Ambient Intelligence: A Computational Platform Perspective

Computational platforms are a key enabling technology for materializing the Ambient Intelligence vision. Ambient intelligence devices will require a widely ranging computational power under widely ranging system-level constraints on cost, reliability, power consumption. We coarsely group computational architectures in three broad classes, namely: fixed-base network (the workhorses), wireless base network (the hummingbirds), and wireless sensor network (the butterflies). Speed and power requirements for devices in these three classes span six orders of magnitude. In this paper, we analyze commonalities and differences between these three classes of computational architectures, and moving from the analysis of representative state-of-the-art devices, we survey design trends directions of research.