Car: Cloud ran assisted-network realization for enhanced vehicular communication

Abstract

The rapid evolution of wireless communication has enabled new possibilities for connected and autonomous vehicles (CAVs), positioning vehicular communication as a key enabler for Intelligent Transportation Systems (ITS). Among the various modes, Vehicle-to-Infrastructure (V2I) communication plays a critical role in ensuring real-time data exchange between vehicles and roadside units (RSUs) to improve road safety, traffic efficiency, and driving automation. However, achieving seamless V2I connectivity under highly dynamic and mobile environments presents significant challenges, including frequent handoffs, connection breaks, and fluctuating channel conditions. Traditional wireless technologies and network architectures fall short of fulfilling the stringent demands for ultra-high reliability, low latency, and high data throughput required by next-generation vehicular networks. Therefore, leveraging higher frequency bands such as millimeter-wave (mm-wave) and terahertz-wave (THz-wave), along with advanced radio access network (RAN) architectures, becomes essential for the future of V2I communication. In this thesis, we first propose a Traffic-Aware Hybrid CRAN Scheme (TRASH) to address the mobility-induced challenges in vehicular networks. The TRASH framework intelligently hybridizes the usage of mm-wave and micro-wave spectrum within a sub-RAN architecture, where Macro-RRHs (MRs) and Pico-RRHs (PRs) are strategically deployed to adapt to varying vehicular traffic densities. MRs are responsible for wide-area coverage, while PRs are activated in dense zones to enhance localized communication. Through comprehensive simulations, it is demonstrated that TRASH significantly improves energy efficiency, CoMP utilization, and coverage probability compared to conventional CRAN-based vehicular systems, particularly under dynamic traffic conditions. Building upon the traffic-aware design, we propose a hybrid network architecture on the basis of speed, that integrates mm-wave and micro-wave communications to further enhance V2I performance. In low-speed scenarios, vehicles are served through CoMP transmissions from multiple Access Points (APs), leading to improved signal strength and reduced outage probability. In high-speed scenarios, vehicles are dynamically connected using selection combining, choosing between Macro-Stations (MSs) and APs based on link quality. This dual approach effectively minimizes outage and enhances the rate performance. The simulation results, supported by analytical derivations, validate the superiority of the proposed hybrid scheme in maintaining reliable communication for both low and high mobility conditions. To exploit the benefits of even higher frequency bands, the thesis further introduces a novel CRAN-based hybrid architecture integrating sub-6 GHz bands via Macro-RRHs and THz frequencies via Terahertz Stations (TSs). The network distinguishes between dense and sparse traffic zones, dynamically allocating resources to maximize system performance. Analytical closed-form expressions for coverage probability are derived for both zone. Simulation results reveal that the THz communication can significantly improve coverage. Next, we find that achieving optimal performance of the handoff-aware rate necessitates determining the optimal density of TSs, as it plays a critical role in maintaining coverage. These findings underline the importance of optimal network planning for THz-based vehicular deployments. Recognizing the vulnerability of THz-enabled networks to inter-THz station (TSs) interference, we propose a reinforcement learning (RL)-based interference mitigation framework for high-mobility vehicular environments. In the proposed scheme, a central processor (CP) within the CRAN architecture dynamically assigns optimal interference weights through a convex optimization-driven RL strategy. This allows real-time adjustment of CoMP transmission parameters, suppressing unwanted interference from non-serving TSs. Analytical models for outage probability and handoff-aware rate are developed under the RL-assisted CoMP framework, and simulation results confirm substantial improvements over conventional CoMP approaches. This work highlights the scalability and adaptability of RL techniques in managing the complexity of future vehicular communication networks. In conclusion, this thesis presents a comprehensive study aimed at enhancing V2I communication performance through the integration of spectrum hybridization, advanced RAN architectures, and reinforcement learning-based optimization strategies. Firstly, to address the dynamic variations in high-mobility vehicular environments, we propose a traffic-aware TRASH framework. Building on this, we integrate mm-wave and micro-wave technologies and analyze critical mobility metrics such as connection break probability and handoff rate. We then introduce a THz hybridization approach to enable high-data-rate communication, followed by the development of a reinforcement learning-based interference control mechanism to further enhance network reliability. Together, these four contributions offer a robust solution to the challenges of vehicular communication in dynamic scenarios, paving the way for the deployment of next-generation intelligent transportation systems that support seamless, reliable, and high-capacity vehicular connectivity in real-world conditions.