On improved channel estimation for OFDM systems using the additive eigen value inflation factor approach

TLDR: In this article, a biased generalized ridge regression estimator (GRRE) was proposed to obtain a reduced MSE over the unbiased estimator for the specular wireless channel, and the proposed estimator elegantly reduces to the optimal MLS solution at high SNRs and hence does not experience any MSE floors.

Abstract: Knowledge about the true temporal tap delay locations of the wireless channel allows construction of the CRLB attaining modified-Least squares or the MLS channel estimator However, there exist several biased estimation paradigms which breach the CRLB for the unbiased counterparts, and which render the MLS inadmissible for orders of parameter estimation greater than 2Harnessing the fact that the specular wireless channel is often characterized by more than two taps, we propose to exploit the much forgotten Bhattacharya's algorithm to construct the biased generalized ridge regression estimator [GRRE] The algorithm provides us with optimal values of the so-called additive Eigen value inflation factors which when incorporated within the GRRE framework, guarantee a reduced MSE over the MLS estimator We analytically derive the attainment of the CRLB for biased estimators and then prove that there always exists a positive threshold SNR over which the proposed estimator outperforms the MLS estimator Further, we show that the proposed estimator elegantly reduces to the optimal MLS solution at high SNRs, and hence does not experience any MSE floors We conclude by showcasing the optimality of the proposed estimator through simulations involving vehicular channel estimation