Pilot contamination posts a elementary limit on the performance of huge multiple-input–multiple-output (MIMO) antenna systems owing to failure in correct channel estimation. To address this drawback, we tend to propose estimation of solely the channel parameters of the specified links during a target cell, however those of the interference links from adjacent cells. The desired estimation is, nonetheless, AN underdetermined system. During this paper, we show that if the propagation properties of huge MIMO systems will be exploited, it's potential to get a correct estimate of the channel parameters. Our strategy is impressed by the observation that for a cellular network, the channel from user instrumentality to a base station consists of solely a number of clustered methods in space. With an awfully massive antenna array, signals may be discovered under extraordinarily sharp regions in space. As a result, if the signals are discovered within the beam domain (using Fourier transform), the channel is around thin, i.e., the channel matrix contains only a little fraction of huge elements, and different elements are near zero. This observation then permits channel estimation based on thin Bayesian learning strategies, wherever thin channel components may be reconstructed employing a little variety of observations. Results illustrate that compared to traditional estimators; the planned approach achieves far better performance in terms of the channel estimation accuracy and doable rates in the presence of pilot contamination.