Abstract: STBC-OFDM systems can provide diversity and coding gains.However,ML(Maximum Likelihood) has expensive computational complexity,which can not satisfy the need of high-velocity and real-time communication system.This paper presents a sphere decoding algorithm that searches new vector in a sphere radius.The central part of the algorithm is a modulation-independent sphere decoding framework formulated in the complex domain.This paper presents a new better searching method based on(reordered) vectors in complex field.This approach optimizes sphere radius iteration,increases the searching speed of signal vector.In order to compare with other algorithms,this paper develops two decoding(approaches): a modulation independent approach applicable to any memoryless modulation method,a QAM-specific fast decoding algorithm performing nearest-neighbor signal point search.The computational complexity of the algorithms is investigated via both analysis and simulation.The result of simulation shows this approach has more superiority on computational complexity and increases the efficiency of decoding,which demonstrate that the proposed algorithm can significantly reduce the decoding complexity.We observe up to 75% reduction in the required FLOP count percode block compared to previously existing methods without noticeable performance degradation.