Anomaly detection and self-healing in industrial wireless networks

A self‐healing block in self‐organizing network consists of two modules, namely cell outage detection and cell outage compensation (COC). This chapter presents a data‐driven analytics framework for autonomous outage detection and coverage optimization in an LTE network that exploits the minimization of drive test functionality as specified by 3GPP in Release 10. The outage detection approach first learns a normal profile of the network behaviour by projecting the network measurements to a low‐dimensional space. For this purpose, the multi‐dimensional scaling method in conjunction with domain and density based detection models, one class support vector machine based detector and local outlier factor based detector, respectively, are examined for different network conditions. The low‐dimensional representation of network measurements facilitates data modelling and allows the anomaly detection algorithms to obtain a better estimation of data density. To optimize the coverage and capacity of the identified outage zone, a fuzzy based reinforcement learning algorithm for COC is proposed