The main objective of this dissertation is to propose a low-complexity hardware realization of the channel This dissertation is proposing a low-complexity hardware realization of the channel estimation for the orthogonal frequency-division multiplexing (OFDM) system with electronically steerable parasitic array radiator (ESPAR) antenna. Consisting of a single active element surrounded by multiple parasitic elements, ESPAR antenna can achieve the similar reliability and capacity to that of multiple antenna systems but only using one RF front-end hardware set. ESPAR antenna has a potential of realizing multiple antenna systems with high energy efficiency. However, the ESPAR based system requires that the channel estimation should be realized in time-domain because the received signal for each antenna elements will be overlapped each other both in the frequency and time domain. Among the existing methods, compressed sensing (CS) technique has been proven to be a suitable channel estimation for ESPAR antenna. However, due to its very large sensing matrix, the CS algorithm suffers from a computational burden, making it infeasible for many practical applications.
This dissertation proposes three methods to reduce the computational cost of CS in ESPAR antenna deployment. The first method is a multi-column CS which utilizes only one small segment of sensing matrix to detect all of the channel impulse response (CIR) locations simultaneously. The second method is the matrix strength reduction by taking advantage of the structures of sensing matrix and discrete Fourier transform (DFT) matrix. The third method is the observation vector optimization by randomly selecting a small set of optimal pilot locations at the receiver side, which can reduce the CS computation complexity involved in CIR location search. The proposed hardware can achieve 90% reduction in the computational cost and also 1700% faster computation time.