Jitter-Awareness User Allocation in Edge-based Intelligent Video Systems via Traffic Simulation

Recently, Edge-based Intelligent Video Systems (EIVSs) have shown great potential to mitigate damages caused by the mobility of traffic participants in urban areas, leveraging the advantage of computing resources close to edge users.In EIVSs, a key challenge is the Edge User Allocation (EUA) problem under limited edge resources.However, existing EUA research overlooks allocation jitter, which can be defined as the consecutive reallocation of users on servers. Consequently, the allocation jitters could undermine server availability, lead to the omission of critical information from edge users, and thereby jeopardize traffic safety.To address this issue, we formulate the Jitter-Awareness Edge User Allocation (JA-EUA) problem, and propose a novel prediction-based method named the Range-variable Jitter Controlled Prediction (RJCP), that incorporates flexible server coverage.Our introduced Urban Traffic Simulation and Prediction Framework (UTSPF) integrates heterogeneous results from RVO and SUMO simulators, validating our approach's capability to mitigate jitter and improve edge camera allocation.