论文部分内容阅读
基于最小互熵(minimumcrossentropy,MCE)的迭代多用户检测算法在高度互相关的系统(即非扩频系统)中可以渐近地获得单用户的性能,是一种最优检测算法。但标准的MCE算法的计算复杂度极高,因而使其难以实用。该文提出了一种实现MCE算法的快速算法,避免了标准MCE算法在计算度量函数时的重复运算,在用户间符号同步和异步两种情况下均可以将计算复杂度降低K倍(K为用户数),而无性能损失。对于异步系统,还可以此基础上,通过对度量函数作合理近似,使运算度进一步减少1/3,在加性高斯白噪声信道下的仿真结果表明,这种近似所产生的性能损失低于0.2dB。
Iterative multi-user detection algorithm based on minimum cross-entropy (MCE) is an optimal detection algorithm for asymptotic gain of single-user performance in a highly cross-correlation system (ie non-spread spectrum system). However, the computational complexity of the standard MCE algorithm is extremely high, making it difficult to use. In this paper, a fast algorithm for implementing MCE algorithm is proposed, which avoids the repeated computation when standard MCE algorithm is used to calculate the metric function. It can reduce the computational complexity by K times in both the case of symbol synchronization and asynchronization between users (K is Users), without loss of performance. For the asynchronous system, on the basis of this, through reasonable approximation of the metric function, the degree of operation can be further reduced by 1/3. The simulation results under additive white Gaussian noise channel show that the performance loss of this approximation is lower than 0.2dB.