Cross-layer sequential sensing with effective throughput maximization in time-slotted cognitive networks

This paper studies slotted cognitive networks which include multiple spectrum bands and time is divided into slots. In order to increase the opportunity to use idle band, we consider that a secondary user sequentially senses multiple bands within a time slot. To maximize effective throughput excluding sensing overhead, transmission overhead and useless transmission while interference is guaranteed, we shall determine sensing time, frame size, and which band a secondary user sends frames in. We study two sequential sensing policies, namely forward sensing and recall sensing, which determine which band to transmit; the former policy forwardly senses bands one by one, and the latter policy senses a band and records the related information which is used to return back to the band if the effective throughput of other being sensed bands is lower than that of the band. Analyses verified by simulations are developed for the sensing policies; based on the analysis, we find optimal sensing time in physical layer, frame size in medium access control layer. We compare the forward and recall sensing policies with a sequential sensing merely with information in physical layer. Numerical results show that (i) the forward and recall sensing policies outperform the compared sequential sensing policy and (ii) the recall sensing policy outperforms the forward sensing policy at light load. In addition, the effects of sensing time and frame on the forward and recall sensing are extensively studied.