Distributed IoT Analytics across Edge, Fog and Cloud

Internet of things (IoT) is a computing paradigm that connects devices (things) with the network and each other. The huge amount of data generated by IoT devices makes them promising targets for deep learning and machine learning algorithms required for analysis. The computational and memory costs associated with such algorithms is huge, that makes it prohibitive for deep learning algorithms to be executed on resource constraint IoT devices. In this paper, we define an optimization problem for distributing the deep neural network model across different compute hierarchies such that the communication between the compute hierarchies is minimized. The model is assumed to be a directed acyclic graph (DAG) and the algorithm for splitting the DNN model is based on particle swarm optimization (PSO). PSO splits the DAG across edge (R_e), fog (R_f) and cloud (R_c) in such a way that the communication between the vertices placed on different compute nodes is minimized.