Mapping and Priority Assignment for Real-Time Network-on-chip with Static and Dynamic Applications

Abstract

Networks-On-Chip (NoC) has emerged as a solution for addressing the limitation of bus-based communication in multi-processor system designs. Some of these systems are designed for executing real-time services which are associated with deadlines, failing to meet these deadlines may lead to failure of the system. The problem of Task Mapping and Priority assignment becomes more crucial when multiple applications are targeted on a platform in realtime scenario. Real-time schedulability analysis is performed assuming worst case scenario to get the lower bound of schedulability for tasks and flows. The term schedulability defines the number of tasks and flows meeting the deadlines. A broader range of Embedded System consists of both static and dynamic applications. The resources for static and dynamic applications are to be allocated separately. Considering mixed requirement of static and dynamic application and with major focus on dynamic scenario in real-time system, a novel approach for mapping and flow priority assignment has been reported in this paper. It extends the state of the art works in Deadline and Energy Aware Dynamic Mapping and Scheduling (DEAMS) algorithm with flow priority assignment approach obtained by modifying Graph based Exhaustive Search Algorithm (GESA). Current DEAMS algorithm does not consider the network congestion and flow priorities while mapping the tasks instances. In order to get better schedulability in dynamic scenarios, GESA has been modified to work in dynamic situations, which considers the flow dependencies as well. The graph based algorithm puts check for schedulability in a more deterministic way, making it more appropriate for real-time systems. The motivation behind the work presented in this paper lies in probable application of Cyber Physical Systems, having static and dynamic resource allocation for applications in real-time.