CN114301508A

CN114301508A - Dual-time scale optimization method in reconfigurable intelligent surface-assisted MIMO transmission based on over-time CSI

Info

Publication number: CN114301508A
Application number: CN202111370756.1A
Authority: CN
Inventors: 吕铁军; 曹亚帅
Original assignee: Beijing University of Posts and Telecommunications
Current assignee: Beijing University of Posts and Telecommunications
Priority date: 2021-11-18
Filing date: 2021-11-18
Publication date: 2022-04-08

Abstract

The invention provides a double-time scale beam forming method for a reconfigurable intelligent surface (IRS) auxiliary MIMO system, which is used for configuring IRS relatively infrequently on the basis of statistical Channel State Information (CSI) on a large time scale and frequently updating power distribution of a Base Station (BS) on the basis of outdated CSI on a small scale. The specific method is that a transmission structure for overcoming outdated CSI is designed on a small scale, a particle swarm optimization (mbs-PSO) algorithm based on small-batch sampling is provided on a large scale by utilizing a closed solution for obtaining optimal small-scale power distribution, and large-scale IRS configuration is predictively optimized under the condition of reducing channel sampling. The invention adopts the IRS-assisted MIMO double-time scale transmission scheme, can effectively resist the interference between signal streams caused by outdated CSI, and the algorithm does not need excessive cascade channel estimation and IRS configuration overhead in a fast time-varying channel.

Description

Dual-time scale optimization method in reconfigurable intelligent surface-assisted MIMO transmission based on over-time CSI

Technical Field

The invention relates to a double-time scale design method in an IRS (intelligent resilient system) assisted MIMO (multiple input multiple output) communication system. Specifically, the scheme considers the outdated effect faced by actual channel estimation, designs an improved transceiving structure aiming at the outdated CSI, jointly optimizes long-term IRS configuration and short-term power allocation by taking the maximized average reachable rate as a target, and belongs to the technical field of wireless communication.

Background

In order to meet the demand for high data rates in mobile communication, MIMO technology is proposed in 5G to support high spectral efficiency. IRS has been identified as an attractive complementary technique to MIMO systems to improve spectral and energy efficiency. Since IRS is mainly composed of low-cost passive metamaterial elements, they can be densely deployed with existing MIMO systems, greatly improving spectral efficiency in a cost-effective manner. It can be demonstrated that the transmit power required by the BS can be reduced approximately twice as the number of IRS elements increases, and the achievable energy efficiency of the IRS is higher compared to existing relay systems.

Perfect instantaneous CSI (I-CSI) is crucial to obtain the expected benefits of IRS. Since the CSI estimation technique is not perfect in practical implementation, and the CSI obtained when the BS applies the signal processing technique is usually outdated, it is not easy to introduce the IRS into the MIMO system. The concatenated channel estimation associated with IRS has been widely studied and made significant progress, however, due to non-negligible feedback delay, it is difficult for the BS to obtain perfect I-CSI, especially in Time Division Duplex (TDD) systems. The BS applies sophisticated precoding techniques, such as Singular Value Decomposition (SVD), based on outdated CSI, which can result in severe inter-signal stream interference.

On the other hand, non-negligible overhead in IRS configuration can severely reduce spectrum utilization. Existing IRS configuration algorithms are mostly developed under perfect I-CSI. In practical cases, CSI acquisition is a computationally intensive task and takes into account the additional feedback delay of I-CSI, so it is of great significance to solve the problems of outdated CSI effect and IRS feedback overhead in IRS-assisted MIMO transmission design.

Disclosure of Invention

In view of this, the present invention provides an effective method for dual-time scale beamforming design in an IRS-assisted MIMO system in consideration of outdated CSI, so as to reduce IRS configuration overhead and suppress outdated CSI effect, where the method mainly includes large-scale IRS reflection coefficient configuration and small-scale power allocation. We consider a MIMO transmission system in which both the BS and the user are equipped with MIMO antennas and the quality of communication between the BS and the user is improved by an IRS-supplemented link. The BS adopts the classical SVD precoding, and further adopts a Zero Forcing (ZF) rule to resist the influence of the over-time CSI, so that the interference between signal streams is inhibited. On the basis, the Average Achievable Rate (AAR) is maximized through the dual-time scale beamforming design.

The method comprises the following three operation steps:

(1) and (3) designing a transceiving structure based on outdated CSI: BS is based on outdated effective channels, i.e.

In a time-varying channel

Performing SVD precoding

BS performs transmit beamforming

Wherein Λ ═ diag ([ P ])₁,…,P_M]) Representing the power allocation of the BS. The user performs receive beamforming W based on the SVD results_r. In order to resist the effect of the over-time CSI, the BS sends a pilot frequency sequence before precoding, and the user equipment estimates a virtual channel according to the received pilot frequency

Performing ZF detection of transmission signals based on virtual channels, i.e.

