Conference paper
Stochastic Modeling and Performance Analysis of Multimedia SoCs
Reliability and flexibility are among the key required features of a framework used to model a system. Existing approaches to design resource-constrained, soft-real time systems either provide guarantees for output quality or account for loss in the system, but not both. We propose two independent solutions where each modeling technique has both the above mentioned characteristics.
We present a probabilistic analytical framework and a statistical model checking approach to design system-on-chips for low-cost multimedia systems. We apply the modeling techniques to size the output buffer in a video decoder. The results shows that, for our stochastic design metric, the analytical framework upper bounds (and relatively accurate) compare to the statistical model checking technique.
Also, we observed significant reduction in resource usage (such as output buffer size) with tolerable loss in output quality.
Language: | English |
---|---|
Publisher: | IEEE |
Year: | 2013 |
Pages: | 145-154 |
Proceedings: | 2013 International Conference on Embedded Computer Systems: Architectures, Modeling, and Simulation (SAMOS XIII)International Conference on Embedded Computer Systems: Architectures, MOdeling, and Simulation |
ISBN: | 1479901032 , 1479901040 , 9781479901036 and 9781479901043 |
Types: | Conference paper |
DOI: | 10.1109/SAMOS.2013.6621117 |
Analytical models Calculus Model checking Real-time systems Stochastic processes Streaming media System-on-chip analytical framework upper bounds formal verification logic design low-cost multimedia systems modeling technique multimedia systems output buffer output quality performance evaluation probabilistic analytical framework probability real-time systems resource allocation resource usage reduction resource-constrained soft-real time system design statistical analysis statistical model checking approach stochastic design metric stochastic processes system-on-chip system-on-chip design video coding video decoder