HAMLET: Heterogeneous Adaptive Mapping and Low-Energy Task Scheduling for Heterogeneous Multicore IoT Devices

The rapid growth of Internet of Things (IoT) deployments and the increasing integration of multiple cores on a single chip have made managing power consumption and ensuring thermal safety critical challenges for multicore IoT devices. This paper proposes HAMLET, a heterogeneous adaptive mapping and low-energy task scheduling framework that jointly manages energy consumption, timing constraints, thermal safety, and reliability in multicore IoT platforms through a multi-objective genetic optimization approach. In the offline phase, task-to-core mapping and replication decisions are optimized to minimize peak power and energy consumption while preserving schedulability and reliability targets. Moreover, predictive temperature control is achieved in the runtime phase using a Long Short-Term Memory (LSTM) model, which anticipates core temperature evolution and enables proactive control actions. This approach ensures that tasks are allocated to processing cores to prevent thermal violations while maintaining system-level reliability. Furthermore, DVFS and DPM mechanisms are adaptively applied at runtime based on predicted thermal states to reduce energy consumption and avoid thermal emergencies. ... mehrExperimental results demonstrate that the proposed method achieves up to 84.39% reduction in peak power and up to 81.98% reduction in energy consumption, while improving schedulability by up to 20.7% compared to state-of-the-art techniques, confirming its effectiveness for energy- and reliability-aware multicore IoT systems.