论文部分内容阅读
MIMO系统中的最大似然检测可以表示为一个整数二次规划问题.将基于变量二分的深度优先分支定界法引入到MIMO检测中,利用这种结构,可以对更多的节点进行剪枝.在分支定界法的每一层,采用有效集法来求解对偶子问题.为进一步降低复杂度,在有效集法的迭代过程中采用Cholesky分解更新求解一个线性系统问题.通过松弛剪枝条件,给出了准分支定界法,实现了性能和复杂度的较好折衷.数值仿真表明,基于分支定界法的MIMO检测算法复杂度很低,尤其在低信噪比和高阶调制时,其优越性尤为明显.
The maximum likelihood detection in MIMO system can be expressed as an integer quadratic programming problem. The depth-first branch-and-bound method based on variable dichotomy is introduced into MIMO detection. With this structure, more nodes can be pruned In each layer of the branch and bound method, the effective set method is used to solve the duality problem.In order to reduce the complexity further, a Cholesky decomposition update is used to solve a linear system problem in the iterative process of the effective set method. By relaxing the pruning condition , A quasi-branch demarcation method is given and a good compromise between performance and complexity is achieved.The numerical simulation shows that the complexity of the MIMO detection algorithm based on the branch-and-bound method is very low, especially at low SNR and high order modulation , Its superiority is especially obvious.