Internet of Things (IoT) is part of a cyber-physical system (CPS)(1) which focuses on automating a job to make things easier for humans. In its development, the IoT system is expected to be a reliable system that makes it easier for users according to the functionality of the system it is building. Most of the goals of developing an IoT system apart from automating human work are to inform important things such as weather predictions and increase security and convenience, for example, in smart home systems (2). Another goal besides this is to increase the efficiency of energy use to reduce carbon gas emissions (3). Unfortunately, not all the purposes of developing an IoT system can be realized. An example is when the weather forecasting system experiences certain problems, this can certainly affect the weather forecasting process. The obstacles that arise allow the existence of abnormal data, which is then called an anomaly.

Data anomalies can occur in IoT systems due to the user’s habitual pattern of using the system (4), relatively low bandwidth, delays, disturbances from the sensor side, and network disturbances (5). These disturbances can also cause a set of data received to experience data loss, making it difficult to find anomalous data. In the data stream type, incomplete data is a common thing to see. Depending on the data type and the studied case conditions, several special methods are applied to complete the data. The anomaly data detection process can occur when the data is complete.

The appearance of anomalous data can certainly affect system performance. In large-scale IoT system implementation, system performance disrupted at one point can affect other sensor nodes nearby. Of the problems that may arise due to anomalies in the data, it is necessary to find out where these abnormalities exist so that system repairs can be done immediately. The search for anomaly data in an incomplete live stream data set can be approximated by the interpolation method, which is synthesized by several methods such as statistics and machine learning.

In detecting anomalies, several researchers have made efforts to detect anomalies (6–8). The anomaly detection process in the CPS domain, including in the network domain (9), IoT (10,11), and other physical computing systems. The models used to detect these anomalies are also quite diverse, starting from the application of statistical or conventional models (12), deep learning models (7), (13,14), to other mathematical models (15). These studies also produce varying results depending on the environmental conditions that are the object of research.

Research carried out by applying deep learning models can have optimal results (16). However, using deep learning in detecting anomalous data requires quite a large amount of computing time in the learning process.

If you look at the data used in existing studies, most of these data are large datasets with the aim of research to test the methods developed. On the other hand, not all studies apply models only to large datasets but to live data (17), which tend to be small. Using live data is challenging in developing an anomaly detection system (18). Live data is susceptible to problems with the data sending and receiving system. Several things, including network interference and errors in the program or sensors, cause this. On the other hand, under normal system conditions, if the network bandwidth is small or the traffic is quite heavy during the data transfer process, this can trigger the same problem. These constraints can impact data loss or delays in data entry to the storage system so that the monitoring system built cannot provide accurate information.

Incomplete data in live data can be said to be anomalous data. On the other hand, incomplete data is not an anomaly but a case of data that needs to be corrected so that the data is complete. These two possibilities depend on the type of case to be studied. If examined more deeply, data loss can not be an anomaly if it occurs in the data transportation flow. When data transmission goes well, it is possible that the condition is normal and no anomalies are found. Therefore, analysis of incomplete data needs to be carried out so that anomalous data can be identified precisely.

If you want to learn more about IoT, you can join the IoT Lab, Faculty of Informatics, Telkom University. There, you can not only find out the theory about IoT, but you can also put it into practice directly, starting with making hardware, building networks, and even sensor monitoring systems.

“Ayo #RaihMasaDepanmu bersama Telkom University!”

Come on #AchieveYourFuture with Telkom University!

References:

