Intelligent Task Offloading in VANETs: A Hybrid AI-Driven Approach for Low-Latency and Energy Efficiency

Tariq Qayyum 1 , Asadullah Tariq 1 , Muhammad Ali 2 , Mohamed Adel Serhani 3 , Zouheir Trabelsi 1 , Maite López-Sánchez 4 · 2025

Read Intelligent Task Offloading in VANETs: A Hybrid AI-Driven Approach for Low-Latency and Energy Efficiency by Tariq Qayyum 1 , Asadullah Tariq 1 , Muhammad Ali 2 , Mohamed Adel Serhani 3 , Zouheir Trabelsi 1 , Maite López-Sánchez 4 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-20735.txt.

Introduction

Tariq Qayyum, Asadullah Tariq, Muhammad Ali, and three co-authors tune computing for connected vehicles in this two thousand twenty-five paper. Vehicles in ad-hoc networks can offload work to roadside units and edge servers for real-time processing. But the setting is harsh and live: terms shift in ways no one can call, bringing latency, wasted energy, and failed tasks. The team puts forth an AI frame to call what to offload, and where.

Supervised learning models call offloading plans from current terms. Reinforcement learning adapts the calls as the network shifts. Particle swarm optimization tunes the trade between latency and energy use. Wide runs showed cuts in latency and energy use, and task success rates and network throughput beat the ways set off against them too. The team sets the frame out as a scalable base for vehicular uses. They run from safety systems to traffic intelligence.

Catalog record

CREATORTariq Qayyum 1 , Asadullah Tariq 1 , Muhammad Ali 2 , Mohamed Adel Serhani 3 , Zouheir Trabelsi 1 , Maite López-Sánchez 4
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-20735

