Centralized and Distributed Architectures: Approximation of the Response Time in a Video Surveillance System of Road Traffic by Logarithm, Power and Linear Functions

In this article, we propose mathematical models that result in logarithmic, power and linear functions for the predictive evaluation of the response times of video surveillance systems in smart cities. Most of the QoS measurements and evaluations used in the current literature are hardware based and do not take into account the influence of the technical architecture. We have therefore proposed a decomposition process of video surveillance systems to obtain mathematical approximations of each treatment time. The integration of these components guided us towards generic mathematical models validated by experimentation. The measurements resulting from the experiment were approached with logarithmic, power and linear functions before adopting the most precise model between distributed and centralized architectures. The comparison between these architectures shows a much lower response time for the distributed architecture especially with the determination coefficient \(R^2\) of the functions.