On Finding Sub-optimum Signature Matrices for Overloaded CDMA Systems

TL;DR

This paper proposes methods to design optimal signature matrices for overloaded CDMA systems using genetic algorithms, particle swarm optimization, and distance criteria, including suboptimal large matrices for computational efficiency.

Contribution

It introduces a combined evaluation approach using capacity, BER, and distance measures, and proposes suboptimal large matrices derived from smaller ones for practical implementation.

Findings

Genetic Algorithm and PSO effectively optimize signature matrices.

Distance criteria provide an additional measure for optimality.

Suboptimal large matrices reduce computational complexity.

Abstract

The objective of this paper is to design optimal signature matrices for binary inputs. For the determination of such optimal codes, we need certain measures as objective functions. The sum-channel capacity and Bit Error Rate (BER) measures are typical methods for the evaluation of signature matrices. In this paper, in addition to these measures, we use distance criteria to evaluate the optimality of signature matrices. The Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) are used to search the optimum signature matrices based on these three measures (Sum channel capacity, BER and Distance). Since the GA and PSO algorithms become computationally expensive for large signature matrices, we propose suboptimal large signature matrices that can be derived from small suboptimal matrices.