Multiple-symbol differential sphere decoding

In multiple-symbol differential detection (MSDD) for power-efficient transmission over fading channels without channel state information, blocks of N received symbols are jointly processed to decide on N - 1 data symbols. The search space for the maximum-likelihood (ML) estimate is therefore (complex) (N - 1)-dimensional, and MSDD quickly becomes computationally intractable as N grows. Sphere decoding (SD) is a widely used approach to find the ML estimate in such high-dimensional spaces. In this paper, we devise the application of SD to accomplish MSDD, and we refer to the resulting technique as multiple-symbol differential sphere decoding (MSDSD). We present an efficient algorithm for MSDSD, whose excellent performance versus complexity trade-off is verified by various simulation results and by comparisons with other, suboptimum approaches known from the literature.