Research on Energy Consumption Optimization Using a Lyapunov-Based LSTM-PSO Algorithm

As the Internet of Things (IoT) proliferates, there’s a growing demand for efficient task processing and top-tier service. This paper introduces a Lyapunov-based LSTM-PSO (LLPSO) predictive offloading algorithm, tailor-made to optimize both user task and computation node resource allocation in rechargeable networks. Simulations reveal that our method eclipses conventional strategies, including the original, First-Come-First-Serve (FCFS), and Greedy algorithms, in terms of energy optimization. Notably, our approach adeptly predicts user task attributes and offloads tasks based on these insights, promoting a more judicious resource allocation. This not only trims down energy expenditure but also amplifies system performance. Taking into account the unique characteristics of rechargeable networks, we leverage Lyapunov functions for dynamic resource tuning, effectively curbing energy usage and bolstering resource efficiency. Experimental benchmarks underscore our algorithm’s edge in harmonizing computation node load, augmenting energy conservation, and refining task processing efficiency. In essence, our LLPSO predictive offloading algorithm stands out as a pivotal tool for elevating resource allocation prowess and user experience within IoT frameworks.