Downlink Performance Analysis of Distributed Antenna Systems with Linear Precoding

Abstract

Distributed antenna system (DAS) has emerged as a promising architecture for future mobile communications. Compared with collocated antenna system (CAS), DAS has advantages in reducing transmission power and improving the cell coverage area. In this paper, the downlink performance of DAS with linear precoding is analyzed. The optimization problem of minimizing the total transmitted power while satisfying signal-to-interference-and-noise ratio (SINR) constraints is studied to optimize the beamforming for all user jointly, which can be solved by conic optimization method. However, the problem is non-convex and can not be solved efficiently. Suboptimal methods are proposed by optimizing power allocation after given the direction of precoding. In simulation, the performance of different linear precoding methods are analyzed for the downlink of DAS. Moreover, the performance of DAS and CAS are also compared with the same antenna configuration and direction of precoding. It is shown that compared with CAS, DAS achieves the same performance with much less transmission power. Introduction More attention was paid to the distributed antenna system (DAS) because of its potential to improve system coverage and reduce the energy consumption. In DAS, the remote antenna units (RAUs) are separately and remotely located in a cell in accordance with certain rules. The RAUs are connected with the baseband processing unit (BPU) via Fiber-optic backbone network or dedicated radio link [1]. By introducing the RAUs, the distance between the mobile terminals (MTs) and the antennas of the RAU is greatly reduced. Thus, the large path loss of the radio signals can be avoided. Furthermore, the RAUs are located in different locations, which can be used to form a distributed multiple-input multiple-output (MIMO) system to further improve the performance of radio links. DAS has been extensively studied since 1980s. [2] proposed the idea of distributed antenna system, for solving the “blind spot” and large-scale indoor wireless communications fading problem. [3] pointed out that the DAS can be used to improve the capacity and coverage of the cell. [4] analyzed the capacity of CDMA DAS. In [5], the authors compared the general distributed antenna system (GDAS) and conventional cellular system where the same conclusion as [3] was drawn. In the downlink of DAS, the request for better performance with less complexity lead researchers to optimize the transmitter. Linear precoding, as the key technology to reduce the interference and maximize the throughput, has been extensively studied. Linear precoding mainly consists of maximum ratio transmission (MRT) precoding, zero forcing precoding (ZF) and signal-to-leakage-and-noise ratio (SLNR) precoding [6] [7] [8]. In order to improve the performance, many researchers proposed the optimal precoding according different criterions. [12] proposed a solution to solve the multiuser downlink problem with individual SINR constraints in a multiple-input single-output (MISO) system. The algorithm is based on the duality between the uplink and the downlink, and solved the problem iteratively in the uplink before switching to the downlink. [13] proposed a solution to solve the multiuser downlink problem with SINR constraints in a MIMO system. The algorithm is also based on the duality between the uplink and the downlink. For a change, [9] utilized convex optimization method to solve the multiuser downlink problem 3rd International Conference on Machinery, Materials and Information Technology Applications (ICMMITA 2015) © 2015. The authors Published by Atlantis Press 786 with the transmitted power constraints in MISO system. In this paper, we introduce convex optimization method to solve the multiuser downlink problem with SINR constraints in DAS. In order to reduce complexity, suboptimal methods are proposed by optimizing power allocation after given the direction of precoding. The performance of DAS with optimal precoding method and suboptimal precoding methods are analyzed. In addition, the performance of DAS and CAS are also compared with the same antenna configuration and direction of precoding. Numerical results show that the performance of DAS is better that of CAS. The following notations are used in this paper. Standard lower case letters (e.g. a) denotes scalars, boldface lower case letters (e.g. a) denotes column vectors and boldface upper case letters (e.g. A) denotes matrices. IM denotes an identity matrix of size M and diag(.) denotes diagonal matrix. (.), (.) and (.) denote the transpose, the conjugate transpose and pseudo-inverse, respectively. E(.) denotes the mean, N(μ,σ) and CN(μ,σ) denote respectively Gaussian distribution and complex Gaussian distribution, which μ is mean and σ is variance.