Much of the ongoing research in our group deals with validating our ideas under "real world" conditions. While the results we have produced so far have been promising in the limited user studies and wireless simulations, the most compelling validation of our vision of the wireless future will come from demonstrating a running system with long-term user experience. We currently are building, testing, and experimenting along three main lines of research.
Deployable personal router systems
We have built a demonstrate laptop and local-area-network based version of the Personal Router system. These pieces implement much of the functionality that is required of a PR enabled wireless world. Our current system served effectively as our initial research and development platform, but we consider it vital to move beyond the prototype and build a fully functioning system. We want to progress to the point where our system can support networks of low cost devices accessing wide area wireless services offered by a diverse set of providers. After the codebase stabilizes and some more functionality is added, we plan to release our code to the wireless community.
Long term user studies
Our second avenue of research is to extend the user studies that we have conducted. Our limited user studies have demonstrated that a personal router effectively can learn user preferences and translate those preferences into appropriate service selection decisions. We now want to explore how the PR performs over longer time durations. The learning techniques we employ should improve the PR device's preformance results with time.
Dynamic configuration of collocated access points
Finally, on the provider side of the Personal Router project, we plan to implement our protocols for dynamically configuring networks of collocated wireless access points. Our basic algorithmic approach has been validated in simulation, but we want to build a Linux HostAP based network of wireless access points. Particularly in wireless networks, where simulations so often do not correspond to reality, validation through real world implementation is essential.