Hierarchical DWT based optimal diversity power allocation for video transmission in MIMO wireless systems with quantized feedback

Abstract

In this work we propose novel algorithms for singular mode diversity order based optimal power allocation for video transmission in Multiple-Input Multiple-Output (MIMO) wireless systems. We motivate a discrete wavelet transform (DWT) and hierarchical block motion estimation algorithm (HBMA) based spatio-temporal video decomposition that is naturally suited to the ordering and diversity properties of the MIMO channel singular modes for video layer allocation. Based on this paradigm, we propose a framework for the MIMO singular value decomposition (SVD) based optimal power allocation, specifically for the context of MIMO video transmission. Further, we consider a practical MIMO scenario with channel state information (CSI) available only at the receiver and employ a codebook based limited feedback technique to feedback the index of the quantized beamforming vector. This significantly reduces the communication overhead on the reverse link of the wireless system. We demonstrate that the proposed hierarchical video decomposition based optimal diversity power allocation employing limited CSI feedback can be formulated as a constrained optimization problem. Subsequently, it is demonstrated that the solution to this problem can be computed by solving an iterative sequence of convex cost minimization problems. A closed form expression for the optimal power allocation is presented for the convex optimization problem at each iterative step. Simulation results demonstrate significant performance improvement of the proposed optimal power allocation schemes over the suboptimal equal power allocation scheme.