Spectrum agility (SA) has been proposed as an important means of improving the efficiency of spectrum utilization, yielding greater bandwidth availability to network applications. Recently, there have been a number of proposals for SA protocols in literature. Despite the associated advantages, the fundamental operations involved in realizing SA can also adversely affect application performance. For example, application traffic is interrupted when a channel is scanned, which directly affects a network application’s delay performance. Such unwarranted side-effects of SA have not been addressed in the SA protocols proposed thus far; this can potentially negate the advantages of SA for many applications. In this project we take an application-centric view towards SA. We argue that application requirements or hints can serve as key parameters in evaluating channel conditions relative to application needs, leading to an informed SA mechanism. We propose Context-Aware Spectrum Agility (CASA) that uses application hints together with current channel state, in order to make intelligent decisions in manipulating key agility parameters in an adaptive fashion. To enable cross-layer communication from the application layer to spectrum agility module (SAM), we use a lightweight cross-layer communication module, called the Cross-Layer Interaction Framework (CLIF). Our evaluation through simulations shows that CASA improves throughput by more than 35%, and is also 50+% more effective in meeting application demands than conventional SA.
Currently we are in the process of evaluating CASA on a wireless testbed. This involves implementation of a basic SA protocol in wireless cards. We are experimenting with Atheros-based wireless cards that can operate using open source Mad-wifi drivers in Linux.
- Ashwini Kuamr, and Kang G. Shin, Extended Abstract: Towards Context-Aware Wireless Spectrum Agility, Proceedings of the 13th ACM International Conference on Mobile Computing and Networking (MobiCom’07), Montreal, Canada, pp 318-321, September 2007. PDF