Research On Adaptive Equalization Based On Swarm Intelligence Algorithm For Asynchronous Cooperative Communication

In some application scenarios of asynchronous cooperative communication,such as military operation and emergency rescue,the deployment of base station or central control node for collaborative scheduling may be impractical and there is a strict limit of system overhead.As a result,channel estimation,relay selection and some other technologies are unavailable.In order to ensure that the destination node is able to receive the signal,the signals from the source node will be retransmitted by every relay under the condition that AF protocol is adopted.For the correct decision of the received signal,adaptive equalization technique is carried out at the receiver to remove inter-symbol interference,which is caused by the retransmissions of the asynchronous relays.As one of the most used adaptive equalization algorithm,LMS algorithm has an obvious disadvantage on convergence speed.In the above background,adaptive equalization for asynchronous cooperative communication is investigated in this thesis,and LMS algorithm is improved by various swarm intelligence algorithms.The main contents are as follows:(1)A modified LMS algorithm based on PSO-sphere-searching is proposed,where the search ability of PSO algorithm is utilized to accelerate the convergence of LMS algorithm,and meanwhile,the local convergence of PSO algorithm is overcomed.PSO algorithm is used to search the optimal solution in the neighborhood of the error,which can be obtained in each iteration during the training stage of the equalizer.The output of PSO algorithm is utilized to update the coefficients according the formula of LMS algorithm.Simulation results of time-domain and frequency-domain equalization show that the proposed algorithm outperforms conventional LMS algorithm in convergence and bit-error-rate(BER)performance.(2)A PSO-LMS joint iterative algorithm with a lower calculation amount is proposed according to the feature of PSO algorithm that the motion of the particles refers to the personal best and the global best at the same time,where only one particle is enough for the searching process.In the hybrid algorithm,the output of the equalizer is set as the personal best of PSO algorithm,while the desired signal is set as the global best.During the iterative process,the personal best converges to the global best,that is,the output of the equalizer converges to the desired signal.Simulation results indicate that the devised hybrid scheme outperforms conventional LMS algorithm in time-domain and frequency-domain equalization.(3)By means of the thought of PSO-sphere-searching mentioned above,a modified LMS algorithm based on DE-sphere-searching is proposed,where the search ability of DE algorithm is utilized to improve the convergence performance of LMS algorithm.Simulation results show that the proposed scheme accelerates the convergence of LMS algorithm and a lower steady state error is got.It achieves better BER performance than conventional LMS algorithm at high signal-noise-ratio(SNR).(4)On the basis of the above paragraphs,a modified LMS algorithm based on CSO-spheresearching is proposed,where the search ability of CSO algorithm is utilized to accelerate the convergence of LMS algorithm.Simulation results indicate that the devised algorithm has an obvious improvement in convergence and BER performance compared to conventional LMS algorithm.