Research Projects

Vehicle-to-everything (V2X) communication plays an important role in the intelligent transportation system (ITS), requiring stringent low latency, high reliability, and seamless connectivity. An unexpected Quality of Service (QoS) degradation in the network is one of the critical challenges, where performance parameters such as latency, reliability, throughput, etc. fail to meet required thresholds. This can create challenges including accidents, traffic disruptions, safety concerns, and failure in time-critical applications like autonomous driving, remote surgery, etc. Ultra-reliable low latency communication (URLLC) and xURLLC in 5G and beyond networks promise to achieve 99.999% reliability and 1-millisecond latency. Many existing methods and algorithms focus on scenarios involving homogeneous, traditional, ideal, or standard conditions. However, in real-time, these methods fail, because heterogeneous networks in non-standard conditions pose many challenges including range, connectivity, power levels, communication protocols, reliability, latency, etc. The complexities of these challenges grow again when (x)URLLC stringent requirements are introduced especially in V2X communication. This proposal proposes a novel scheme to provide context-based adaptive QoS support in the 5G and beyond networks for V2X communication, which will be supported by AI, edge devices, drones, and software-defined networks (SDN). The proposed scheme optimizes spectrum and resource allocation using AI and deep learning algorithms in heterogeneous networks operating under non-standard conditions. It efficiently enhances coverage and reliability at the intersection of heterogeneous networks, improving communication latency, seamless connectivity, and reliability. In addition, the integration of EdgeAI and SDN reduces the round trip times and improves decision-making efficiency. The proposed scheme includes two core components: First, a predictive QoS framework uses statistical and deep learning algorithms for analyzing QoS from both Random access network and core network aspects. Second, the EdgeAI-driven adaptive resource allocation framework uses deep-reinforcement learning and federated learning algorithms to analyze, predict, and assign optimal resources to RSUs and drones and preserve data privacy. The project outcomes include EdgeAI models for QoS prediction and adaptive optimal resource allocation, and its testbed for real-time V2X communications. These outcomes significantly impact the ITS with seamless connectivity for autonomous vehicles, reduced road accidents, optimization of traffic flows, and improved response time, reliability, accuracy, etc. The successful real-time development, testing, and deployment of the proposed scheme could result in the next level of intelligent transportation systems, especially, for autonomous vehicles which could have seamless mobility and this can be one of the major objectives of the smart cities in India.