8 April 2025 in IEEE Transactions on Machine Learning in Communications and Networking
Vehicular Ad-hoc Networks (VANETs) have gained significant attention as a key enabler for intelligent transportation systems, facilitating vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications. Despite their potential, VANETs face critical challenges in maintaining reliable end-to-end connectivity due to their highly dynamic topology and sparse node distribution, particularly in areas with limited infrastructure coverage. Addressing these limitations is crucial for advancing the reliability and scalability of VANETs. To bridge these gaps, this work introduces a heterogeneous UAV-aided VANET framework that leverages unmanned aerial vehicles (UAVs) for enhancing data transmission. The key contributions of this paper include: (1) the design of a novel adaptive dual-model routing (ADMR) protocol that operates in two modes: direct vehicle clustering for intra-cluster communication and UAV/RSU-assisted routing for inter-cluster communication; (2) the development of a modified density-based clustering algorithm (MDBSCAN) for dynamic vehicle node clustering; and (3) an improved UAV trajectory planning method based on a multi-agent soft actor-critic (MASAC) deep reinforcement learning algorithm, which optimizes network reachability. Simulation results reveal that the UAV trajectory optimization method achieves higher network reachability ratios compared to existing approaches. Also, the proposed ADMR protocol improves the packet delivery ratio (PDR) while maintaining low end-to-end latency. These findings demonstrate the potential to enhance VANET performance, while also providing valuable insights for the development of intelligent transportation systems and related fields.
1 January 2011 in Transportation Research Record: Journal of the Transportation Research Board
Travel demands are derived from the needs and the desires for participating in various activities, such as work, eating, and shopping. An understanding of the interaction between people's activity and travel choice behavior can serve as a base for long-term transport service planning and policy evaluation purposes. This paper proposes an activity-based network user equilibrium model for dealing with daily activity travel pattern (DATP) scheduling problems in congested networks, in which interdependency between people's activity and travel choices is comprehensively investigated. A diagonalization solution algorithm is developed to solve the scheduling problems. A remarkable merit of the developed solution algorithm is that it enables an automatic generation of individuals’ DATPs, implying obviation of an explicit enumeration of those patterns that often were required in many previous related studies. A numerical example is provided to illustrate the application of the proposed model and solution algorithm.