Resource Allocation in Cell-Free Massive MIMO System under Mixed LoS/NLoS Channels

Abstract

Cell-Free (CF) massive multiple input and multiple output (mMIMO) has emerged as a promising technology for the next generation wireless communication systems. It comprises a large number of access points (APs) serving a relatively small number of user equipments (UEs) distributed over a wide area. Since the UEs are now proximal to the APs, CF mMIMO can provide a high likelihood of coverage, and a line-of-sight (LoS) connectivity may also exist between a UE and one or more connected APs nearby. Most of the state-of-the-art works on the CF mMIMO either consider Rayleigh fading or Rician fading channels. However, due to vagueness in the physical distribution of various blockages (e.g., buildings in urban areas, trees, hill rocks, etc., in rural areas), it is impossible to correctly assess the presence of a LoS link between the APs and UEs. As a result, characterizing the performance of CF mMIMO system with probabilistic LoS components becomes necessary. Since, the CF mMIMO system aim to provide uniform coverage to all the users connected in the network, however, under mixed LoS/NLoS channel conditions CF mMIMO system suffers some following problems: the capture of the system by the UEs with LoS paths, and hence the service-outage of the NLoS users at moderate AP densities even under accurate channel state information (CSI); limited feedback capacity between the APs and the central processing unit (CPU); the rapid degradation of energy transfer efficiency over large distances due to significant path loss; and the problem of pilot contamination with limited pilot lengths. Therefore, in this thesis, we focus on performance optimization of the CF mMIMO system under mixed LoS/NLoS channel conditions considering all the above-mentioned problems. The first problem in this thesis address the issue of the service-outage of the NLoS users at moderate AP densities even under accurate CSI via the use of appropriate power control in both the uplink and downlink of a CF mMIMO system under mixed LoS/NLoS channels. We find that simplistic power control techniques, such as channel inversion-based power control, perform sub-optimally as compared to exhaustive search-based max–min power control. As a consequence, we propose a particle swarm algorithm (PSA) based power control algorithm to optimize the performance of the system under study. We then use numerical simulations to evaluate the performance of the proposed PSA based solution and show that it results in a significant improvement in the fairness of the underlying system while incurring a lower computational complexity. Following this, in the second problem, we analyse the uplink performance of a CF mMIMO system under probabilistic LoS/NLoS channels with limited feedback capacity between the APs and the CPU. We note that due to the limited feedback capacity, the information shared by the APs needs to be quantized before transmission, and use this fact to derive the rates achievable by the different users for different numbers of bits being used by the different APs. We then discuss the max-min feedback bit allocation strategy to develop the analogous optimization problem, and observe its computational in-feasibility. Following this, we develop a user-centric feedback bit allocation strategy that exploits the significantly different channel qualities offered by LoS and NLoS channels. Finally, we validate the performance of the proposed bit allocation scheme via extensive simulation experiments and find that it leads to a performance almost as good as the unquantized case. Next, we consider the problem of simultaneous wireless information and power transfer (SWIPT) in a CF mMIMO system under a mixed LoS/NLoS channel. In this context, we derive expressions for the achievable downlink rate and the harvested energy under an energy split scenario and show that the problem of joint downlink rate and harvested energy maximization is non-convex and combinatorially complex. Following this, we derive a particle swarm algorithm (PSA) based optimization algorithm for maximizing the joint downlink achievable rate and harvested energy. Finally, we demonstrate the performance of this algorithm and the effect of various system parameters on the achievable rate and the average harvested energy via detailed numerical simulations. Finally, we examine the pilot contamination problem in the CF mMIMO system, which arises due to a large number of UEs being served simultaneously by multiple APs under constraints on pilot length. Reduction of pilot contamination becomes challenging in the context of CF mMIMO system mainly because of two reasons:(i) all the APs serving a subset of UEs need to obtain good local channel estimates; and (ii) there is no inherent AP-UE clustering that can facilitate local orthogonality among the clustered UEs. Thereby, we cast the joint problem of pilot length and contamination minimization for a CF mMIMO system with a more practical mixed LoS/NLoS channel model, and analyse it using tools from stochastic geometry. We first characterize the probability of a given UE having an LoS link as a function of the AP density and use this to find the number of UEs whose accurate CSI is required. Following this, we recast a pilot allocation strategy based on graph coloring to our system model and analyze the values taken by various parameters using tools from stochastic geometry. We validate our results using Monte-Carlo simulations and evaluate the performance of the graph coloring-based pilot allocation algorithm.