In this paper, the problem of radio and power resource management in underlay cognitive radio heterogeneous networks is investigated, where macro and pico Base Stations (BSs) are considered as primary BSs while femto BSs are considered as secondary BSs. The goal is to minimize the total primary power consumption and maximize the secondary utility of the network while satisfying the primary user quality of service determined by target data rate and interference constraints. Furthermore, a green communication algorithm is implemented based on a sleeping strategy. Simulations study investigates the performance of the proposed scheme and shows an important saving in terms of total power consumption.