A learning-based algorithm for fog computing deployment in IoT network

Abstract

The number of connected Internet of Things (IoT) devices has increasing rapidly due to the benefits and various use cases of IoT, which results in network congestion in traditional cloud. Fog computing has been recognized as a promising technology to meet the requirements of IoT devices by bringing computing, storage, and networking resources to the edge of the network. Fog computing is a dispersed architecture consisting of geographically distributed fog nodes. The selection for locations of fog nodes is an essential precondition for fog computing implementation since an effective fog nodes deployment method can reduce the service response time and improve the efficiency of energy consumption. In this work, regarding the space-time characteristics of sensed data of IoT devices, we formulate the fog nodes deployment as an uncertain programming problem with the aim to reduce the energy consumption of the devices. A learning-based algorithm is proposed to solve this problem with neural network embedded to speed up the solving process. Finally, we evaluate the effectiveness of the algorithm by a set of simulations and the results show the advantages of the algorithm compared to other baseline methods.