Modeling and Analysis of Response Time in Vehicular Networks using Markov Chains

Masood Kalamati¹Hasan Raei¹Bulbula Kumeda Kussia²Sadiq Hussain³Emrah Arslan⁴Javad Hassannataj Joloudari⁵Silvia Gaftandzhieva^6*

• ¹Islamic Azad University Mashhad Branch, Mashhad, Iran
• ²Jinka University, Jinka, Ethiopia
• ³Dibrugarh University, Dibrugarh, India
• ⁴KTO Karatay Universitesi, Konya, Türkiye
• ⁵University of Birjand Faculty of Electrical and Computer Engineering, Birjand, Iran
• ⁶Plovdivski universitet Paisij Hilendarski, Plovdiv, Bulgaria

With the rapid evolution of vehicular networks, delay-sensitive applications such as autonomous driving and real-time navigation have gained significant attention. However, vehicles with limited computational resources often fail to meet the low-latency demands of these services, becoming a critical bottleneck in system performance. In this study, we propose the use of Mobile Edge Computing (MEC) to offload time-sensitive tasks to roadside servers, thereby reducing response times and improving overall efficiency through localized processing and network intelligence. To evaluate this approach, we developed a mathematical model based on Markov Chains to capture traffic dynamics and response behaviors in vehicular environments. This model allows system designers and researchers to test different configurations analytically, without relying on time-consuming simulations. By inputting selected parameters, one can predict performance outcomes with reasonable accuracy. Our results demonstrate that the modeled average response times closely match those obtained via traditional simulation methods, validating the practicality and effectiveness of our approach. Beyond vehicular networks, this modeling framework can also support smart technologies in university environments—such as campus mobility solutions, real-time IoT management, and intelligent infrastructure systems. These applications align well with initiatives aimed at advancing smart technology in the university, highlighting the broader potential of our work. Quantitatively, across 20– 120 vehicles, the analytical model estimated average response times in the range 0.023–0.505 s, close to the algorithmic baseline (0.019–0.482 s), with a relative error between 3.4% and 24.0% (mean 11.7%).