Utility-Based Resource Allocation for Multi-Stream Video Analytics: Experimental Scripts and Results

This Zenodo record provides the experimental scripts and summarized results used in the comparative evaluation of utility-based resource allocation strategies for multi-stream video analytics systems. The dataset includes profiling outputs from YOLO-based object detection models under multiple bitrate, resolution, and frame-rate configurations, as well as allocation results produced by Alpha Fairness–based, Markov Chain–based, and Lyapunov drift-plus-penalty optimization frameworks. The resources are intended to support reproducibility and further research on accuracy–bitrate–utility trade-offs in bandwidth-constrained video analytics, particularly for smart city and edge computing scenarios.