1. Ochoa SF, Fortino G, Di Fatta G. Cyber-physical systems, internet of things and big data. Future Generation Computer Systems. 2017 Oct 1;75:82–4.
2. Li Jiang, Da-You Liu, Bo Yang. Smart home research. Proceedings of 2004 International Conference on Machine Learning and Cybernetics (IEEE Cat No04EX826) [Internet]. [cited 2022 Jan 2];2:659–63. Available from: http://ieeexplore.ieee.org/document/1382266/
3. Shrivastava A, Bhardwaj A, Hasteer N. IoT in automobile sector: State of the art. Proceedings of the Confluence 2020 – 10th International Conference on Cloud Computing, Data Science and Engineering. 2020 Jan 1;254–9.
4. Alghayadh F, Debnath D, Alghayadh F, Debnath D. A Hybrid Intrusion Detection System for Smart Home Security Based on Machine Learning and User Behavior. Advances in Internet of Things [Internet]. 2021 Jan 20 [cited 2022 Jan 4];11(1):10–25. Available from: http://www.scirp.org/journal/PaperInformation.aspx?PaperID=106859
5. Effective Anomaly Detection in Sensor Networks Data Streams | IEEE Conference Publication | IEEE Xplore [Internet]. [cited 2022 Jan 4]. Available from: https://ieeexplore.ieee.org/abstract/document/5360301
6. Lindemann B, Maschler B, Sahlab N, Weyrich M. A survey on anomaly detection for technical systems using LSTM networks. Comput Ind. 2021 Oct 1;131:103498.
7. Choi K, Yi J, Park C, Yoon S. Deep Learning for Anomaly Detection in Time-Series Data: Review, Analysis, and Guidelines. IEEE Access. 2021;9:120043–65.
8. Alghanmi N, Alotaibi R, Buhari SM. Machine Learning Approaches for Anomaly Detection in IoT: An Overview and Future Research Directions. Wireless Personal Communications 2021 [Internet]. 2021 Aug 27 [cited 2022 Jan 4];1–16. Available from: https://link.springer.com/article/10.1007/s11277-021-08994-z
9. Amoozegar M, Minaei-Bidgoli B, Rezghi M, Fanaee-T H. Extra-adaptive robust online subspace tracker for anomaly detection from streaming networks. Eng Appl Artif Intell. 2020 Sep 1;94:103741.
10. Fahim M, Sillitti A. An Anomaly Detection Model for Enhancing Energy Management in Smart Buildings. 2018 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids, SmartGridComm 2018. 2018 Dec 24;
11. Violatto G, Pandharipande A, Li S, Schenato L. Anomalous occupancy sensor behavior detection in connected indoor lighting systems. IEEE 5th World Forum on Internet of Things, WF-IoT 2019 – Conference Proceedings. 2019 Apr 1;335–9.
12. Cho SB, Park HJ. Efficient anomaly detection by modeling privilege flows using hidden Markov model. Comput Secur. 2003 Jan 1;22(1):45–55
13. Ahmad Z, Khan AS, Nisar K, Haider I, Hassan R, Haque MR, et al. Anomaly Detection Using Deep Neural Network for IoT Architecture. Applied Sciences 2021, Vol 11, Page 7050 [Internet]. 2021 Jul 30 [cited 2022 Jan 4];11(15):7050. Available from: https://www.mdpi.com/2076-3417/11/15/7050/htm
14. Zhang C, Song D, Chen Y, Feng X, Lumezanu C, Cheng W, et al. A Deep Neural Network for Unsupervised Anomaly Detection and Diagnosis in Multivariate Time Series Data. Proceedings of the AAAI Conference on Artificial Intelligence [Internet]. 2019 Jul 17 [cited 2022 Jan 4];33(01):1409–16. Available from: https://ojs.aaai.org/index.php/AAAI/article/view/3942
15. Liu Y, Chen X, Kadambi D, Bari A, Li X, Hu S, et al. Dependable Visual Light-Based Indoor Localization with Automatic Anomaly Detection for Location-Based Service of Mobile Cyber-Physical Systems. ACM Transactions on Cyber-Physical Systems [Internet]. 2018 Aug 29 [cited 2022 Jan 4];3(1). Available from: https://dl.acm.org/doi/abs/10.1145/3162051
16. Sharma B, Sharma L, Lal C. Anomaly Detection Techniques using Deep Learning in IoT: A Survey. Proceedings of 2019 International Conference on Computational Intelligence and Knowledge Economy, ICCIKE 2019. 2019 Dec 1;146–9.
17. Ariyaluran Habeeb RA, Nasaruddin F, Gani A, Targio Hashem IA, Ahmed E, Imran M. Real-time big data processing for anomaly detection: A Survey. Int J Inf Manage. 2019 Apr 1;45:289–307.
18. Cook AA, Misirli G, Fan Z. Anomaly Detection for IoT Time-Series Data: A Survey. IEEE Internet Things J. 2020 Jul 1;7(7):6481–94.

IoT Lab Locations:
Telkom University Landmark Tower,
Jl. Telecommunications Canal Buah Batu Indonesia 40257, Bandung, Indonesia,
(022) 7566456
https://telkomuniversity.ac.id/