In this system, the signal to interference and noise ratio (SINR) of the mth data stream can be characterized as

(2) Small time scale power allocation: based on the dual-time scale beamforming strategy, the IRS only needs to be configured once with the unchanged statistical CSI on each time frame. The beamforming of the BS and the users is frequently updated on the basis of outdated CSI on each mini-slot within the remaining time of the frame.

(21) In this system, based on dual time scale beamforming, the AAR maximization of the system can be modeled as

In the above optimization problem, the first constraint is used to limit the sum of the transmission powers of the M antennas of the BS to P_totThe second constraint ensures that the configured phase shift coefficient of the IRS is unity modulo. Given the IRS configuration Θ, diag (θ), at the tth slot, according to the known obsolescence

The AAR maximization problem can be modeled as a small time scale power allocation problem as follows:

(22) the constructed small time scale power allocation problem can be shown to be convex with respect to p, and to derive its closed form solution, a function is first defined

Introduction of a relaxation variable mu convertible by Lagrangian

The closed form solution of power can be obtained by derivation:

(3) large time scale IRS configuration: under the large scale IRS configuration scheme, our goal is to maximize the AAR of the system. By the improved particle swarm optimization method, the optimal IRS configuration coefficient can be obtained.

(31) Since the objective function of the AAR maximization problem contains the desired operator, the IRS configuration at each time frame is a random optimization problem. To solve this problem, we first convert it to a deterministic optimization problem:

specifically, we are at L_BThe achievable rate averaging over a (large) number of random channel samples is done to approximate the expectation in the target.

(32) The constructed deterministic optimization problem is non-convex, with the goal of modeling it as a Particle Swarm Optimization (PSO) problem to find the optimal solution Θ, diag (θ). Firstly, a group of particles are randomly generated under a constant modulus constraint, each particle represents a potential theta solution, the BS acquires all outdated CSI samples in a time frame, and an effective channel is calculated according to each particle theta

In correspondence with

And (3) applying the conclusion of the step (2) to obtain an optimal power distribution strategy so as to obtain the maximum reachable rate corresponding to each particle. The AAR of each particle over all outdated CSI samples is taken as the fitness function of the PSO, which is evaluated in each iteration based on the fitness function of each particle to determine whether the current position implies a good solution. Each particle can record its own found optimal position, and the global optimal position is selected from the optimal positions of all particles.

(33) In fitness evaluation, an mbs-PSO algorithm is developed, aiming at greatly reducing the number of channel samples for evaluating a fitness function, thereby reducing complexity while maintaining the fitness functionCommunication performance is verified. First, a small batch recursive sampling proxy function is introduced as a fitness function. All L_BA random number of channel samples is divided into N_BEach batch using L_mb＝L_B/N_BAnd (4) sampling. In the ith iteration, the proxy function is set to

(34) Order to

And

respectively representing the optimal positions of p particles and the global optimal positions of all the particles in i iterations. At i iterations, p particles are dependent on their velocity

Updating the position:

if the fitness value of the last optimal position is lower than the fitness value of the new position, the optimal position of the particle is updated using the new position. Otherwise, the optimal position of the particle is unchanged. After a round of fitness value evaluation, the global optimal positions of all the particles are updated.

(35) And (4) repeating the steps (33) to (34) until the fitness function value converges.

In the invention, the problems of outdated CSI effect and frequent IRS feedback overhead are considered, and a dual-time-scale beam forming design strategy is provided for an IRS-assisted MIMO communication system. The method has the advantages that under the condition of ensuring the transmission rate, the data transmission rate is improved by utilizing the infrequent IRS reflection coefficient configuration and the power allocation of the short time slot, so that the interference among signal streams caused by outdated CSI is overcome, and the feedback overhead in communication is reduced. The proposed AAR maximization problem is a complex random non-convex problem, and it is difficult to obtain a globally optimal solution. To this end, we transform the original problem to a convex power allocation problem on a small scale and derive its closed-form solution. In the large-scale IRS coefficient optimization, an mbs-PSO algorithm is adopted to obtain an optimal IRS configuration coefficient, so that the high calculation complexity based on large-scale data in the traditional PSO algorithm is overcome. The invention is an effective transmission scheme design which can improve the average transmission rate of an IRS auxiliary MIMO system, reduce the overhead and inhibit the outdated CSI effect.

Drawings

Fig. 1 is an application scenario of the present invention: IRS assisted MIMO transmission system model diagrams based on outdated CSI.

Fig. 2 is a flow chart of the design based on dual time scale beamforming in the present invention.

FIG. 3 is a simulation diagram of the convergence behavior of the mbs-PSO algorithm in the embodiment of the present invention.

