Developing a distributed clustering algorithm to enable self-managing groups for 802.11 access points

Co-channel interference is the most significant degrader of quality of service in 802.11 Wi-Fi in dense residential areas. The number of observed neighbouring networks can grow as large as 20-30 networks in a modern apartment building. This makes interference from neighbouring networks a big issue. It can degrade the perceived quality of service to the point where private consumer networks can no longer provide any- where near the throughput guaranteed by the ISPs’ service level agreements. This problem would be easier to solve if access points in residential areas had the opportunity to cooperate and coordinate the distribution of channels in a way similar to what some centralized solutions for enterprise networks offer today. This thesis is focused on developing a clustering algorithm that can define clusters of access points in a distributed and chaotic network topology. These clusters could provide a future framework for communication and coordination between access points. We specify a set of requirements for the distributed clustering algorithm and then go through the development stages of the algorithm, beginning with a minimal working algorithm, and iteratively progress until a satisfactory solution is found that meets the requirements. Lastly, we suggest technologies that could facilitate the deployment of this clustering algorithm in a real world implementation on access points, and consider how these technologies could interface with each other to provide the required services