【摘要】 在多用户CDMA系统中,为了提高扩频通信的抗多址干扰(MAI)能力,改善系统容量.提出了方便实用的能产生最佳扩频序列的系统模型,该模型由两级多层前馈网络(2-MFNN)级联组成,前级网络用于重建混沌吸引子,后级网络对前级网络输出的混沌序列进行变换,所得序列具有良好的相关性.模型产生序列的方法满足MAI最小的最佳扩频序列的要求.MFNN的学习算法采用具有对复杂的多峰值函数进行全局优化能力的克隆选择算法,以提高网络的学习能力,最后,应用柏内庭Lyapunov指数对该网络产生的序列是否具有混沌性态进行检验,并对网络输出的序列性能通过对整个序列集合的相关函数的均方值取平均进行了评估.计算结果表明,由该网络产生的序列不但具有混沌性态而且具有优良的相关特性,可以将其作为良好的混沌扩频序列发生器,适合于大容量CDMA通信系统.更多还原

【Abstract】 In order to enhance ability of combating Multiple Access Interference (MAI) and improve system capacity in multi-user CDMA system,a new system model,which is compose of two Multiple Forward Networks(2-MFNN)in series,is proposed to achieve optimum spreading sequences minimizing MAI.The former reconstructs chaotic attractor to generate Chaos Sequences,and the latter further simulate a transform relation to make Chaos Sequences from the former have better cross-correlation and auto-correlation.The method meets the condition of generating optimum spreading sequences minimizing MAI.Due to the disadvantage of conventional BP(Back-Propagation)learning algorithm used to MFNN,Clonal selection algorithm(CLONAL),which has the competence of searching for globally optimal or near-optimal solutions for multi-peak value function,is used as learning algorithm of MFNN.Whether the sequences that generated by the system model have chaotic characteristic,which is tested by Lyapunov Number,and the performance of output sequences of 2-MFNN is evaluated by using of root-mean-square of correlation function of all sequences.Simulation results indicated that the sequences obtained by 2-MFNN not only possess chaotic characteristic but also has better correlation.It can be served as generator of good sequences for great capability CDMA system.更多还原