Work-in-Progress: Deadline-Constrained Multi-Resource Allocation in Edge-Cloud System

In an edge-cloud system, end devices can offload computation intensive tasks to servers for processing, to satisfy deadline requirements of time-critical tasks, or maintain a good quality of service. Because the system has limited bandwidth and computation resource, it can be very challenging to determine where tasks should be offloaded and processed (task mapping), and how much bandwidth and computation resource should be allocated to each task (resource allocation). In this paper, we propose a task mapping and multi-resource allocation problem with both communication and computation contentions in an edge-cloud system, which aims to maximize the total profit gained by the system while meeting the deadlines of mapped tasks. Besides, the backhaul network of the proposed edge-cloud system is modeled as a directed incomplete graph with bandwidth contention on every edge of the graph. We formulate the problem into a nonconvex Mixed-Integer Nonlinear Programming (MINLP) problem and provide a linearization method to reformulate the MINLP problem into an Integer Linear Programming (ILP) problem formulation, which can be solved with ILP solvers.