INTELLIGENT CAUSALITY-GUIDED SWARM OPTIMIZATION FRAMEWORK FOR ENERGY-EFFICIENT RESOURCE ALLOCATION IN INDUSTRIAL INTERNET OF THINGS
Shalu Saraswat, Surendra Mahajan, Ritesh V. Patil, Lalit V Patil
Abstract: The availability of energy, diverse resource requirements and demanding latency-reliability requirements are barriers to IIoT deployments in large-scale industrial settings where inexpensive embedded nodes are required to operate continuously with varying workload conditions. An all-encompassing, energy-conscious resource distribution framework which takes into account the sensing, computing, communication, and cloud-edge cooperation is required as the incremental optimizations at individual layers are no longer applicable in order to achieve scalable and reliable IIoT functionalities. A causality-based, swarm-optimal, and stability-enforced Energy-Efficient Resource Allocation Framework of Industrial IoT can be used to deal with these challenges. Task-based energy latency fingerprints are generated by gauging the sensing, processing, and transmission prices of ESP-based IIoT nodes. Using these fingerprints, a Hierarchical Edge Cloud Load Balancing module with Graph-Bandit Offloading (HECLB-GBO) can dynamically assign jobs to device, edge and cloud levels to form a Causal Energy-Delay Attribution Graph (CEDAG) explicitly identifying the main control variables that lead to energy escalation and SLA violations and generate interpretable constraints and optimization weight. The central one is Energy-Efficient Swarm Resource Allocation (EESRA-SO) in which a causality-seeded multi-swarm optimizer adjusts sampling rates, batching windows, scaling CPU frequencies, transmission power, compression ratios as well as duty cycles to the smallest possible energy consumption with strict SLA constraints. Last but not the least, Uncertainty-Gated Federated Policy Distillation offers lightweight, drift-resistant deployment of nodes with limited resources, and Lyapunov-Safe Primal-Dual Resource Control guarantees real-time queue stability, the reduced energy consumption of individual tasks, cloud bandwidth consumption, SLA non-conformance, and enhanced adaptations of both workloads and channels.
Keywords: Energy-Efficient IIoT; Swarm Optimization; Edge–Cloud Resource Allocation; Causal Energy Modeling; ESP-Based Industrial Sensing
DOI: 10.24874/PES08.03.032
Recieved: 29.05.2026 Revised: 15.07.2026 Accepted: 12.08.2026
UDC:
Reads: 2 