Fig. 4 is an AAR simulation diagram of an IRS-assisted MIMO system under different CSI feedback delays in the embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the accompanying drawings.

Referring to fig. 1, the application scenario of the present invention is: the deployed IRS enhances communication between the BS equipped with multiple antennas and the user equipment. There is line-of-sight (LoS) communication between the IRS and the BS. The IRS to user equipment, and BS to user equipment channels are subject to lie fading, where the LoS component is known statistical CSI and remains unchanged over a time frame; unknown non-line of sight (NLoS) components are subject to a first order autoregressive process with temporal correlation. The BS may implement acquisition of all CSI by using channel reciprocity through a Time Division Duplex (TDD) protocol, but there is a CSI feedback delay. And jointly optimizing the phase shift configuration of the IRS on a large time scale and the BS power distribution on a small time scale by the BS based on the SVD decomposition and ZF correction receiving criteria according to the I-CSI and the statistical CSI.

Our goal is to reduce IRS feedback overhead and suppress inter-signal stream interference due to CSI feedback delay, maximizing the average sum rate of IRS-assisted MIMO transmission. First, the problem of construction is a complex random non-convex problem that is difficult to solve directly. In a dual time scale beamforming scheme, power allocation is considered on the large time scale and the small time scale, respectively, in phase shift optimization for IRS. Aiming at the power distribution problem on a small time scale, a power closed-form solution of each iteration is obtained according to Lagrange transformation. Aiming at the phase shift optimization of the large-time-scale IRS, the problem is converted into a deterministic optimization problem, and because the problem is non-convex and the function form is complex, an mbs-PSO algorithm is designed, and the optimal IRS phase shift coefficient is effectively obtained by iterative solution with the advantages of few parameters, simple calculation operation and fast convergence.

In order to demonstrate the utility of the present invention, the applicant conducted a number of simulation experiments. The transmission system model in the test system is an application scenario shown in fig. 1, and the result of the simulation test is shown in fig. 3. In the baseline scheme of fig. 3, we show the AAR performance results of the proposed algorithm and the comparative algorithm under CSI feedback delay.

The above description is only a preferred embodiment of the present invention, and should not be taken as limiting the invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims

1. The invention provides a double-time scale beam forming method for overcoming outdated CSI effect in an IRS auxiliary MIMO transmission system; for the following scenarios: the communication between the BS and the user equipment is enhanced by the assistance of the IRS; the BS can acquire outdated CSI from the user to the BS and outdated CSI from the user to the IRS, and the BS optimizes the IRS phase shift configuration on a large time scale according to the statistical CSI and the outdated CSI.

(1.1) receiving and transmitting structure design based on outdated CSI: the BS performs SVD on the time-varying channel based on the outdated CSI so as to guide the BS to perform transmission beam forming and power distribution; in order to resist the effect of the over-time CSI, the BS transmits a pilot sequence before beamforming to support the user equipment to estimate a virtual channel according to a received pilot so as to perform ZF detection on a transmission signal.

(1.2) small time scale power allocation: based on a double-time scale beam forming strategy, only one time of IRS is needed to be configured on each time frame by using invariable statistical CSI; the beamforming of the BS and the users is frequently updated on the basis of outdated CSI on each mini-slot within the remaining time of the frame.

(1.3) large time scale IRS configuration: under large scale IRS configurations, our goal is to maximize the AAR of the system; by the improved particle swarm optimization method, the optimal IRS configuration coefficient can be obtained.

2. The method of claim 1, wherein in step (1.2), the small-time-scale power allocation scheme comprises the following operations:

and (2.1) constructing the AAR maximization problem of the system based on double-time scale beam forming. Given IRS configuration, on the t-th time slot, converting the AAR maximization problem into a small-time scale power distribution problem;

and (2.2) introducing a relaxation variable by utilizing a Lagrange multiplier method to obtain a power closed-form solution corresponding to a specific IRS configuration.

3. The method of claim 1, wherein step (3) further comprises the following operations:

(3.1) to solve the AAR maximization problem for large scale IRS configurations, we transform the stochastic optimization problem into a deterministic optimization problem;

(3.2) to solve this non-convex deterministic optimization problem, we model it as a PSO problem to find the optimal solution for IRS configuration. Firstly, based on all outdated CSI samples acquired by a BS in a time frame, calculating corresponding AAR according to each particle;

(3.3) introducing a small-batch recursive sampling proxy function as a fitness function according to the provided mbs-PSO algorithm in fitness evaluation, and updating the fitness function in each iteration;

(3.4) updating the position and the speed of the particle in each iteration, and updating the optimal position of the particle record and the global optimal positions of all the particles after one iteration;

(3.5) repeating the steps (3.3) - (3.4) until the fitness function value converges.