Making Transportation Smarter: BS (Hons) Student Co-authors Chapter on AI-assisted Vehicular Networks

Rapid and reliable data exchange is crucial for developing and operating intelligent transportation systems using Vehicular Ad Hoc Networks (VANETs). However, the dynamic nature of VANETs, marked by high vehicle mobility and constantly changing network structures, pose challenges for traditional communication protocols.

Addressing these challenges, Maanit Shah, a BS (Hons) Class of 2026 student at Ahmedabad University, has authored a book chapter exploring how Reinforcement Learning (RL) can enhance data dissemination in these dynamic networks. His work talks about creating smarter communication systems for vehicles that can adapt instantly to traffic, road conditions, and vehicle movements.

The chapter discussed how existing methods pose a few challenges, namely the complexity of VANETs due to rapid vehicular movements and constantly altering positions and connections, making it difficult for traditional routing protocols to keep pace. Consequently, outdated methods fall short, as current routing approaches often struggle to handle rapid network changes or guarantee high Quality of Service (QoS), essential for timely message delivery. Furthermore, Machine Learning (ML) has limitations as it often relies on historical data. However, VANET's unpredictability also proves that past solutions may not apply to current conditions.

Maanit studied how RL can improve data sharing in intelligent transportation systems, making it more reliable and efficient for critical applications like accident warnings, traffic management, and even self-driving cars. RL stands out because it learns in real-time through continuous interaction with its environment. This capability allows RL-based routing algorithms to adaptively optimise communication decisions by dynamically adjusting to fluctuating network conditions, including vehicle density, mobility patterns, and communication link quality. Essentially, RL enables routing protocols to intelligently respond to constant changes for more efficient data flow.

Maanit Shah's contribution is featured in the book chapter, "Intelligent Data Dissemination in Vehicular Networks: Leveraging Reinforcement Learning," published in the Springer volume Deep Learning Based Solutions for Vehicular Adhoc Networks (Studies in Computational Intelligence, Vol. 1207). He coauthored this work with researchers from Nirma University, the Indian Institute of Technology Gandhinagar, Adani Institute of Infrastructure Engineering, and the team from Sahana System Ltd. To learn more about his work, visit the link