Cellular Automata-based Differential Evolution Algorithm And Its Applications In Communication Systems

Differential evolution(DE) algorithm is a parallel population iterative optimization algorithm and has been widely used in solving various types of static optimization problem. However, due to the premature convergence and search stagnation, which limit the optimization ability and application scope of DE, especially applied to solve dynamic optimization problems. It urgently needs to be studied and improved for DE. In the field of communication systems’ processing, there are many optimization application problems, such as pilot-based fast-fading channel estimation in which various parameters about fast-fading channel cannot be obtained promptly, the existing algorithms have a lot of limitations. The resource allocation of multi-relay cooperative communication system often has to dipose approximately for high complexity of theoretical computation. So the intelligent heuristic algorithms can provide a new effective way to tackle those the analytical class problems. To overcome the premature convergence and search stagnation of DE algorithm, a novel cellular DE algorithm has been proposed against. What is more, the cellular DE algorithm has been used to resolve the channel estimation algorithm of the wireless receiver and the optimization problem of the resource allocation algorithm of the cooperative communication system. The main contents and innovation points are as follows:(1) An improved DE algorithm based on cellular automata(CA) has been proposed. For the premature convergence shortcoming of DE algorithm, we have studied the evolution of CA with diffusion characteristics and parallel computing, which can be used to maintain the diversity of evolutionary population and to avoid prematurity. The neighbor structure and evolutionary rules of CA is used in the proposed DE algorithm as an effective means of tuning the pressure of the control parameters’ selection and thereby the exploration and exploitation tradeoff. The proposed DE algorithm employs the opposition-based learning(OBL) for population initialization and the orthogonal crossover for selecting winners from multiple offsprings into the next-generation evolution. The aim of those strategies is to improve the global convergence speed, maintain the diversity of population and avoid premature convergence.(2) A new chaotic local search(CLS) based cellular DE algorithm(cc DE) has been proposed. In order to defeat the search stagnation shortcoming of DE algorithm, we study the CLS-based update mechanism of evolutionary individual in DE algorithm. By using the ergodic of the chaotic sequence, the evolutionary individual should be reinitialize when it is not updated within the setting number of iterations and falling to a local optimum, which can help it out of local optimal solution. The way can greatly improve the chances of obtaining the global optimal solution and find the global optimal solution as soon as possible. The proposed DE algorithm can avoid falling into local optimal solution and search stagnation by using the ergodic and stochastic characteristics of CLS.(3) A novel cc DE-based Maximum Likelihood(ML) channel estimation algorithm has been proposed to get the effective channel length in real time. For high complexity and low transmission efficiency shortcoming of the Minimum Mean Square Error(MMSE) channel estimation algorithm based on the pilot, we first study the classic Linear Minimum Mean Square Error(LMMSE) channel estimation algorithm. Then, by the judgment of the signal-subspace dimension, the compromise between computational complexity and estimation performance is obtained. The computational complexity may be reduced effectively by reducing the rank of the matrix without degrading estimation performance substantially. Meanwhile the error rate performance of data detection may be improved by a secondary filter. In a addition, Least Squares Support Vector Machine(LS-SVM) and Extreme Learning Machine(ELM) prediction mechanism-based channel estimation algorithm are suggested to save the number of pilots and improve transmission efficiency. Finally, a cc DE-based ML channel estimation algorithm has been proposed for the effective channel length obtaining.(4) A cc DE-based resource allocation scheme of the Orthogonal Frequency Division Multiple Access(OFDMA) cooperative communication system has been proposed. The resource allocation of OFDMA relay cooperative communication system is a complex joint optimization problem, which is difficult to locate the optimal solution. The existing distribution methods generally adopt a simplified approximation process and then seek suboptimal solution by converting it to a convex optimization problem. It is a new idea by utilizing the intelligent heuristic algorithm to solve the analytical class problem. The cc DE-based power allocation algorithm has been proposed and has better performance compared with other classical distribution methods. By introducing the maximization criterion, the cc DE-based resource allocation approach of OFDMA cooperative communication system has been put forward while taking into account both multi-user’s fairness and overall system performance.