|
Research
WCET Analysis
 |
Proposed ESOMICS
We propose a novel scheme called ESOMICS, to obtain the appropriate values of WCETopt for HC tasks to improve QoS and utilization while reducing the number of mode switches. In this scheme, we propose a Machine-Learning (ML)-based approach (which can be generalized to any embedded application) to determine the appropriate value of WCETopt . This scheme evaluates the
model functionality and performance based on the generated data sets to train and validate different prediction techniques. This proposed ML model is utilized in MC systems to obtain WCETopt for any application, to improve resource utilization and QoS, while reducing the number of mode switches, compared to state-of-the-art research works. To the best of our knowledge, this is the first work that obtains WCETopt, utilizing ML models while guaranteeing real-timeliness, improving the QoS, and reducing the mode switches with no timing overhead at run-time.
|
|