Blind null-space tracking for MIMO underlay cognitive radio networks

Blind Null Space Learning [1] has recently been proposed for fast and accurate learning of the null-space associated with the channel matrix between a secondary transmitter and a primary receiver. In this paper we propose a channel tracking enhancement of the algorithm, namely the Blind Null Space Tracking algorithm, that allows transmission of information to the Secondary Receiver while simultaneously learning the null-space of the time-varying target channel. Specifically, the enhanced algorithm initially performs a sweep in order to acquire the null space. Then, it performs modified Jacobi rotations such that the induced interference is kept lower than a given threshold P_Th with probability p while information is transmitted to the secondary receiver simultaneously. The learning process is performed based on sensing whether the transmit power of the primary user has increased or decreased between adaptations. We present simulation results indicating that the proposed approach has strictly better performance over the Blind Null Space Learning algorithm for channels with independent Rayleigh fading at a low Doppler frequency.