Throughput Maximization for Result Multicasting by Admitting Delay-Aware Tasks in MEC Networks for High-Speed Railways

The rapid expansion of high-speed railways (HSRs) and the growing demand for diverse data services during long journeys require efficient computing services. Mobile Edge Computing (MEC) emerged as a promising platform to fulfill this demand. We envision a scenario wherein passengers interact with each other on the same or different trains in real-time by offloading computationally intensive and delay-sensitive tasks to the track-side MEC networks for HSRs and computation results are multicast to the receivers. To improve the quality of data services, we propose a novel approach to optimize network throughput by admitting as many tasks as possible, subject to delay constraints, and multicasting the maximum number of results. The high mobility of trains and the frequent handovers during train-ground communication are factored into our scheme, which presents significant challenges to jointly consider the dynamic multicast grouping and admission/rejection policies for tasks/results. We introduce the multi-group-shared Group Steiner tree (GST) model and propose an efficient heuristic algorithm that reduces the multicast routing problem to finding a GST for each candidate cloudlet. The effectiveness of our proposed algorithm is demonstrated through simulations and the results are promising.