Machine Learning Quorum Decider (MLQD) for Large Scale IoT Deployments

The usage of sensors and actuators in industrial applications has increased widely. Cyber Physical Systems (CPS) has improved the usage of closed loop control systems using sensors and actuators. However the health of sensors & actuators and their impact on the efficiency of the industry is still a challenging area. The other important aspect is the network related issues that rise due to the large scale of sensors & actuators. There need an efficient mechanism that manages the network delay and packet drops associated with the large scale sensor deployments. There are machine learning techniques which are used to predict the sensor values when it is absent due to various reasons. Similarly, there are anomaly detection techniques that are used to detect faulty sensors. Both the prediction approaches fail to address the properties related to lifetime and service needs of sensors & actuators. This paper proposes a machine learning framework that combines the results of individual learning modules for achieving collective decisions. This proposed method provides the solution to address the delays and failures occurring in large scale deployments. (C) 2019 The Authors. Published by Elsevier B.V.