EE7V82 Special Topics In Electrical Engineering - Light-weight Embedded Systems - Spring 2008

Instructor: Roozbeh Jafari
Time: MW 2:30PM-3:45PM

Abstract

Recent technological advances have led to the introduction of a variety of COTS (Commercial-Off-The-Shelf) wireless and embedded platforms. These platforms can measure physical attributes such as temperature and acceleration, perform limited local computation and storage, and communicate within a short range. Distributed embedded system platforms enable ubiquitous presence of sensing, computing and communication capabilities and hence, enable a large number of application domains. In particular, they can be mounted on human body or clothing, or even be woven into the very fabric that we wear to realize various health monitoring applications. We take special interest in such systems, generally referred to as Body Sensor Networks (BSN), due to the unparalleled significance of their application domain and their very specific requirements and implications. Sensor platforms integrated into clothing provide the possibility of enhanced reliability of accident reporting and health monitoring. Such devices improve the independence of people needing living assistance.

This course intends to give students a hands-on experience on design and analysis of light-weight and distributed embedded systems. In addition, the course investigates several real-time signal processing and pattern recognition techniques.

Course Outline

• Introduction to distributed and light-weight embedded systems
  • A survey on existing SW and HW platform
  • Introduction to HW platform
  • Introduction to SW platform (tinyOS components, modules, configurations, interfaces, events and commands)
• Applications
• Sensing
• Computation
• Wireless communication and medium access
• Storage
• Information Theory
• Pattern recognition
  • Segmentation, feature extraction and classification
  • Supervised and unsupervised learning
  • Maximum likelihood estimates, k-nearest neighbor, linear discriminant functions and neural networks
  • Error bounds for normal densities
• Collaborative signal processing
  • Algorithmic techniques

Course Contents (restricted to registered students)