Abstract：Two low-complexity Particle Filtering (PF) detections for Multi-Input Multi-Output (MIMO) systems, namely sphere-constrained PF and multi-level mapping PF, are proposed by reducing the sample size and the search space of signal detection, respectively. In the proposed sphere-constrained PF, a sphere bound is first obtained based on zero-forcing principle, then the sphere bound is utilized to decrease the number of particles resulted by the importance sampling of each stage in the PF procedure. While the proposed multi-level mapping PF partitions the high-order Quadrature Amplitude Modulation (QAM) constellation of size 4L into L 4-QAM constellations with the aid of multi-level mapping, which reduces the search space of signal detection. Simulation results show that the first method can reduce the computational complexity of PF detection effectively without performance degradation especially when the number of transmit antennas is large; and the second method can significantly reduce the computational complexity at the cost of little performance degradation.
2009, Vol. 31 
