Pervasive Computing Research
The Mobile and Pervasive Computing Lab is focused on systems research and experimental aspects of pervasive computing, emphasizing four main research thrusts: fundamentals, applications, enablers and smart space deployments
Fundamental research in pervasive computing addresses defining approaches to: (1) cyber-physical interfacing — integrating sensors, devices, and other computing and communication elements into a single system that we call a “smart space”. Such integration will not be scalable unless it is autonomic—stemming from the devices and requiring no engineering specialties (human system integrators). Autonomic integration makes it possible to rapidly deploy instrumented spaces in a radically cost-effective manner. The next fundamental challenge is to turn such deployments into smart spaces, not by any ad hoc means, but rather by properly (2) programming such pervasive spaces. Such programmability is key and fundamental to the progress and realization of the pervasive and ubiquitous computing vision. Anything less than full programmability is prototyping, which is not the ultimate goal. To achieve programmability, novel programming models specifically conceived of and designed for pervasive spaces are needed. We are currently working on several programming models and their implementation in terms of middleware, algorithms, and run-time support. We are researching models that can be used to program a space directly through its sensors and actuators, in addition to others that can program a space through (3) sentience abstractions of sensors (e.g., events, contexts, phenomena, activities, behavior, etc.). Once we realize programmable pervasive spaces, it becomes inevitable to see the early signs of an ecosystem in which pervasive space applications can be developed and provisioned by third parties as services to these spaces. Hence, at the outset, we must think of cloud computing and how smart space applications and services will be provisioned through the cloud. To this end, we are currently developing (4) cloud-sensor architectures, algorithms, and optimizations to enable energy-efficient and sentience-efficient smart space operation.