Multi-user Detection For MC-CDMA System Based On An Advanced Clonal Selected Particle Swarm Optimization

As a multi-carrier code division multiple access technology, MC-CDMA combines the advantages of CDMA and OFDM technologies. It is characteristic of such merits as high spectral efficiency and good inter-symbol interference (ISI) suppression. Besides, its modulation/demodulation can be implemented by the fast Fourier Transform (FFT). Thus MC-CDMA has become one of the most popular research topics for the next generation wireless communication system.Same as in CDMA, in MC-CDMA, the multi-users share the same frequency band, they are identified only by their spreading codes. In fact, for reasons such as multipath fading, it is difficult to maintain the orthogonal among the spreading sequences, and this will produce multiple access interference (MAI), which seriously affects the performance of the system. Conventional single user detector views MAI as Gaussian white noise, and can not eliminate the MAI. Thus multi-user detective technique is needed to remove MAI and further to improve the performance of MC-CDMA.Multi-user detection (MUD) makes full use of effective information of interference signals. It can effectively eliminate the MAI and increase the system capacity, and does not require high system power control accuracy. Optimum multiuser detection has high computational complexity, is difficult to apply in practice. So a variety of sub-optimal multi-user detection has become the research focus.In this paper, the predigested MC-CDMA system model is established, including the model of transmitter, wireless channel and receiver. Then come to multi-user MC-CDMA system model. Base of it, several commonly used linear and nonlinear suboptimal multiuser detectors are studied, and their advantages and disadvantages are analyzed respectively.Since multi-user detection can also be viewed as combinatorial optimization problem, in this paper, two intelligent algorithms used for combinatorial optimization are introduced to multi-user detection for MC-CDMA system. Then the two algorithms are compared with traditional detector and decorrelation detector. The simulation results show that the traditional detector has the worst performance because it can not eliminate MAI. The decorrelation detector enlarges noise at the same time of removes MAI, and has poor performance in low SNR. Colonal selection (CS) algorithm focuses on local optimization, but its global optimization is worse, the performance improvement is not enough. The particle swarm optimization (PSO) has good global and local search capabilities, and its performance is better than other algorithms. But PSO has weak local search problem with the addition in the number of iterations, which limits the performance of the algorithm. To solve this problem, this paper proposed a new algorithm based on clonal selection and particle swarm optimization. By adding cloning and mutating operations every time of PSO iteration, the new algorithm strengthens the particles searching in the local near the best particle, and the introduction of decreasing inertia weight strategy to further strengthen accuracy of the searching. The simulation results show the improved algorithm is better than the original algorithm against the multiple access interference, accommodates more users and improves the convergence rate. Besides, the new algorithm's complexity does not increase a lot.