In a cognitive radio network (CRN), allocating spectrum resources to terminals is a fundamental problem due to dynamic characteristics of wireless spectrum. However, distributed spectrum allocation schemes are usually sub-optimal due to lack of global information, and existing centralized schemes merely assume the existence of global information collection method without implementing the protocol. In this paper, we propose to use SDN framework for bandwidth allocation and global information collection. Our system, called software defined multi-hop cognitive radio network (SDM-CRN), has a centralized controller executing high level strategies of resource allocation by communicating with the terminals via the OpenFlow protocol. In SDM-CRN, all routers have caching capabilities so that a user request may be satisfied by local routers. Consequently, the traffic in CRNs may be unicast, muticast, converge-cast or hybrid. To satisfy these traffic demands, SDM-CRN jointly optimizes the selection of radio interfaces, channels, and power levels. We conjecture that this problem is NP-hard, and propose a heuristic algorithm. This algorithm constructs a 4-dimensional conflict graph (4-DCG), generates a set of maximal independent sets, and allocates spectrum to terminals by linear programming. Extensive simulations show that SDM-CRN can efficiently utilize the bandwidth and improve the user experience in multi-hop CRNs.
展开▼
