ADAPTIVE FILTERING AND DYNAMIC COMPUTATION OFFLOADING FOR RESILIENT TASK EXECUTION IN IIOT

Relevance. The execution of time-sensitive tasks in Industrial Internet of Things (IIoT) systems requires decentralized and fault-tolerant computing models that can operate under dynamic workloads, unstable node availability, and limited resources. Object of research: task management and offloading processes in edge–fog IIoT environments. Purpose of the article. Development of a method for decentralized, latency-aware task processing using adaptive filtering and dynamic computation offloading, which ensures high availability, efficient resource usage, and responsiveness in distributed edge systems. Research results. The paper presents the AFDCO method, which integrates local data filtering, node availability evaluation, latency-based offloading, and lightweight redundancy into a unified framework. This approach enables autonomous and fault-resilient task execution without relying on centralized controllers. Simulation results confirm that AFDCO reduces response time, improves task deadline compliance, and minimizes network and energy overhead. Conclusions. Compared to static or centralized task allocation models, the proposed method demonstrates better adaptability and robustness under variable conditions by dynamically adjusting execution and replication strategies based on system feedback. Scope of application of the obtained results: distributed IIoT systems, edge–fog computing platforms, and lowlatency industrial automation scenarios requiring decentralized execution and fault recovery.