Nonasymptotic Analysis of Massive MIMO under Different Wireless Scenarios

Abstract

The data rate demand is increasing as per Moore's law over the past two decades. This surge in data demand is a key source of motivation for the researchers to build robust wireless network and smarter wireless devices like smartphones, tabs, PCs, laptops, etc. From 2G to 4G, throughput maximization was the main driving force for the researchers of information and communication technology (ICT) industry. On the other side, energy efficiency (EE) is an essential figure of merit for the next generation (5G) wireless technology. 5G technology also improvises data rates, latency, massive connectivity, and network reliability significantly. Massive multiple-input-multiple-output (MIMO) emerges as one of the vital technologies for next-generation wireless communication. 5G technology, a base station (BS) can have hundreds or even more antenna that can improvise the spectrum efficiency (SE) and energy efficiency (EE), significantly. It can achieve all merits of conventional MIMO to a much larger scale. Despite the advantage of massive MIMO, challenges like hardware mismatch (HM), antenna correlation, pilot contamination, improper resource allocation, etc., undermine the benefits.
The first part of this thesis considers a massive multi-user MIMO network, which utilizes time-division duplexing (TDD) scheme. HM is a severe concern in massive MIMO, which limits the potential of time-division multiplexing (TDD) scheme. A novel hardware calibration technique, which helps for generating the downlink (DL) channel matrix using the estimated uplink (UL) channel at a minimal computational cost, is proposed. HM causes amplitude and phase impairment in the received signal and makes the wireless channel non-reciprocal. Considering complex Gaussian hardware response at each antenna terminal of the base station (BS) or user terminal (UT) side, we derive the probability density function (PDF) of amplitude and phase mismatch are derived individually. The joint PDF of amplitude mismatch (AM) between each terminal of BS to the UT and vice-versa is also derived at a constant phase response. This joint PDF reduces the computational cost of the UT by processing signal at the BS end. The DL system performances are evaluated with the proposed algorithm under three different linear precoder like matched filter (MF), regularized zero-forcing (RZF), and zero-forcing (ZF).
In the second part of this thesis considered three different antenna correlation environments based on the placement of antennas. It also considers a very large but finite number of BS antenna, and spacing between adjacent antenna elements is half of the wavelength. Using three linear detectors (maximum ratio combiner (MRC), ZF, and minimum mean square error (MMSE)), UL data rate and power efficiency in antenna correlation regime is obtained and compared to the independent and identically distributed (IID) wireless channel. Considering the impact of adjacent antenna element misalignment, for different range of degree of misalignment is also validated through the numerical simulation.
In the third part of this thesis, a relay assisted cooperative network, where the base station (BS) and relay station (RS) have a very large but finite number of the antenna, is considered. An analytical expression for the UL rate in different channel conditions (perfect/imperfect) is derived, and the impact of a large number of BS and RS antenna over the UL rate is verified. Different cooperation protocol for improvising the ease of cooperative selection diversity have also been incorporated. On the other side, suitable linear precoder and decoder improvise the end-to-end SNR and end-user capacity in a dual-hop wireless network. When MIMO size increases asymptotically; the random matrix theory (RMT) helps to obtain the closed-form solution. In such a network, when statistics of the channel matrix and precoding matrix are known then, the SNR and achievable rate is easily obtained through the RMT.