Near-capacity performance soft output sphere decoding based on short list detection and metrics combining

We introduce a low complexity iterative soft output sphere decoding algorithm for coded transmissions over multiple antenna channels. Before the iterative detection and decoding starts, a modified hard decision sphere decoder produces a short (base) list of vectors with maximum likelihood metrics. In subsequent iterative soft detections, two competing lists with a small number of vectors are further generated for each coded bit, by utilizing the base list vectors and a priori information from the channel decoder. The corresponding likelihood metrics of the vectors in each competing list are combined to produce soft detection output that approximates the optimal maximum a posteriori (MAP) solution. The performance improves as the base list size increases and a short list (hence a low number of competing vectors) can provide near-capacity performance after a few iterations. Compared with existing methods that adopt the max-log approximation and select only a single best competing vector, the proposed algorithm approaches the optimal performance better with significantly lower complexity requirements.