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Intelligent Task Offloading in VANETs: A Hybrid AI-Driven Approach for Low-Latency and Energy Efficiency Tariq Qayyum 1 , Asadullah Tariq 1 , Muhammad Ali 2 , Mohamed Adel Serhani 3 , Zouheir Trabelsi 1 , Maite López-Sánchez 4 Abstract Vehicular Ad-hoc Networks (VANETs) are integral to intelligent transportation systems, enabling vehicles to offload computational tasks to nearby roadside units (RSUs) and mobile edge computing (MEC) servers for real-time processing. However, the highly dynamic nature of VANETs introduces challenges, such as unpredictable network conditions, high latency, energy inefficiency, and task failure. This research addresses these issues by proposing a hybrid AI framework that integrates supervised learning, reinforcement learning, and Particle Swarm Optimization (PSO) for intelligent task offloading and resource allocation. The framework leverages supervised models for predicting optimal offloading strategies, reinforcement learning for adaptive decision-making, and PSO for optimizing latency and energy consumption. Extensive simulations demonstrate that the proposed framework achieves significant reductions in latency and energy usage while improving task success rates and network throughput. By offering an efficient, and scalable solution, this framework sets the foundation for enhancing real-time applications in dynamic vehicular environments. I Introduction Vehicular Ad-hoc Networks (VANETs) play a crucial role in modern intelligent transportation systems by enabling vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication. These networks enhance road safety, traffic efficiency, and autonomous driving by facilitating real-time data exchange. However, the increasing reliance on advanced sensors, cameras, and computational modules in vehicles has led to a surge in data processing demands [ 1 ] . Many applications, including collision avoidance, traffic monitoring, and in-vehicle infotainment, generate significant computational workloads that exceed the processing capabilities of individual vehicles. Computation offloading, wherein tasks are dynamically allocated to Roadside Units (RSUs), Mobile Edge Computing (MEC) servers, or cloud infrastructures, emerges as a viable solution to mitigate computational constraints and improve service responsiveness [ 2 ] . Despite its potential, offloading decisions in VANETs present significant challenges due to their highly dynamic and decentralized nature. Rapid topology changes, intermittent connectivity, variable computational loads, and stringent latency requirements complicate the decision-making process. Traditional rule-based or heuristic offloading strategies often struggle to adapt to real-time traffic variations, resulting in increased latency, suboptimal resource utilization, and inefficient energy consumption. Therefore, an intelligent and adaptive mechanism is required to balance local execution and offloading, considering factors such as network conditions, vehicle mobility, and available computational resources [ 3 ] . Artificial Intelligence (AI) is a powerful tool for managing task offloading in VANETs, leveraging data to predict network conditions and make adaptive decisions. Hybrid frameworks combining supervised learning, deep learning, and reinforcement learning enhance offloading efficiency. Supervised models like Random Forest predict feasibility, while CNNs and LSTMs analyze traffic patterns. RL dynamically learns optimal policies, and heuristic techniques such as PSO optimize resource allocation to reduce latency and energy consumption [ 4 ] , [ 5 ] . This research introduces a novel hybrid AI framework that addresses the limitations of traditional offloading approaches. By integrating predictive analytics, real-time adaptability, and heuristic optimization, the proposed framework enhances computational efficiency, reduces latency, and balances energy consumption. Furthermore, the incorporation of federated learning principles allows AI models to be collaboratively trained across multiple vehicles, improving decision-making while preserving data privacy and reducing communication overhead. Contributions: The key contributions of this research are: (1) the development of a predictive offloading model that dynamically determines optimal execution strategies based on real-time traffic conditions and resource availability, (2) an RL-based adaptive offloading mechanism that continuously learns optimal policies through real-world traffic simulations, and (3) the application of bio-inspired optimization techniques to enhance resource allocation efficiency. Extensive simulations demonstrate that the proposed framework significantly improves latency reduction, energy efficiency, and network throughput compared to conventional offloading strategies. By addressing the inherent challenges of task offloading in dynamic vehicular environments, this research advances AI-driven decision-making for scalable and intelligent VANET architectures. The proposed framework provides a roadmap for the integration of AI-based solutions in next-generation vehicular networks, paving the way for more efficient and autonomous transportation systems. Figure 1 shows the proposed system architecture that includes AI based task offloading leveraging PSO and RL. Fig. 1: Overview of the proposed system for intelligent task offloading in VANETs. The system integrates reinforcement learning and Particle Swarm Optimization (PSO) for optimal task distribution between vehicles and RSUs, ensuring low latency and energy efficiency. II Related Work In this section, we present a state-of-the-art review of AI-based approaches for task offloading in VANETs. Firdose et al. [ 6 ] proposed a comprehensive survey addressing task offloading challenges in Edge and Cloud Computing. The study highlights the limitations of centralized Cloud systems for delay-sensitive applications and emphasizes the potential of Edge Computing to mitigate these issues. Yen et al. [ 7 ] focuses on a game-theoretic approach and adaptive type selection for resource sharing, which may not effectively handle the high variability in VANET environments. Unlike our proposed hybrid AI framework, which integrates supervised learning, reinforcement learning, and PSO for intelligent task offloading, the existing model lacks a comprehensive AI-driven optimization mechanism, leading to potential inefficiencies in latency reduction, energy optimization, and adaptive decision-making in dynamic vehicular scenarios. Han et al. [ 8 ] proposed an intelligent computational offloading scheme, CODIA, to address the NP-hard problem of dependent task offloading in dynamic IoT environments. CODIA employs a prioritized subtask scheduling strategy and an Actor-Critic-based reinforcement learning algorithm for efficient decision-making. It transforms task dependency into device state transitions, ensuring low latency and energy efficiency. Guanjin et al. [ 9 ] proposed the Deep Meta Reinforcement Learning-based Offloading (DMRO) framework, which integrates DRL and meta-learning for efficient IoT task offloading in dynamic MEC environments. DMRO achieves faster adaptation, strong portability, and 17.6% better performance over traditional DRL algorithms. Ying et al. [ 10 ] propose the Efficient Multi-Vehicle Task Offloading (EMT) algorithm for 6G networks using Cybertwin-enabled MEC. EMT minimizes system costs and queue lengths, balancing performance with dynamic green energy utilization. Kai et al. [ 11 ] introduces a task offloading framework in heterogeneous vehicular networks, leveraging V2X communication technologies and federated Q-learning to minimize resource costs and offloading failure probabilities, ensuring optimal system performance. Moreover, authors in [ 12 ] propose an efficient SDN-assisted forwarding strategy within NDN-based IoV, designed to support vehicular mobility while leveraging cellular networks to ensure low-latency control message delivery Trabelsi et al. [ 13 ] propose a fuzzy-based V2V offloading method for low-latency, priority-aware IoV tasks. Their approach optimizes task scheduling, ensuring reduced system delay and efficient resource utilization. Ai et al. [ 14 ] proposed a hybrid optimization framework using reinforcement learning and deep reinforcement learning to minimize power consumption in MEC systems while ensuring low-latency task offloading and resource allocation. III System Model Vehicular Ad-hoc Networks (VANETs) have emerged as a cornerstone of modern intelligent transportation systems, enabling vehicles to communicate with each other and with roadside infrastructure to enhance road safety, traffic efficiency, and autonomous driving capabilities. As vehicles are increasingly equipped with advanced sensors, cameras, and computational modules, the demand for real-time processing of massive amounts of data has surged. Applications such as autonomous driving, collision avoidance, traffic monitoring, and in-vehicle infotainment generate extensive computational workloads that often exceed the limited processing capabilities of individual

Read the complete text