SeAC: SDN-Enabled Adaptive Clustering Technique for Social-Aware Internet of Vehicles

Aamir Akbar, Muhammad Ibrar, Mian Ahmad Jan, Lei Wang, Nadir Shah, Houbing Herbert Song

Abstract

As the Internet of Vehicles (IoV) evolves into the Social Internet of Vehicles (SIoV), vehicles are increasingly treated as social entities forming relationships based on shared interests and ownership. However, maintaining stable network clusters in such highly dynamic environments remains a significant challenge. This paper presents SeAC (SDN-Enabled Adaptive Clustering), a novel technique that leverages Software-Defined Networking (SDN) and a heuristic 3D modeling approach to balance physical proximity with social compatibility. By mapping vehicles into a logical 3D space defined by location, social ties, and interest similarity, SeAC optimizes cluster formation to enhance stability and reduce network overhead. Simulation results in diverse urban settings show that SeAC significantly improves cluster lifetime and reduces both packet loss and end-to-end delay.

Introduction

The Social Internet of Vehicles (SIoV) represents a paradigm shift where vehicles exchange data not just as sensors, but as social nodes with specific interests—such as traffic updates, infotainment, or shared ownership. Traditional clustering methods often struggle in these environments because they typically prioritize either physical distance or social ties, but rarely both effectively. This leads to frequent cluster reformations and high communication overhead. SeAC addresses these issues by centralizing cluster management through an SDN controller and employing a multi-dimensional optimization strategy.

System Architecture and Methodology

The SeAC framework is organized into a three-layer architecture designed to decouple data forwarding from control logic, thereby improving network flexibility and efficiency.

The three-layer SDN-enabled architecture of SeAC, comprising Perception, Control, and Application layers.
  1. Perception Layer: Consists of vehicles and Roadside Units (RSUs) that collect real-time data on location, velocity, and social attributes.
  2. Control Layer: An SDN controller maintains a global view of the network. It processes incoming data from the perception layer to calculate vehicle trajectories and manage network topology.
  3. Application Layer: This layer hosts the SeAC algorithm, which performs the core clustering logic.

The 3D Modeling Innovation

To solve the stability problem, SeAC maps each vehicle into a logical 3D space based on three parameters: Physical Location ($P_L$), Social Ties ($S_L$), and Interest Similarity ($I_S$). By treating these factors as dimensions, the algorithm divides the space into geometric "octants." Vehicles falling within the same octant are both physically close and socially compatible, making them ideal candidates for a stable cluster. Since finding the optimal cluster configuration is NP-hard, SeAC utilizes a heuristic approach to provide real-time solutions. Additionally, SeAC uses hash functions to create "Logical Addresses" (LAs) based on this 3D model, allowing for relationship-based data routing.

Experimental Evaluation

The performance of SeAC was evaluated using a combination of SUMO for traffic simulation and OMNeT++ with the INET framework for network modeling. The researchers tested the algorithm on two distinct real-world road topologies imported from OpenStreetMap:

  • Downtown New York (USA): Characterized by high vehicle density and complex grid-like junctions.
  • Islamabad (Pakistan): A more open road network with different traffic flow dynamics.

Key Results and Discussion

The simulations compared SeAC against a baseline clustering approach that lacks the 3D multi-parameter optimization.

Cluster Stability

A primary metric for success in SIoV is cluster lifetime. As shown in the results, SeAC consistently produced more stable clusters across varying vehicle speeds and densities. Even at high velocities (up to 600 km/h in simulation), the 3D-enabled approach maintained significantly longer cluster durations than the baseline.

Mean cluster lifetime comparisons in Downtown NY and Islamabad across varying vehicle velocities and densities.

Network Performance

  • Reduced Latency: SeAC demonstrated lower end-to-end delay. Although the centralized SDN processing introduces some overhead, it is far outweighed by the reduction in delays typically caused by frequent cluster re-elections in decentralized systems.
  • Lower Packet Loss: The increased stability of SeAC clusters directly translates to more reliable data delivery. In high-density scenarios (e.g., 120 vehicles), SeAC maintained a much lower packet loss rate, as vehicles were less likely to unexpectedly drop out of their assigned groups.

Conclusion and Future Work

SeAC offers a robust solution for the SIoV by integrating social awareness into physical network management via SDN. While the current model significantly improves stability and performance, future research will focus on enhancing security against malicious nodes and exploring distributed SDN architectures to prevent the central controller from becoming a bottleneck in massive-scale deployments.

Summary updated

Abstract

Since millions of smart vehicles in Internet-of-Vehicles (IoV) produce and relay data to analyze road conditions, creating social networks of vehicles in IoV is an important factor for the future Intelligent Transportation System (ITS). Likewise, the IoV architecture has seen vertical fragmentation of approaches used to meet the needs of different work domains. Therefore, IoV in combination with social networking, called Social IoV (SIoV), was created to address these alleged problems. However, one of the challenges in SIoV is that the social relations between vehicles grow and deplete very fast due to the extremely dynamic and unstable nature of the IoV. Therefore, a clustering-based scheme for SIoV, which is efficient in terms of stability can overcome this problem. We propose SeAC: an SDN-enabled adaptive clustering technique for SIoV. SeAC uses a 3D modeling approach to construct logical clusters that are based on factors such as physical location, social tie, and interest similarity among vehicles. Therefore, SeAC improves the stability of clusters and the efficiency of the underlying SIoV architecture. Additionally, by minimizing the trade-off between social and physical distances, SeAC lowers communication and computation costs. We evaluate SeAC, and the simulation results show that for two different topologies, the adaptive approach using SeAC can produce better results in terms of a stable cluster formation.
BibTeX
@article{akbar2023seac,
  author = {Aamir Akbar and Muhammad Ibrar and Mian Ahmad Jan and Lei Wang and Nadir Shah and Houbing Herbert Song},
  title = {{SeAC: SDN-Enabled Adaptive Clustering Technique for Social-Aware Internet of Vehicles}},
  journal = {IEEE Trans. ITS},
  year = {2023},
  doi = {10.1109/tits.2023.3237321},
}
Figure