关键词: 多符号差分检测, 球形译码, 半径, 复杂度

Abstract: The depth-first sphere decoding is a classical sub-superior detect algorithm in the Multiple Symbol Differential Detection（MSDD） system. However, from the view of complexity, it still has a high computational complexity, and there are difficulties in pipeline and parallel operations. To solve these problems, it focuses on two ways to improve the algorithm at the present： first, choosing appropriate radius to reduce complexity； second, combining with K-Best（M） algorithm to solve the parallel of operational problems. The paper mainly studies the former, and proposes two radius selecting methods based on existing theory, namely linear radius and nonlinear radius. Simulation results show that the sphere decoding with two radiuses has lower complexity than the Maximum Likelihood（ML） detection with the loss of their performance less than 0.5 dB.

Key words: Multiple Symbol Differential Detection（MSDD）, sphere decoding, radius, complexity