Urban traffic congestion and inefficient parking management have emerged as major challenges in modern smart cities due to the rapid growth in the number of vehicles and increasing urban population density. In many metropolitan regions, drivers spend a considerable amount of time searching for vacant parking spaces, which leads to excessive fuel consumption, increased traffic congestion, environmental pollution, and driver frustration. Conventional parking systems primarily rely on manual supervision or sensor-based infrastructure, which often involves high operational costs, limited scalability, delayed response times, and inefficient utilisation of parking resources. These limitations highlight the urgent need for intelligent, automated, and real-time parking management solutions capable of supporting future smart transportation environments.
This paper proposes a novel Smart Edge-based Vehicle Parking System (SE-VPS) that integrates Convolutional Neural Network (CNN)-based image classification, Vehicular Ad-Hoc Networks (VANETs), and edge computing technologies to develop an efficient and intelligent parking management framework. The proposed system utilizes surveillance cameras installed in parking zones to continuously capture parking area images. These images are processed locally using an edge server equipped with a trained CNN model capable of accurately detecting parking occupancy and classifying vehicles into categories such as Light Motor Vehicles (LMVs) and Heavy Motor Vehicles (HMVs). Based on vehicle classification and parking availability, the system allocates appropriate parking slots dynamically and efficiently.
The proposed architecture consists of several interconnected components including Parking Side Units (PSUs), Road Side Units (RSUs), On-Board Units (OBUs), edge servers, and a centralized Traffic Management Bureau (TMB). PSUs monitor parking slot occupancy and communicate parking status information to the edge server. RSUs facilitate Vehicle-to-Infrastructure (V2I) communication by transmitting parking updates, congestion information, and routing guidance to nearby vehicles. OBUs installed within vehicles receive real-time parking information and assist drivers in navigating toward available parking slots. The Traffic Management Bureau supervises system coordination, authentication mechanisms, congestion monitoring, and communication management to ensure secure and efficient system operation.
The integration of VANET communication with edge computing enables low-latency information exchange and real-time decision-making within highly dynamic vehicular environments. Unlike conventional cloud-based parking systems, the proposed edge-based framework performs image processing and parking occupancy analysis locally, thereby reducing communication delays, bandwidth consumption, and dependency on centralized cloud infrastructure. This localized processing significantly improves response time and system scalability, making the framework suitable for large-scale smart city deployments.
The CNN model employed in the proposed system is trained using parking lot image datasets under varying environmental and lighting conditions to ensure robust performance and high classification accuracy. Experimental analysis of the proposed framework targets vehicle classification accuracy above 96%, precision exceeding 95%, and image processing time below 50 milliseconds per frame. These performance metrics demonstrate the capability of the system to provide fast and reliable parking management services in real-time urban traffic environments.
The proposed SE-VPS framework aims to minimize driver parking search time, reduce unnecessary vehicle movement, decrease fuel consumption, and lower carbon emissions associated with urban traffic congestion. Additionally, the intelligent parking allocation mechanism improves parking space utilization and enhances the overall efficiency of urban transportation infrastructure. The proposed system contributes toward the development of sustainable smart city ecosystems by integrating artificial intelligence, edge computing, and vehicular communication technologies into a unified intelligent parking management solution.
Smart Parking, VANET, CNN, Intelligent Transportation System, Edge Computing, IoT, Vehicle Classification.
Mohit Sanan, Dr. Karthik Kovuri. CNN-Based Smart Parking System Using VANET: A Novel Intelligent Transportation System Framework. Indian Journal of Modern Research and Reviews. 2026; 4(5):279-285
Download PDF
