CENS Technical Seminar Series

CENS Technical Seminar Series - Roy Shea

Invited Speaker: Roy Shea
Date: July 17, 2009
Time: 1:00 PM - 2:00 PM
Venue: Boelter Hall 4760

Abstract

Severe resource constraints in bottom tier distributed
embedded wireless systems hinder attempts to understand runtime behavior of deployed systems. Detailed diagnostic logs can provide the required insight, but must be obtained without violating bandwidth limitations, timing constraints, and other challenges of the target domain. This talk introduces the Log Instrumentation Specification (LIS) language that is used to describe focused logging tasks and drive the instrumentation of source code to collect the requested logs. The LIS language is easy for developers to use directly in their daily work flow and expressive enough to act as an intermediate language targeted by higher level analysis. The runtime logging support needed by LIS fits comfortably on low end sensing devices, allowing LIS to be applied to a wide range of systems. We will examine recent applications of LIS to diagnose problems in sensor networks and observe general runtime behavior. The end of the talk will focus on two new techniques for gathering call traces. The proposed techniques, local identifier logging and control flow logging, significantly reduce the bandwidth required to gather call traces compared to the standard call tracing technique currently used within the embedded sensor network community.

Biography

Roy Shea is a Ph.D. Candidate in Computer Science in the Networked
Embedded Systems Laboratory (NESL) and the Center for Embedded Networked Sensing (CENS) at the University of California, Los Angeles. Roy received a joint B.S. degree in Mathematics and Computer Science from from Harvey Mudd College (HMC) and his M.S. degree in Computer Science from UCLA. His graduate career has focused on research for the sensor network domain and spanned from designing dynamic operating systems targeting bottom tier sensor network devices to exploring new techniques to identify and trace faults in embedded systems.