Reweighted nuclear norm minimization of interference alignment for cognitive radio networks

Abstract

A cognitive radio network allows the primary users (PU) and secondary users (SU) transmit simultaneously given that the primary user's transmission is not disrupted. The secondary users are able to transmit their signals by aligning the signal direction to the primary user's unused direction. However, the performance of SU is heavily affected by the degree of freedom (DoF) of the cognitive radio network in a static flat-fading multiple input multiple output (MIMO) interference channel. A rank constraint rank minimization (RCRM) method has been used to maximize the DoF but the optimization problem happens to be a non-convex problem thus, a tighter convex approximation will help to solve this problem. One of the most popular methods is the nuclear norm minimization method which provides a convex envelope of the approximation but discovered to be not optimal in finding the maximum achievable DoF. This paper proposes a reweighted nuclear norm minimization method in a cognitive radio network with the presence of PU and multiple SU with the interest at the SU's receiver side. The proposed method allows the PU and SU to coexist in the same frequency band and transmit simultaneously without disturbance from SU to PU while avoiding degradation of performance for SU at the same time. The weight matrix is placed at the receiver and updated iteratively according to the current environment, resulting in a tighter convex approximation and thus, enhances the performance of SU.