A Primer on MIMO Detection Algorithms for 5G Communication Network

Abstract

In the recent past, demand for large use of mobile data has increased tremendously due to the proliferation of hand held devices which allows millions of people access to video streaming, VOIP and other internet related usage including machine to machine (M2M) communication. One of the anticipated attribute of the fifth generation (5G) network is its ability to meet this humongous data rate requirement in the order of 10s Gbps. A particular promising technology that can provide this desired performance if used in the 5G network is the massive multiple-input, multiple-output otherwise called the Massive MIMO. The use of massive MIMO in 5G cellular network where data rate of the order of 100x that of the current state of the art LTE-A is expected and high spectral efficiency with very low latency and low energy consumption, present a challenge in symbol/signal detection and parameter estimation as a result of the high dimension of the antenna elements required. One of the major bottlenecks in achieving the benefits of such massive MIMO systems is the problem of achieving detectors with realistic low complexity for such huge systems. We therefore review various MIMO detection algorithms aiming for low computational complexity with high performance and that scales well with increase in transmit antennas suitable for massive MIMO systems. We evaluate detection algorithms for small and medium dimension MIMO as well as a combination of some of them in order to achieve our above objectives. The review shows no single one detector can be said to be ideal for massive MIMO and that the low complexity with optimal performance detector suitable for 5G massive MIMO system is still an open research issue. A comprehensive review of such detection algorithms for massive MIMO was not presented in the literature which was achieved in this work.

A Review to Massive MIMO Detection Algorithms: Theory and Implementation

“…First, the ZF 1 The nomenclature section provides definitions. 2 It is widely believed that it does not exist any solution to solve NP-hard problems in polynomial time.…”

“…Hence, low computational complexity algorithm achieving near-optimal performance is required; many existing detection algorithms like zero forcing (ZF), minimum mean-square error (MMSE), and successive interference cancelation (SIC) are used to deal with massive MIMO detection. In [1,2], the authors presented surveys on various MIMO and massive MIMO detection techniques from algorithmic viewpoints. Although many classical massive MIMO detectors have been proposed in the literature, herein, new recent algorithms based on the application of machine learning, geometrical techniques, and bioinspired methods are presented and discussed.…”

5G Wireless Communication Network Architecture and Its Key Enabling Technologies

“…In this scenario as dimensions become large, matrix operations like inversion becomes less operationally complex and simple series expansion techniques can be used to calculate it [17]. This is what makes linear algorithms like ZF and MMSE where matrix inversion operation is used to be near optimal in performance [30], a situation that is almost impossible without mMIMO favorable propagation condition [31]. In 5G HetNet, wireless backhaul is preferred to wire backhaul and mMIMO readily provide this service because of its ease of deployment compared to wired backhaul.…”

5G Small Cell Backhaul: A Solution Based on GSM-Aided Hybrid Beamforming

“…With SM, inter-antenna synchronization (IAS) and inter-channel interference (ICI) are entirely done away with and the complexity of the receiver in decoding the received symbols, in terms of mathematical computations performed, increases in a linear dimension as the constellation size increases and the number of transmitting antennas increases. Again, in SM the number of receive antennas is not restricted as we have in V-BLAST scheme, which requires > before the minimum mean square error (MMSE) detector can be operational [23]. Finally, SM has a higher spectral efficiency than single-input single-output (SISO) and orthogonal STC systems due to the employment of antenna indices as an additional source of information, [24].…”

“…At the receiver, for us to detect the symbols transmitted, the receiver using CSI must know a priori the CIRs of all the Tx-Rx wireless links and in general, a total of CIR need to be estimated and using the maximum likelihood algorithm [23] rule is given in equation 2…”