Joint optimisation of time and energy consumption for data aggregation in fog-enabled IoT networks

Abstract

Fog computing extends cloud capabilities to the network edge, enabling Internet-of-Things (IoT) devices to offload computation to nearby fog nodes rather than a remote cloud. Offloading aggregation tasks reduces data redundancy and accelerates analytics while easing device energy use and backhaul load. Yet end-to-end completion time—comprising execution, transmission, and queueing—can still be substantial, creating a challenging time-energy trade-off. We formulate data-aggregation offloading as a multi-objective optimization problem that jointly minimizes latency (makespan) and energy under compute and bandwidth constraints. To solve it, we develop an NSGA-III-based method that searches for Pareto-optimal offloading and scheduling decisions across sensor and fog nodes. Comprehensive simulations and systematic experiments demonstrate that our approach consistently outperforms state-of-the-art baselines, delivering lower latency and energy consumption with better scalability.