Initial Radius Selection of Sphere Decoder for Practical Applications of MIMO Channels

Sphere Detection (SD) method has been used as a powerful mean to find Maximum Likelihood (ML)solution to the detection problem of Multiple-Input Multiple-Output (MIMO) channels with lower complexity. This method searches closest lattice point inside a sphere. This paper discusses the effect of initial radius of the sphere on the complexity and performance of SD in different noise variances, constellation sizes and number of receive and transmit antennas. According to the discussion, a function is introduced which shows the relation between computational complexity and initial radius. As a result, a useful method for selection of suitable initial radius in practical applications is derived.