摘要
针对多调制指数连续相位调制全状态序列检测复杂度较高的问题,提出一种基于倾斜相位、特征值分解和状态空间分类的Multi-h CPM低复杂度序列检测算法。利用特征值分解获得信号降秩表示,将信号状态进行分类融合,得到减少状态网格,引入判决反馈进行状态转移约束,实现匹配滤波器和状态数的减少。给出最大似然序列检测误码性能下界并进行仿真实验,实验结果表明,以ARTM CPM TierII信号为例,该算法可将匹配滤波器数减少为全状态所需的3/64,网格状态减少为全状态所需的1/8,在误码率为10-5时性能损失仅为0.2dB。
A reduced-complexity sequence detection algorithm for Multi-h CPM signals based on tilted phase combined with eigenvalue decomposition（EVD）and state-space partitioning（SSP）was proposed to solve the complexity-reducing problem.The optimal low-rank approximation was achieved through the eigenvalue decomposition of the signal correlation matrix.A reducedstates trellis was formed through classifying and fusing states and decision feedback was introduced to restrain the states transition.In this way,the number of filter and states was reduced.Results of analysis and simulation show that the algorithm can reduce the number of match filters and trellis states sharply with low cost.For ARTM CPM,it can cut the amount of match filters and states down to 3/64 and 1/8of the original respectively with 0.2dB loss when the error probability is 10-5.
出处
《计算机工程与设计》
北大核心
2015年第11期2926-2930,共5页
Computer Engineering and Design
