A METHODOLOGY FOR ESTIMATION OF PERFORMANCE IN REAL TIME DISTRIBUTED EMBEDDED SYSTEMS

As embedded systems become increasingly networked, and interact with the physical world, Real Time Distributed Embedded Systems are emerging. This research work mainly focuses on the performance estimation and analysis of Real Time Distributed embedded systems. Performance evaluation is one of the key challenge in the system analysis. Major design parameters that influence the performance includes real-time properties, such as ask execution times, communication delays, the degree of parallelism in computations, and the throughput of the communication architecture. Here the performance estimation is done by discrete event simulation methodology which is based on repetitive simulations of the model, and therefore does not suffer from memory consumption limits. Moreover, it can provide partial results in case the models are too large for exhaustive analysis. They provide the platform for the implementation of cyber-physical systems that run in open environments, in less predictable conditions by providing a highly adaptive infrastructure for reusable resource management services than previous generations of real-time embedded systems that are specialized for specific application domains. The proposed algorithm is scalable for large-scale systems as well, and performs well for finding task mappings that satisfy real-time constraints. Obtaining a feasible mapping for a model of this size will be in the order of seconds. The proposed algorithm is implemented and the results are compared with that of Random Simulations