To Relay or Not To Relay in Cognitive Radio Sensor Networks

Abstract

A recent investigation on MAC layer in cognitive radio networks proposed the primary packet relaying by the secondary node maintaining an extra queue used for this particular addable functionality. Nevertheless, relaying of primary packets may introduce delays on secondary’s packets, called secondary delay, and may require an additional power budget in order to forward the primary packets. Power budget is especially crucial when a type of sensor network is deployed using devices of limited power resources. In this paper, admission control is employed in order to manage efficiently this packet-wise relaying process in cognitive radio sensor networks. To be specific, we assume a cognitive packet relaying scenario with two pairs of primary and secondary users, i.e. transmitter and receiver. We analyse and formulate the secondary delay and the required power budget of the secondary sensor node in relation with the acceptance factor (i.e., admission control parameter) that indicates whether the primary packets are admitted for relaying or not. Having defined the above, we present a tradeoff between the secondary delay and the required power budget by tuning the acceptance factor, that can be tailored to specific chosen values. Based on this behaviour, we formulate an optimization problem to minimize the secondary delay over the admission control parameter subject to a limit on the required power budget. Additionally, the constraints related to the stabilities of all individual queues at the primary and secondary networks are taken into account in the proposed optimization problem, due to their interdependencies. The solution of this problem is provided using iterative decomposition methods, i.e. dual and primal decompositions, using Lagrange multipliers that simplifies the original complicated problem and results in a final equivalent dual problem that includes the initial Karush Kuhn Tucker (KKT) conditions. We obtain the optimal acceptance factor while in addi- ion we highlight the opportunities for extra delay minimization that is provided by relaxing the initial constraints through changing the values of the Lagrange multipliers. Finally, we present the behaviour of the secondary delay assuming infinite and finite queues and assessing thereby the overflow and blocking probabilities respectively.