Joint task offloading, DNN pruning, and computing resource allocation for fault detection with dynamic constraints in industrial IoT

Niazmand, Vahidreza (2024) Joint task offloading, DNN pruning, and computing resource allocation for fault detection with dynamic constraints in industrial IoT. Masters thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (8MB)

Abstract

In an industrial Internet of Things (IIoT) environment, maintaining high system efficiency and stability is critical to achieving industrial automation. Deep neural networks (DNNs) have been integrated into IIoT systems to improve the intelligence and efficiency of industrial task processing. On the other hand, the execution of DNN model inference for task processing also imposes a significant computation load on end devices (e.g., monitoring sensors). To address the challenges of satisfying stringent task computing/processing requirements (e.g., latency and accuracy) in IIoT environments, offloading tasks to edge servers offers a promising solution. However, solely relying on edge-assisted offloading can introduce prolonged communication delays due to fluctuating wireless channel conditions. To enable efficient processing of a high volume of sensed industrial data for facility fault diagnosis on industrial washing machines, in this thesis, we investigate a joint task offloading, DNN model pruning, and edge computing resource allocation (JOPA) problem under a layered IIoT networking architecture. Specifically, we aim to maximize the overall network resource utilization while guaranteeing diverse and time-varying task processing delays and accuracy requirements for generated processing/computing tasks of the fault detection service. To capture the network dynamics, we formulate a stochastic optimization problem with the objective of maximizing the long-term network resource utilization with per-time-slot constraints on the end-to-end task latency and accuracy. Considering the network state transitions and the relations between network states and policies, we transform our problem into a Markov reward process (MRP) formulation where the state transitions are independent of the actions taken. To deal with the large problem size and dynamic quality-of-service (QoS) constraints, we design a deep-reinforcement-learning (DRL) solution framework based on the soft actor-critic (SAC) algorithm, where the actor networks, critic networks, and target networks are customized to accommodate hybrid actions, achieve robust policy evaluation, and stabilize the training process, respectively. Extensive simulation results are provided to demonstrate the effectiveness of the proposed scheme and the advantages over benchmark approaches in terms of 1) achieving high network resource utilization, 2) balancing the trade-off between resource utilization and QoS satisfaction, and 3) adapting to the network load variation and dynamic QoS requirements.

Actions (login required)

View Item