Computer Science · Computer Vision and Pattern Recognition
Atlas- Architecture for Sensor Network Based Intelligent Environments
Abdelsalam (sumi Helal, Hen-i Yang, Jeffrey King, Raja
See on the mapAbstract
Moving beyond the current generation of pervasive computing spaces (known as integrated environments) requires a middleware that can automate interconnection and integration tasks and ensure the pervasive space’s openness and extensibility. The required middleware must also enable programmers to develop device-based applications dynamically without having to interact with or fully understand the physical world of sensors and actuators. Based on lessons learnt from early prototyping of an Assistive Home Environment for the elderly we have arrived at a set of high-level goals and functional requirements for such a middleware. We have fully implemented the Atlas middleware and have deployed it in several real-life settings including a 2,500 sq ft. house. In this paper, we report on the goals and requirements that influenced the design and implementation of Atlas. We present the Atlas architecture, components and programmer API. We also present several case studies of actual Atlas deployments. Finally, we present a performance evaluation of Atlas to quantify its merits and potential.