Deterministic pilot design for structured sparse channel estimation in MISO systems

To reduce the pilot overhead, compressive sensing techniques have been widely adopted for sparse channel estimation. In this paper, we explore the spatial common sparsity across different channels and investigate the deterministic pilot design with superimposed pattern for the downlink channel estimation in multi-input single-output systems. Previous works generally allocate pilots randomly when using the superimposed pilots to estimate the jointly sparse channels, which is storage-hungry and time-consuming. For the joint optimization of pilot locations and symbols, this paper proposes two pilot design schemes, named by Algorithms 1 and 2, to minimize the intrablock and interblock coherence simultaneously. Algorithm 1 sequentially optimizes pilot locations and symbols, while Algorithm 2 integrates the allocation of pilot symbols into the optimization of pilot locations and alternately optimizes the pilot locations and symbols. Both schemes can flexibly select the design criteria to ensure the small intrablock and interblock coherence. Simulation results show that the proposed pilot designs outperform the existing ones in terms of channel estimate mean-squared-error and bit-error-rate.