CANSAS: Crowdsourcing Access Network Spectrum Allocation using Smartphones
As wireless devices proliferate and bandwidth demands grow, it is more important than ever for wireless infrastructure providers to increase spectrum efficiency in access networks by adaptively allocating limited spectrum in response to
changing demands. Optimal spectrum allocation requires measurements from both the AP and terminal ends, since both ends can experience independent interference. One way to gather this data is for network providers to conduct periodic site surveys
using dedicated devices, but these measurements are expensive to perform and yet fail to represent actual user experience and temporal fluctuations in channel usage and conditions. While client-side measurements are more representative, taking
measurements by active terminals can interrupt their link traffic and reduce spectral efficiency. Finally, harnessing measurements from both ends of the link to improve spectral efficiency remains an open problem, especially when there are terminals
or infrastructure providers competing for limited spectrum resources.
To address these challenges this project introduces the concept of crowdsourcing access network spectrum allocation using smartphones (CANSAS). CANSAS utilizes the rapidly-growing number of smartphones, which are always on but mostly idle, to perform triggered and periodic observation of both their own wireless performance and that of other nearby active terminals. Channel measurements from (idle) smartphones are used as inputs to new radio resource management algorithms which improve spectral efficiency through dynamic allocation of channels and transmission powers, and also through scheduling uplink and downlink transmissions. Game-theoretic approaches are used to incentivize both terminal data collection and cooperation between competing terminals and networks in order to achieve socially-optimal spectrum allocation within a variety of environments. In addition to developing optimal spectrum allocation schemes based on network utility maximization and graph theory, the project will implement an experimental CANSAS system for WiFi networks called PocketSniffer.