Low complexity channel estimation for SISO, MIMO and massive MIMO OFDM wireless system

In OFDM based wireless communication systems, whether employing single or multiple antennas, channel state information has to be estimated accurately and that too within a fraction of time, making channel estimation very crucial. Against various state-of-the art on channel estimation, this thesis presents several low complexity channel estimation techniques for SISO, MIMO and massive MIMO OFDM systems by exploiting the structure and some of the constraints of communication problem.;We first present a reduced complexity optimal interpolation technique for SISOOFDM systems based on MMSE criteria. By utilizing the structure of channel frequency correlation, it is shown that if pilots are placed appropriately across OFDM subcarriers, the matrix inversion in conventional MMSE estimation can be completely avoided with no loss in performance. Next, we present a blind ML algorithm for joint channel estimation and data detection for MIMO-OFDM systems with Alamouti coding where the complexity is reduced by again utilizing the correlation structure and the finite alphabet property of symbols. A semi-blind algorithm is also introduced which has much lower complexity than the blind algorithm but at the cost of few training symbols.;As for the massive MIMO systems, the complexity is of primary concern because with increased number of base station antennas (BS), the number of unknown channel parameters also grow large. Unlike the optimal MMSE approach, which is prohibitively complex, we present a distributed MMSE algorithm whose complexity is linear in the number of BS antennas while at the same time achieves near-optimal performance by sharing the information locally in a large antenna array. A data-aided version of distributed algorithm is also presented to minimize the pilot overhead in massive MIMO. Finally, we investigate the effect of pilot contamination (i.e., interference due to reuse of pilots) on MSE performance of various algorithms. We use stochastic geometry to derive closed-form expressions for channel MSE under both noise and pilot contamination regime, which are validated by simulations. Our results indicate severe implications of pilot contamination on channel estimation performance.