Tongxin xuebao (Apr 2021)
Recovery mechanism of large-scale damaged edge computing network in industrial Internet of things
Abstract
Given the limited resources at early stages for recovery, a failure recovery mechanism of the edge computing network considering both computational demands and repair costs was proposed, which intends to tackle the problem of the high probability of large-scale cascading failure caused by the interdependence between the edge computing network and other subnetworks in industrial Internet of things (IIoT).Considering the network structure (topology and link capacity) and network dynamics (computational demands), a joint link recovery selection and computation migration optimization problem was formulated under the conservation of node computing requirements.By leveraging the Benders decomposition algorithm, the NP-hard problem was transformed into a main problem and a sub-problem, which were interdependent and could be solved in polynomial time through the approximation of cutting planes.A local branching method was further introduced to guarantee the non-increasing nature of the Benders upper bound, thus accelerating the convergence of Benders decomposition.Simulation results demonstrate that the proposed algorithm outperforms the conventional topology-based recovery algorithm in system utility, and can perform well in multiple scenarios.
