Pilot allocation by Genetic Algorithms for sparse channel estimation in OFDM systems

The comb-type pilot-aided channel estimation in OFDM systems is here considered. The number of pilot subcarriers is generally tuned to tradeoff between performance and spectral efficiency. For time domain sparse channels, the application of the recent theory of Compressed Sensing has proved its efficiency to both ameliorate the estimation accuracy and reduce the transmitted overhead. The performance and uniqueness guarantees remain however closely related to the pilots placement, which are optimized prior to transmission. Few works have addressed the pilot allocation issues. To avoid the computationally prohibitive exhaustive search over all possible allocations, they proposed sub-optimal solutions. This contribution proposes an enhanced pilot design scheme based on Genetic Algorithms. First, we set the adopted problem formulation in the frame of Genetic Algorithms. Then, we propose two algorithms, tailored for the decomposition matrix coherence minimization. Simulation results prove the proposed schemes effectiveness and performance enhancement compared to former pilot design schemes.