A Two-Stage Task Offloading Algorithm for Cloud-Edge Collaborative Computing

Cloud-edge collaborative computing has become an emerging computing paradigm due to its ability in fully utilizing the computing power for cloud and edge servers. However, in the process of task offloading, it is faced with the problem of long solution time due to high computational complexity and large scale of solving multi-objective optimization problems, which result in excessive cost of time and energy of the local devices. To solve this problem, we design an algorithm with a two-stage task offloading strategy. To be specific, we divide the overall task offloading into two stages. The first stage is to make the decision of allocating each task module to the suitable computational platform by considering each task respectively as a mixed integer linear programming (MILP) problem and solving it by branch and bound algorithm. The second stage is to allocate the selected task nodes at the edge platform reasonably by turning the allocation problem into a knapsack problem, which can be solved by dynamic programming algorithm. In this way, we can not only narrow the problem scale by dividing one large overall constraint model into several smaller ones, but also reduce the computational complexity. Performance evaluation shows that compared with the similar algorithm, our algorithm is optimized by more than 50% in terms of task waiting time and by more than 5% in terms of the optimization effect.