A near-ML Performance Adaptive Dijkstra Algorithm for Large Scale MIMO Detection

Hellassa, Wassim; Boukari, Karima

Synthèse
Volume 26, Numéro 1, Pages 52-62
2020-04-20

A Near-ml Performance Adaptive Dijkstra Algorithm For Large Scale Mimo Detection

Authors : Hellassa Wassim . Boukari Karima .

Abstract

Employing Maximum Likelihood (ML) algorithm for signal detection in large scale Multiple-Input- Multiple Output (MIMO) system with high modulation order is a computationally expensive approach. In this paper an adaptive search algorithm is proposed for ML detection based MIMO receiver that can be classified as a derivative of Dijkstra’s Search (DS) algorithm based best first search algorithm; hence naming Adaptive Dijkstra’s Search (ADS) algorithm. The proposed ADS exploits the resources available in the search procedure to reduce the required number of nodes to be visited in the tree. Results are obtained depending on a tunable parameter, which is defined to control the number of the best possible candidate nodes. Unlike the conventional DS, the ADS algorithm results in signal detection with low computation complexity and quasi-optimal performance for systems under low and medium SNR regimes. Simulation results demonstrate 25% computational complexity reduction, compared to the conventional DS. For Symbol Error Rate (SER) of 10-2, such computation complexity reduction is also a trade-off with 2 dB SNR degradation, while attaining the same SER with conventional DS. The reduction of the computation complexity with the proposed ADS is non-linearly proportional to the dimension of MIMO combination as well as the modulation order.