Adaptive Strategy for Optimizing Latency-Energy Tradeoffs in Edge Computing Task Offloading

The rapid growth of big data and the Internet of Things has imposed significant challenges in balancing latency-sensitive and energy-efficient computation for mobile edge devices. This paper addresses the inherent latency-energy tradeoffs in edge computing task offloading by proposing an adaptive optimization strategy. We formulate a bi-objective framework, GPK-MOEA(Guided-Pareto-Knowledge Multi-Objective Evolutionary Algorithm), to simultaneously minimize processing delay and energy consumption, introducing a knowledge-guided evolutionary mechanism that hybridizes historical Pareto solutions with stochastic exploration. A key innovation is an adaptive penalty coefficient mechanism, which dynamically balances the contributions of conflicting objectives without manual weight tuning, ensuring equitable optimization across heterogeneous edge environments. Experimental results demonstrate that the proposed strategy achieves an average reduction of 23.09% in total cost compared to full local execution and 21.28% compared to full offloading, outperforming state-of-the-art algorithms (e.g., 10.47% over MOEA/D) under varying scales of end devices. These findings validate the effectiveness of the adaptive mechanism in optimizing latency-energy tradeoffs for edge computing systems.