Local search based near optimal low complexity detection for large MIMO System

This paper presents a low complexity detection technique with near Maximum Likelihood (ML) performance for large multiple-input multiple-output (MIMO) systems. Large MIMO systems have gained popularity very soon because of high spectral efficiency and increased link reliability. ML based detection is known to give optimal result in terms of accuracy but due to extremely high computational complexity involved, detection time increases exponentially as the number of transmitter and receiver antennas increases. We propose an algorithm which gives near optimal performance along with much reduced computational complexity. Our results show that the proposed method outperforms linear detection technique named Zero Forcing (ZF) as well as heuristic based search algorithms named likelihood ascent search (LAS) and Reactive Tabu Search (RTS). Our algorithm finds the best solution restricted to a given Euclidean distance around initial solution. It searches all the neighbors of initial solution falling under dynamically calculated squared Euclidean distance based cost function value. As the number of antennas can vary in the range of tens to few thousands in large MIMO systems, this algorithm could be a substitute for ML based detection algorithm. We have considered Rayleigh fading channel for our simulations and assumed that perfect channel state information at the receiver (CSIR) is available.