A Mixture Density Network Approach to Predicting Response Times in Layered Systems

The archives include the research data associated to the paper: A Mixture Density Network Approach to Predicting Response Times in Layered Systems. The paper is in the Proceedings of IEEE MASCOTS 2021. The research data includes five datasets that correspond to the section 6 of the paper: -(1) FCFS_single_class: The response time distribution in single class queueing network with FCFS scheduling discipline. -(2) PS_single_class: The response time distribution in single class queueing network with PS scheduling discipline. -(3) FCFS_multiclasses: The response time distribution in two-class queueing network with FCFS scheduling discipline. -(4) PS_multiclasses: The response time distribution in two-class queueing network with PS scheduling discipline. -(5) Test: The corresponding test pairs for (1)-(4) and the response time distribution of each pair, saved as histogram. The collection of training sets (1)-(4) is divided into subsections and executed by Condor batch-processing system