Per-subcarrier antenna selection for H.264 MGS/CGS video transmission over cognitive radio networks

In this paper, the problem of efficiently allocating wireless resources to support multiple scalable video sequences over a downlink cognitive radio network is addressed. We consider the coarse grain scalable (CGS) and medium grain scalable (MGS) extension of the H.264 standard as the encoding method and propose to perform a per-subcarrier transmit antenna selection such that spatial diversity is exploited. We formulate the problem of optimally allocating subcarriers and antennas among secondary users as a binary integer program, where optimality is defined in terms of the aggregate visual quality of received video sequences of all cognitive users. The proposed formulation caters to the staircase rate-distortion characteristics of CGS/MGS sequences and avoids allocating more resources than are necessary to attain a given visual quality. However, due to the high computational complexity involved in solving this hard integer program using discrete programming solvers, we reduce the resource allocation problem to a subset-sum problem. Although it remains ${cal NP}$ -hard in general, we present two efficient methods to solve this subset-sum problem based on its structure. Simulation results demonstrate that the methods proposed lead to a solution that is close to the optimal. Moreover, we demonstrate that having multiple antennas at the cognitive base station reduces the outage probability of secondary users.