Resource utilization in data intensive embedded systems
Thomas Gustafsson, from Jonköping Univ/Linköping Univ
- Date and Time
Wednesday, October 21st, 2009 at 13.15
Polacksbaken, room 1212
In many data intensive embedded systems, it is important to use available resources efficiently. In studying an engine control system, data freshness and data similarity were evaluated in context of CPU utilization. Data similarity and its implications on resource utilization, workload determination, and admission control will be discussed. Results show that CPU utilization can be reduced compared to approaches where time is used to measure data freshness. Workload determination centers around determining mean interarrival time of updates of data items. Estimation techniques have been evaluated. Admission control is developed based on data usage in engine control and adapted for data similarity.
Resource utilization in terms of software development efforts is also discussed. Software product lines is a concept aiming at efficiently reusing artifacts derived during development. Feature modeling is an approach to model the variability points in software, and can thus be used to reason about software versions with different functionality. Reasoning about software on the feature model level will be discussed.