NESL Technical Report #: 2010-1-2
Abstract: The ability to deploy a Wireless Embedded System (WES) without preexist- ing infrastructure makes it an attractive solution to sensing tasks. Unfortunately, software defects that survive into deployment are challenging to detect and diagnose due to resource constraints, such as low-bandwidth radio-only communications. This dissertation presents two new techniques that help expose defects within WES software. The first technique casts resource usage into an exclusive resource ownership model. The model provides guidelines within which to reason about resources used by their software. Violations to this model are caught at compile time by static analyses that track resource ownership across software component boundaries. The second technique provides a framework for gather- ing compact runtime logs from deployed WESs. This logging technique honors bandwidth limitations of the WES domain by exploiting program structure to create small token name spaces from which token identifiers are assigned. The small token name spaces facilitate compact identifier assignments that that can be efficiently packed into bit aligned data structures. The resulting logs provide developers with much needed insight into runtime WES behavior.
Publication Forum: Ph.D. Thesis, UCLA CS Department
Public Document?: Yes
NESL Document?: Yes
Document category: Report