Research On Dynamic Scheduling Mechanism In Cognitive Networks

It is a difficult but fundamental goal to fully utilize various resources to deliver data as efficiently as possible in resource-limited wireless networks. In cognitive net-works, it becomes more challenging due to some unique characteristics. First, the net-work is highly dynamic due to reasons such as the unrestricted mobility of SUs and time-varying channel availability, which makes connections between SUs unstable. Second, in the real world, traffic may arrive to the network at any time with arbitrary rate requirement, so it is hard to achieve the network-wide optimal performance. Third, to balance traffic load and maintain network stability, each SU should carefully control its admitted date rate and data transmission. Finally, without central coordination, each SU has to make decisions only by itself and the control algorithm should be operated in a distributed manner.In this paper, we propose a Joint Rate adaptation, Channel assignment and Route Selection (JRCRS) approach to optimize the resource utility in the multi-channel, multi-hop cognitive network, with the objective of maximizing social welfare. Main contributions are summarized as follows.· We set up the node, link and flow model for multi-channel, mutli-hop cognitive network, and model it as an undirected graph, we extend the queuing technology of Lyapunov theory into multi-channel cognitive network, and set up the real queue model for each node. Based on the queue stability theory, we construct a virtual queue to ensure the minimum rate requirement for flows.· In order to encourage the cooperation among secondary nodes, we introduce the pricing scheme into data transmission and establish a virtual trading market for cognitive networks. We detine the service price, and use it as the routing metric to select the next hop. The routing metric considers the real queue backlog and the distance between the relay and destination node, which efficiently reflects the transmission cost of links.· We formulate the dynamic scheduling problem in cognitive network, using the mixed nonlinear integer programming. Its objective is to maximize the average social welfare. Based on the Lyapunov optimization of stochastic networks, we design a distributed dynamic scheduling scheme JRCRS. It jointly optimizes rate adaptation, channel assignment and route selection in order to satisfy the min-imum rate requirement of flows, assigns interference-free channels, and selects the next hop with lighter workload and shorter distance to the destination node. In our JRCRS, nodes can make their decisions just based on the current network states.· Finally, we develop a simulation system for the multi-channel, multi-hop cogni-tive network in order to illustrate the efficiency of our JRCRS. Comprehensive performance evaluations are conducted to compare JRCRS with related solu-tions, such as BPR and GPSR, in terms of average throughput, queue backlog, end-to-end delay and so on. We test the network stability, and the influence of flow number, which illustrates that our JRCRS outperforms BPR and GPSR.