Parameter Estimation and Prediction using Rotational Invariant Techniques in MIMO System using USRP

In this work, channel estimation and prediction using rotational invariance techniques namely 1D ESPRIT, Unitary ESPRIT and Multidimensional ESPRIT is considered. These techniques account for fading effects caused due to spatial, temporal and polarization changes. Direction of arrival is also estimated using 1D ESPRIT, Matrix Pencil and Unitary ESPRIT. The major contribution in this work is channel prediction performed for antenna spacing less than λ/2 and the analysis indicated normalized mean square error to be same as obtained for antenna spacing greater than λ/2 for prediction interval of 0.3ms. This analysis paves way for achieving low spatial correlation in MIMO for antenna spacing less than λ/2. Experimental setup using Universal Software Radio Peripheral- Networked Series (USRP-N210) operating at a frequency of 2.4 GHz was used to validate the simulated result. The experimental results indicate reduced prediction error at high SNR and similar NMSE performance for antenna spacing less than and greater than λ/2