FPGA-Based Symbiotic Evolutionary Swarm Intelligence for Real-Time Dynamic Task Scheduling in Low-Power Consumer Electronics

In modern computing, effective resource utilization and low power consumption are key requirements especially in real-time systems with the demand for timely executions of tasks. By combining the concepts of symbiotic evolution and swarm intelligence, SESI provides optimized task scheduling that adapts dynamically to the fluctuation of workloads and power. This paper proposes a new methodology called SESI Optimization targeting dynamic task scheduling in low-power multi-threaded real-time systems. Applying concepts of swarm intelligence, SESI realizes an efficient way of conducting resource allocation and task scheduling under real-time constraints while embedding good convergence characteristics for task scheduling aiming at the optimization of system performance while keeping the energy consumption low. The approach has been implemented on Kintex UltraScale+ FPGA KCU116 Evaluation Kit, which runs at 100 MHz and uses Vivado Design Suite 2024.1. The tasks are executed using MiBench and Mälardalen WCET benchmarks combined with interleaved pipelining for time predictability guaranteeing worst-case analysis for validation within the PRET community. The performance improvements are statistically significant at p < 0.01. The proposed research demonstrates energy savings up to 65% and a minimum of 20%, as well as 30% savings in power and fitness function by using the proposed SESI Optimization. The proposed SESI exhibits better performance in optimizing dynamic task scheduling for multi-threaded real-time systems and indicating possible application scenarios for resource-constrained real-world systems.