Abstract: This project is about technique or approach in finding anomalies, which represents deviations from expected patterns, can signify critical events of irregularities, malfunctioning of sensors, demanding accurate detection. Internet Of Things (IoT) represents a framework that links physical devices to the internet, allowing them to communicate and exchange data. The quality of IoT services usually depends on the integrity and accuracy of the data. Time series is a common type of data found in everyday situations like traffic flow, network performances, financial records, etc. Detecting anomalies in time series IoT sensor data is very much needed because of the possibility of noise and unavailability of labels in the sensor readings and it’s also an important research topic with practical uses such as spotting intrusions in networks, monitoring traffic, and identifying errors in sensor data. In this project the Inter-Berkeley Research Lab dataset is used for unlabeled anomaly detection technique and UNSW-NB15 IoT weather board sensor dataset is used for labelled anomaly detection, which is suitable for testing and validating different anomaly detection methodologies. This project is proposed to work on hybrid models such as , LSTM – Autoencoder +Isolation Forest, Bi – LSTM + OneClass SVM, an Ensemble model of DBSCAN, LOF, SVM, and a Statistical approach for anomaly detection in IoT sensor Time Series Data, using the results to understand better about the performance of these proposed models.
Keywords: Internet Of Things (IoT),Bidirectional Long Short-Term Memory( Bi-LSTM),One-Class Support Vector Machine( One-Class SVM),Density-Based Spatial Clustering of Applications with Noise.(DBSCAN),Local Outlier Factor (LOF).
Cite:
Shibzan Shahanas, Afnaj Akthar, Saanna Anand, Rakshitha, Dr. Amirthavalli.M, "ANOMALY DETECTION IN TIME SERIES DATA IN IoT ENVIRONMENT", IJARCCE International Journal of Advanced Research in Computer and Communication Engineering, vol. 13, no. 3, 2024, Crossref https://doi.org/10.17148/IJARCCE.2024.13365.
|
DOI:
10.17148/IJARCCE.2024.13365
