A Data Quality-Aware and Adaptive Communication Framework for Vehicular Networks

Danladi Suleman

doi:10.25907/01068

Smart cities continue to attract the attention of researchers, industry leaders, and policymakers. This momentum is driven by the rise of the Internet of Things (IoT) and a wide range of innovative solutions and strategies within digital technology frameworks. Through these technological advancements, cities are developing into dynamic, adaptable, and responsive ecosystems that promote resilience, sustainability, and intelligence across various application domains. The core innovation propelling this digital transformation in the mobility and infrastructure management domain of smart cities is Cooperative Intelligent Transport (C-ITS). The central tenet of C-ITS is to incorporate transport components into a unified communication ecosystem. At the core of the C-ITS concept is Vehicle-to-Infrastructure (V2I) communication, which enables real-time interactions between vehicles and surrounding roadside units (RSUs) to improve road safety and traffic efficiency. However, V2I applications and use cases involve messages with varying contexts and data quality (DQ) needs. This complexity and dynamism, characterised by high mobility, variable node densities, and constantly changing topologies, create significant challenges in ensuring that data-in-motion satisfies the required DQ conditions for efficient and reliable communications. Given the wide range of communication requirements within V2I ecosystems and the diverse capabilities of available communication technologies, one-size-fits-all strategies used in legacy networks are no longer sufficient. This challenge is reflected in the behaviour of existing IoT application-layer protocols (ALPs), which are built on legacy TCP and UDP transport protocols and therefore exhibit varying performance across different IoT environments and dynamic V2I conditions. This thesis examines a range of IoT application-layer and transport-layer protocol combinations within V2I ecosystems. It begins by addressing a critical gap in contextual DQ strategies in smart cities through an in-depth literature review of DQ and information assurance (IA) solutions and then investigates the role of communication protocols in supporting integrated DQ-by-design. Furthermore, the thesis presents a simulation framework that integrates OMNeT++, SUMO, Veins, and OpenStreetMap data and conducts a large-scale performance evaluation of IoT ALPs over their legacy (TCP and UDP) transport protocols and the alternative SCTP and QUIC transport protocols. The results indicate that CoAP over UDP/QUIC and WebSocket over QUIC are promising protocol combinations for disseminating safety awareness messages. Achieving sub-milliseconds latency and a packet delivery ratio (PDR) between 95% and 100% in simulated scenarios. The analysis was further extended to examine the coexistence of safety and non-safety messages under a shared radio spectrum, addressing another gap in the literature. The proposed multichannel architecture supports the concurrent dissemination of safety messages on the control channel (CCH) and non-safety V2I communication on the service channel (SCH) through systematic pairings of IoT ALPs and transport-layer protocols. The simulation results show that QUIC-based protocol combinations efficiently manage data-intensive payloads in shared spectrum environments. In addition, the thesis introduces a novel protocol Fusion Module (FM), guided by a fuzzy-based multi-criteria decision analysis (MCDA) and contextual multi-armed bandit machine-learning models as decision engines (DEs), to address a core gap and challenge the one-size-fit all communication approaches in legacy networks. To assess the feasibility of the proposed framework, extensive performance evaluations were conducted across multiple scenarios; the results indicate its strong potential to optimise critical performance metrics, with all DEs achieving an average accumulated PDR of at least 75% under high vehicle penetration conditions and higher values under low and mild traffic, aggregated over the set of protocol combinations selected in each communication session. The fuzzy MCDA-based DE maintained sub-milliseconds (under 1ms) latency across several simulated scenarios. In summary, this thesis provides a comprehensive evaluation of higher-layer protocols, highlighting their strengths and limitations in supporting dynamic, performance-centric protocol recommendations in real-time V2I communication scenarios. It contributes to the emerging communication paradigm by challenging the static, one-size-fits-all approach to wireless communication in dynamic environments.

A Data Quality-Aware and Adaptive Communication Framework for Vehicular Networks

Files and links (1)

Abstract

Details

Metrics