony coefficient vector is as follows:

specifically, the CSI predicted at the next time in S6 is as follows:

s7 is specifically as follows:

making m equal to m +1, obtaining Nth step according to the above steps_pPredicted value of CSI at + m +1 times

Images