Anomalies in optimal rate-control and scheduling protocols for Cognitive Radio Networks

Protocols in Cognitive Radio Networks (CRNs) tune the radio and network parameters to utilize the unused frequency spectrum, and thus improve the application performance. Generalized Network Utility Maximization (GNUM) approach has been used to derive optimal algorithms in CRN from a theoretical formulation of the problem. However, there are several practical challenges for realizing these algorithms in a networking system. In this paper, we consider a joint source-rate, routing and scheduling GNUM problem in CRNs. We discuss three issues of protocols that are derived directly from the optimal algorithms: spurious pressure points, bursty scheduling and the need for link pruning. We show that the above practical problems result in serious system implications such as large message passing overheads, packet delays and buffer requirements.