研究MIMO系统检测算法理论及其实现方法的基础上，对已证明较优的算法进行结合和改进，提出了一种改进的K—Best检测算法及其实现方案，并通过仿真验证了方案的可行性。该算法采用预测技术和并行排序相结合的方法，降低了计算复杂度；采用并行流水线结构实现，节省了处理时间；并对方案在xilinx公司的Virtex_5系列n)GA中的资源使用情况进行了统计。研究表明，实现方案可以用于MIMO系统检测算法的硬件实现。-Basic research MIMO system detection algorithm theory and its implementation methods, have proven to be the optimum combination and improved algorithms, we propose a modified K-Best detection algorithm and its implementation, and verified by simulation feasible options sex. The algorithm uses predictive technology and the combination of parallel sorting method to reduce the computational complexity　parallel pipelined architecture to achieve, saving processing time　and program xilinx' s Virtex_5 series n) GA in the resource usage statistics . Studies have shown that the program can achieve detection algorithm for MIMO system hardware.