Performance Enhancement of Cognitive Relay Networks Using Regression Functions

In a spectrum sharing relay-aided secondary network, the design objective shifts from minimizing the violation of the primary system constraints when using the outdated channel state information (CSI), to making the best use of the available (imperfect) channel gain while maintaining the primary system Quality of Service (QoS). To address this trend, we propose a new power allocation based on a linear regression model and we carry out the outage analysis as well as the asymptotical analysis where we deduce the diversity order of the secondary network. The performance results show that the proposed power allocation strategy outperforms the existing schemes in the literature and provides an accurate tight approximation of the ideal case under perfect CSI while respecting the primary system QoS and using outdated CSI.