Anomaly Detection with Linear Regression in IP Multimedia Subsystem

The IP Multimedia Subsystem (IMS) refers to the standard for a telecommunication system that controls multimedia services accessing different networks. Not only the IMS but other telecommunications networks are also required to operate continuously. In addition, any system error can lead to massive losses in revenue and productivity. Therefore, operations, administration, and management (OAM) task has an important role in the IMS network, and it requires an efficient method to detect any inherent errors and failures. To facilitate this task, a machine learning approach is used to detect the anomalies of the IMS. This approach is based on Linear Regression, an algorithm that provides a linear relationship between variables. If some variables are not commensurate with the operation of the network, the centralized OAM server will raise some alarms to notify the network's administrators. As a result, the administrator can rely on this model because it reduces the time and labour of the OAM tasks. Moreover, this method can be applied to other systems that contain many subsystems, which makes it difficult to examine all the connections and statuses of these subsystems.