Abstract:Intelligent Transportation Systems (ITS) leverage Integrated Sensing and Communications (ISAC) to enhance data exchange between vehicles and infrastructure in the Internet of Vehicles (IoV). This integration inevitably increases computing demands, risking real-time system stability. Vehicle Edge Computing (VEC) addresses this by offloading tasks to Road Side Unit (RSU), ensuring timely services. Our previous work FLSimCo algorithm, which uses local resources for Federated Self-Supervised Learning (SSL), though vehicles often can't complete all iterations task. Our improved algorithm offloads partial task to RSU and optimizes energy consumption by adjusting transmission power, CPU frequency, and task assignment ratios, balancing local and RSU-based training. Meanwhile, setting an offloading threshold further prevents inefficiencies. Simulation results show that the enhanced algorithm reduces energy consumption, improves offloading efficiency and the accuracy of Federated SSL.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to effectively offload tasks and allocate resources through vehicle - edge computing (VEC) in intelligent transportation systems (ITS), in order to reduce energy consumption, improve offloading efficiency, and enhance the accuracy of federated self - supervised learning (Federated SSL) in the Internet of Vehicles (IoV) environment supported by integrated sensing and communication (ISAC). Specifically, the paper focuses on the following points: 1. **Excessive computational burden**: With the increase in data exchange requirements in intelligent transportation systems, the pressure on vehicle - local computing resources increases, which may lead to a decline in real - time processing capabilities. The paper proposes to reduce the computational burden on vehicles by offloading some computational tasks to roadside units (RSU). 2. **Uneven resource allocation**: If the computing resources of RSU are not properly allocated, it may not be able to meet the task offloading requests of a large number of vehicles during peak hours. The paper explores how to maximize the utilization rate of a single RSU by rationally allocating the computing resources of RSU to meet the offloading needs of more vehicles. 3. **Optimization in a dynamic environment**: Since vehicles are in motion, network conditions and computational tasks will keep changing, and an algorithm that can adapt to these changes is required. The paper uses the deep reinforcement learning (DRL) method to dynamically adjust the transmission power, CPU frequency, and task allocation ratio to optimize the overall energy efficiency and task offloading efficiency of the system. 4. **Offloading threshold mechanism**: In order to further optimize the offloading efficiency, the paper introduces an offloading threshold mechanism to quickly respond according to different network conditions and computational tasks, ensuring the high efficiency of task offloading. In summary, the main objective of this paper is to propose an efficient task offloading and resource allocation method by combining federated self - supervised learning and deep reinforcement learning, so as to improve the computational efficiency and system stability in the Internet of Vehicles environment.

DRL-Based Federated Self-Supervised Learning for Task Offloading and Resource Allocation in ISAC-Enabled Vehicle Edge Computing

Deep Reinforcement Learning-Based Task Offloading for Vehicular Edge Computing With Flexible RSU-RSU Cooperation

Adaptive Task Offloading in Vehicular Edge Computing Networks: a Reinforcement Learning Based Scheme

DRL-Based Resource Allocation for Motion Blur Resistant Federated Self-Supervised Learning in IoV

Federated Deep Reinforcement Learning-Based Task Allocation in Vehicular Fog Computing.

Adaptive Task Offloading in Vehicular Edge Computing Networks Based on Deep Reinforcement Learning

Federated Deep Reinforcement Learning Based Task Offloading with Power Control in Vehicular Edge Computing

Deep Reinforcement Learning Based Vehicle Selection for Asynchronous Federated Learning Enabled Vehicular Edge Computing

Actor-Critic Based DRL Algorithm for Task Offloading Performance Optimization in Vehicle Edge Computing

DRL-based Task and Computational Offloading for Internet of Vehicles in Decentralized Computing

CU-DRL: A Novel Deep Reinforcement Learning-assisted Offloading Scheme for Supporting Vehicular Edge Computing

Joint offloading decision and resource allocation in vehicular edge computing networks

Cloud-Edge-End Collaborative Task Offloading in Vehicular Edge Networks: A Multi-Layer Deep Reinforcement Learning Approach

Self-Learning Based Computation Offloading for Internet of Vehicles: Model and Algorithm

Federated deep reinforcement learning-based online task offloading and resource allocation in harsh mobile edge computing environment

Cloud-Edge–End Collaborative Task Offloading in Vehicular Edge Networks: A Multilayer Deep Reinforcement Learning Approach

IoT workload offloading efficient intelligent transport system in federated ACNN integrated cooperated edge-cloud networks

Big Data Driven Computing Offloading Scheme with Driverless Vehicles Assistance

Joint Task Offloading and Resource Allocation for Vehicular Edge Computing Based on V2I and V2V Modes

An Efficient Partial Task Offloading and Resource Allocation Scheme for Vehicular Edge Computing in a Dynamic Environment

A Multi-Layer Deep Reinforcement Learning Approach for Joint Task Offloading and Scheduling in Vehicular Edge Networks.