An efficient fog-based attack detection using ensemble of MOA-WMA for internet of medical things
Smart healthcare applications such as smart fitness, smart watches, and elderly remote monitoring devices have reduced the load on traditional healthcare organizations and led to healthier lifestyles. Nevertheless, these devices are under high risk of zero-day attacks because of their fragile nature...
Saved in:
Main Authors: | , , |
---|---|
Format: | Book Section |
Published: |
Springer, Cham
2020
|
Subjects: | |
Online Access: | http://eprints.utm.my/id/eprint/92906/ http://dx.doi.org/10.1007/978-3-030-70713-2_70 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Universiti Teknologi Malaysia |
Summary: | Smart healthcare applications such as smart fitness, smart watches, and elderly remote monitoring devices have reduced the load on traditional healthcare organizations and led to healthier lifestyles. Nevertheless, these devices are under high risk of zero-day attacks because of their fragile nature and network characteristics. Furthermore, few companies developing these devices take care of security requirements. Attack detection using machine learning techniques such as ensemble learning has been adopted to IoT as a second line of defense. However, most of the proposed approaches are either heavy to implement at Internet of Medical Things (IoMT) devices or are cloud based which lead to delay in the detection of attacks. Also, these detections are centralized which are less compatible with IoT. In this study, an ensemble attack detection method is proposed for the detection of stream data attack at fog layer. The base classifiers are stream and incremental based algorithms, which are compatible with IoMT nature and fog devices. A weighted majority algorithm is followed to obtain best accuracy with reduced latency. The results demonstrated that the proposed model is effective for attack detection at fog layer, while it gives better accuracy, higher detection rate and lower false positive rate with average detection time. |
---|