Research Of Multi-User Detector Based On Ant Colony Optimization Algorithms In DS-CDMA

Code Division Multiple Access (CDMA) has more advantages than other access technologies, so it has already became the core technology of the third generation wireless communication systems, but these systems are all interference-limited, the system performance and capacity is limited by the Mufti-Access Interference (MAI). Mufti-user Detector (MUD) is the key interference cancellation technology of the wideband CDMA communication system, the optimal MUD can overcome the MAI theoretically, but it can not be carried out in real time on the current technology condition due to its high complexity. In recent years, in the field of intelligent computation, there are several types of Swarm Intelligence algorithms that have emerged, such as Ant Colony Optimization Algorithms (ACO), Particle Swarm Optimization (PSO), Shoal Optimization and so on. There are several distinctive features in the Swarm Intelligence algorithms that are different from conventional optimization approaches, and they have already been used to solve many classical NP problems. This thesis is dedicated to the application of improved ant colony optimization (ACO) to solve the issue of MUD.Firstly, Analysis of the MUD problem in CDMA communication systems, coming out from the theory of computational complexity.Secondly, introduce the theory of ACO and application in solving NP problem in detail.Lastly, apply improved ACO to MUD problem considering its factual characteristic, Moreover , have proved the improved ACO's validity by emulation.The main contribution of this thesis can be summarized as follows:(1)On the basis of the model of MUD based on ACO in document [6], improved the ACO algorithm in order to realize low bit error rate(BER),pick up rate of convergence when having same complexity.(2) In order to solve the disadvantage of deficient initial pheromone and slow rate of convergence of ACO, we combine ACO with Hopfield neural network which owning fast convergence ability together. Let Hopfield neural network supply good initial pheromone for ACO, so that improves both complexity and BER.