Maximum Likelihood Detection in Communication Using Metaheuristic Search Methods

Maximum-Likelihood (ML) detection problem in communication is known to be NP(Non-deterministic Polynomial time)-hard. The computational complexity of solving ML detection is exponential in the size of the problem with exhaustive search that provides optimal solution. Several suboptimum algorithms have been proposed in the literature that provide reliable performance with reduced complexity. But still there is a large gap between the performance of the sub-optimal detectors and that of the optimal detector. Motivating by this, the main objective of this research is to achieve near-optimal performance of detector while maintaining computational efficiency. In this thesis, we look at several metaheuristic optimization methods to get approximate optimal solution. We improve the performance of (1 + λ) Evolutionary Strategy (ES) based multiuser detector for synchronous Direct Sequence Code Division Multiple Access (DS-CDMA) system by applying hybrid (1 + λ) ES algorithm. We also applied this hybrid method for ML detection in Multicarrier CDMA (MCCDMA) and Multiple Input Multiple Output (MIMO) systems. We proposed Simulated Annealing (SA) algorithm for ML detection and applied to these systems. Based on a new type Evolutionary Computation (EC) algorithms named Estimation of Distribution Algorithms (EDAs), we developed a new detection scheme. We applied an EDA approach named Population-based Incremental Learning (PBIL) algorithm and also modified. Simulation results are presented to demonstrate the efficacy of the proposed algorithms over the existing detectors.