Federated learning-based smart transportation solutions: Deploying lightweight models on edge devices in the internet of vehicles

Presently, the Internet of Vehicles (IoV) promises a revolution in the smart transportation sector. The concept has become an edge computing centralization or distributed transport 5.0 standard. It has massive technological advancements in hardware, tiny- to large-scale integration, and modular application program interface-based software that enables seamless data interaction and relationships between vehicles, checkpoint segments, and cloud platforms. It has an abundance of troubles with data security, limited bandwidth, and computing depth driven by the tremendous quantity of data that IoV devices bring in. Addressed the safety mechanisms of the social application and the on-load, off-load, and live database retrieval problems identified. The next step is encouraging ground-aware strategies and capable modes of transport throughout the IoV ecosystem regions, like cell- or cell-free technology developments. This research effort evaluates the use of lightweight models set up on edge devices for Federated Learning (FL), the distributed algorithm for learning approach. Federated computing secures confidential information and eliminates overhead for communication by allowing team training of statistical models across great independent electronics while disclosing the data. Federated computing protects confidential information and eliminates communication overhead by allowing group training of statistical models across multiple independent electronics while disclosing data. The work focused on integrating FL with lightweight models targeted at edge devices with limited assets, such as roadside assistance devices (RSUs or RSDs) and on-board gadgets (OBUs or OBGs). Lightweight models meet an acceptable compromise between intellectual loads and operational efficiency through a software-defined network of tactics like quantization, elimination, and modeling deformation. That makes them ideal for edge distribution of inside and outside vehicle connections. The book chapter looks into the theoretical foundations of FL and lightweight model fashion, pointing out their own distinctive strengths, edge devices, and complications. This further leads to effective state timeslot priority, strategies for modes of transportation featuring lane utilization enhancement, cognitive vehicle guidance, automatic upkeep, ecological tracking, and participatory or sensory. Analyses with daily life scenarios illustrate how placing FL-based application delivery network on edge gadgets in the IoV is achievable and accurate. Those insights enable the design of effective, safe, and resilient modes of transport by dealing with concerns regarding data privacy, utilizing resources, and promoting intelligent decision-making. Practical boundaries and possible avenues for study are finally outlined, establishing the way for the widespread adoption of these advances in static and dynamic mobility approaches and accelerating the journey to greater intelligence and connected towns.