Energy Efficient Task Mapping & Scheduling on Heterogeneous NoC-MPSoCs in IoT Based Smart City

Multi-Processor System-on-Chips (MPSoCs) are extensively deployed in modern Internet-of-Things based Smart City (IoT-SC) applications to fulfill the ever growing computation demands. The Sensor Nodes (SNs) in IoT-SC are energy constrained and normally powered by a battery source with limited residual energy. Therefore, reduction in energy consumption is one of the challenging technological aspect for IoTSC. In this paper we investigate the problem of scheduling set of tasks with precedence and deadline constraints on Networkon-Chip (NoC) based heterogeneous MPSoCs. Unlike other energy-aware scheduling approaches that separately perform task ordering and voltage assignment from the task mapping, our proposed approach deals with it in an integrated way while explicitly considering the contentions between communications. Moreover, our approach shares the available slack between tasks and communications. We have proposed Energy-aware Integrated Task Mapping, Scheduling and Voltage Scaling (EIMSVS) algorithm. The EIMSVS algorithm uses Earliest Latest Finish Time First (ELFTF) strategy to order the tasks and communications in time. At each optimization step EIMSVS algorithm selects a task or a communication to remap it to a processor and or a voltage level that minimizes total energy consumption. The experiments are conducted on synthetic as well as real-world TGs adopted from Embedded Systems Synthesis Benchmarks (E3S). The experimental results are compared with state of the art approach. The results illustrates that our proposed approach achieves average energy improvement and maximum energy improvement of ~ 21% and ~ 59% respectively.