Mining Over Air: Wireless Communication Networks Analytics by Ye Ouyang & Mantian Hu & Alexis Huet & Zhongyuan Li

Author:Ye Ouyang & Mantian Hu & Alexis Huet & Zhongyuan Li
Language: eng
Format: epub
ISBN: 9783319923123
Publisher: Springer International Publishing

According to the results, GPLSA is able to detect anomalies in a time-dependent context. Global outliers (e.g., on Fig. 7.3b at Apr. 15 16:00 in red) as well as context-dependent anomalies (e.g., at Apr. 15 5:00 in orange) are identified. Off-peak periods are taken into consideration and unusual values specific to those periods detected. Gaussian hypothesis on GPLSA is not really constraining. As shown in Fig. 7.3a, clusters are adaptable and try to fit Gaussian distributions. They are appropriate to represent the value distribution for each class of dates and cluster. Cluster adaptation is shown in Fig. 7.3d. The three clusters represent different level of values. The upper cluster represents higher values, which are more probable during peak periods. The lower cluster represents lower values, with a roughly constant probability. The third cluster in the middle is also useful to obtain a good anomaly detection behavior (results with K = 2 clusters are unable to correctly detect anomalies). About anomaly detection itself, a threshold indicating the number of alerts to be detected can be set. This method of detection is static and relatively simple. Improving this method of detection is possible and straightforward through likelihood computations: inside a cell, an anomaly could be detected with a repetition of low likelihood scores.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.