| Resumen |
In this paper, a novel joint connection-level and packet-level analytical model for the performance evaluation of cognitive radio networks (CRNs) under VoIP traffic is proposed and developed. The proposed teletraffic model captures the most relevant features of both CRNs (i.e., unreliable spectrum sensing, primary users' activity detection) and VoIP traffic (i.e., ON/OFF activity processes for individual VoIP sessions, periodic packet generation of individual VoIP sessions in the ON state). In addition, our mathematical model captures relevant aspects of multiuser VoIP-based networks such as the use of packet buffering, adaptive modulation coding, proportional fair sharing scheduling for secondary users with different link qualities, and call admission control (CAC). Our mathematical analysis is based on the time-scale decomposition technique and employs conventional discrete time Markov chain tools. However, the most important feature of our mathematical formulation is the introduction of the so called cross-level analysis methodology used here to link packet-level and connection-level analyses. This proposed methodology allows to use time-scale decomposition technique when it is no longer possible to perform the connection level analysis in an independent manner from the packet level analysis (i.e., when connection level statistics depend on packet level parameters). |
