Sphere decoding for multiple antenna systems has been shown to achieve near-ML performance with low complexity. However, the achievement of such an excellent performance-complexity tradeoff is highly dependent on the initial choice of sphere radius. In this paper, we present a new sphere decoding algorithm which is even less computationally complex than the original sphere decoder. Moreover, the complexity of the new sphere decoder is relatively insensitive to the initial choice of sphere radius. Thus, by making the choice of radius sufficiently large, the ML solution is guaranteed with low complexity, even for large constellations. In our simulations, we show that with 4 transmit and 4 receive antennas and 64-QAM, our new sphere decoding algorithm achieves the exact ML solution with approximately a factor of 3.5 reduction in complexity when compared to the original sphere decoder, and a factor of 105 reduction when compared to brute-force ML decoding.
ASJC Scopus subject areas
- Computer Networks and Communications
- Electrical and Electronic Engineering