Channel Assignment Based on Routing Decisions (CARD): Traffic-Dependent Topology Control for Multi-Channel Networks

Abstract

Dynamic spectrum access (DSA) holds the promise for more efficient utilization of the spectrum, while requiring greater cooperation between PHY, MAC, and NET layers to allocate resources and dynamically react to changing network conditions. In this paper, we propose Channel Assignment based on Routing Decisions (CARD), a mechanism that combines channel assignment and topology control so that at any given time the cognitive network self-organizes into the topology that is best suited to support the current offered traffic. CARD is a distributed mechanism, and each network node relies on local information only. We show that CARD results in an improvement in route length (and, thus, end-to-end delay), aggregate network capacity, and, in some situations, energy efficiency. In the process of designing a DSA medium access con- trol (MAC) scheme, new abstractions that span the physical, data-link, and network layers must be developed. This design encompasses the traditional PHY/MAC/NET functions along with new issues such as neighbor discovery in a multi-channel environment, rendezvous, and intelligent channel assignment. In this paper, we propose an approach that combines channel assignment and topology control, so that at any given time the network self-organizes into the topology that is best suited to support the current offered traffic. This approach, Channel Assignment based on Routing Decisions (CARD), is a distributed mechanism where each node makes decisions given its local information. Each node inspects the current route and attempts to shorten the route through dynamic selection of channels. It produces, whenever feasible, both an interference-free channel assignment for nodes actively involved in routing and shorter routes from source to destination. Preliminary results indicate that this method of jointly assigning channels and routes has benefits such as reduced delay and increased network capacity. In some cases energy-efficiency improvements are also obtained, and if enough channels are available intra- and inter-flow inter- ference avoidance techniques can be applied to improve network throughput.