A Network Simulator as a Service for Demanding Performance Evaluation of Wireless and IoT Scenarios

This paper presents the design, development and evaluation of a novel Network Simulator as a Service infrastructure. It is based on a prominent open-source discrete-event network simulator which is extended to offer advanced batch simulation execution, results acquisition, post-processing analysis and presentation features, as well as machine learning optimisations to reduce the overall processing load. In that respect, we have developed a software system, which allows researchers to define simulation scenarios along with a set of parameters and to commission simulation executions in large scale thus delivering to the end user a service that is novel, robust, extendible and extensible. It has been thoroughly evaluated within ESDA Lab (UoP), proving its feasibility and demonstrating its ability to leverage machine learning techniques in order to automatically identify and exclude simulations that are classified as ones that will not meet defined performance requirements, thus drastically reducing simulation workload, delay and resource utilisation.