Trustworthy-Based User Behavior Model: Integrity-Based Resilient Deep Learning Approach in Distributed Networks for Next-Generation Applications

The distributed network utilizing the integrity-based resilient machine learning (DRML) approach offers a robust methodology to address challenges associated with complex Internet of Things (IoT) applications, computational resource limitations, and environmental sustainability. In heterogeneous cloud computing environments, where edge and central clouds collaborate to meet diverse demands in various sectors, significant challenges arise in task offloading. This research proposes an IoT and cloud computing framework integrated with a distributed and resilient deep learning (DLRTO) model to optimize system utility and bandwidth allocation for local controller devices (LCDs). The DLRTO model is designed to generate near-optimal offloading decisions for LCDs, edge cloud servers, and central cloud servers, thus maximizing both system utility and bandwidth allocation. Additionally, we introduce the RL-NN integrated round robin-based dynamic task scheduling (RNRRDTS) algorithm on edge and cloud servers to address job scheduling issues, with a focus on renewable energy generation and job migration. Extensive simulations were conducted using real-world IoT application tasks, renewable energy data, and grid electricity pricing data to evaluate the proposed model. The experimental results demonstrate the superior performance of the DLRTO model compared with existing methods. This research provides valuable insights and advances toward the creation of sustainable, resilient, and optimized IoT systems applicable across various domains.