基于树形结构酉空时星座设计的译码方法

针对酉空时码最大似然检测复杂度高的问题，提出了一种酉空时星座的设计及低复杂度译码方法．根据酉空时信号失真测度的定义，依照最邻近原则对星座进行划分，以最小失真准则计算各个星座子集中信号的最佳中心，再通过迭代算法，交替使用两个局部最佳准则，生成由各个信号子集与对应中心所构成的若干子树，形成树形结构的酉空时星座．接收方以接收信号为树根，选择具有最小失真的中心信号．对选定的中心信号对应的子树进行遍历搜索得到对发送信号的估值．该译码方法仅需要对某一中心信号及其对应子树的叶子节点进行计算，避免了最大似然方法对整个星座的全搜索，从而有效地降低了译码复杂度．仿真实验表明，相对于最大似然检测方法，该方法能够以有限的性能损失换取译码计算的有效简化．

Decoding Algorithm Based on Tree-Structured Unitary Space-Time Constellation Design

A decoding algorithm based on the tree-structured constellation design was presented in order to reduce the decoding complexity of unitary space-time codes. The unitary space-time signals of the constellation were classified by using the nearest-neighbor rule and the unitary space-time signals with minimum average distortion were chosen as the optimum center of each group. Two local optimum rules were used alternately in the iterative process of the constellation design and a tree-structured constellation with optimum center signals as the root of each sub tree was generated. Firstly, the receiver determined the sub tree to which the received signal belongs by searching the set of sub tree roots. Then the final decision was obtained for the received signal after searching the leaf node signals in the selected sub tree. Only one sub tree root and its leaf node signals were considered for every received signal in the decoding process and there was no need to fully search the maximum likelihood in the whole constellation. Thus, the decoding complexity could be much lower. Simulations result demonstrates that the proposed algorithm can reduce the decoding computation effectively compared to the maximum likelihood detect method.