| In the last decade, studies performed by the U.S. Federal Communications Commission have revealed poor utilization of the allocated spectrum in US. This fact motivates the concept of cognitive radio that allows secondary networks to utilize licensed radio spectrum when the primary network is temporally not using it.;The Sensing-Throughput problem in cognitive radio has been studied in this work. Unlike previous research, which assumes a steady channel state for the whole frame, this work employs a discrete Markov model to cover all the possible channel state transitions in the frame. Based on the new channel model, an estimation of the interference according to the sensing result is used for the determination of when to transmit. Numerical results compare the new channel model with the ideal model in terms of throughput, interference and energy efficiency change with different length of the sensing period within a fixed-length frame. Given a proper setting of the interference threshold, the proposed model achieves a higher normalized throughput and similar energy efficiency with the corresponding optimal frame structures.;Recent research has also investigated the capability of cognitive radio networks to handle real-time applications, such as video streaming. To support both real-time(RT) and non-realtime(NRT) traffic in a cognitive radio network, existing protocols give absolute priority to real-time packets in spectrum allocation and use an earliest deadline first scheme in MAC layer scheduling. However, this approach doesn't consider the frame level delay constraints nor the dependency issues in decoding P-frames, such as the live video streaming is in H.264 standard. In this work, the scheduling of two types of co-existing traffic, H.264 RT packets and NRT packets, in the downlink of cellular cognitive radio networks has been studied and a cross-layer designed scheme that involves active P-frame dropping is presented. Simulation results indicate the cross-layer scheme greatly preserves the QoS of RT packets, especially the successfully decoding rate of P-frames, with the sacrifices of I-frames. Still the cross-layer scheme has a much higher decodable rate of all frames, and the QoS of NRT traffic is also improved based on the more reasonable utilization of the spectrum. |
