Statistical QoS driven power allocation for cognitive networks under primary user's outage probability constraint

Resource allocation is a critically important issue for cognitive networks, because it needs to not only meet the quality-of-service (QoS) requirements for the secondary users (SU), but also protect the QoS of primary users (PU) from degradation. In this paper, we develop the optimal power allocation scheme for the cognitive network, which can satisfy the QoS requirements of SUs and PUs simultaneously. Specifically, on the one hand, as the deterministic delay QoS provisioning is usually unrealistic for practical wireless networks, we use the theory of effective capacity for SU's statistical delay QoS provisioning. On the other hand, in order to meet the PU's QoS demand, we impose the PU's outage probability constraint on the cognitive network instead of the traditional average and/or peak interference power constraints. Following the above concept, we derive the optimal power allocation scheme for the cognitive network under the SU's average and peak transmit power constraints and the PU's outage probability constraint. We find from simulation results that 1) the effective capacity of the cognitive network decreases as the statistical delay QoS requirement becomes stringent; 2) the performance of the cognitive network can be improved if the PU can tolerate higher outage probability; and 3) under the given average transmit power and PU's outage probability constraints, the cognitive network can achieve better performance while increasing the maximum peak transmit power, but the obtained performance gain for the stringent QoS requirement is more obvious than that for the loose QoS requirement.