DDoS prevention architecture using anomaly detection in fog-empowered networks

Abstract

The world of computation has shown wide variety of wonders in the past decade with Internet of Things (IoT) being one of the most promising technology. Emergence of IoT brings a lot of good to the technology pool with its capability to provide intelligent services to the users. With ease to use, IoT is backed by a strong Cloud based infrastructure which allows the sensory IoT devices to perform specific functions. Important features of cloud are its reliability and security where the latter must be dealt with proper care. Cloud centric systems are susceptible to Denial of Service (DoS) attacks wherein the cloud server is subjected to an overwhelming number of incoming requests by a malicious device. If the same attack is carried out by a network of devices such as IoT devices then it becomes a Distributed DoS (DDoS) attack. A DDoS attack may render the server useless for a long period of time causing the services to crash due to extensive load. This paper proposes a lightweight, efficient and robust method for DDoS attack by detecting the compromised node connected to the Fog node or edge devices before it reaches the cloud by taking advantage of the Fog layer and prevent it from harming any information recorded or from increasing the unnecessary traffic in a network. The chosen technology stack consists of languages and frameworks which allow proposed approach to works in real time complexity for faster execution and is flexible enough to work on low level systems such as the Fog nodes. The proposed approach uses mathematical models for forecasting data points and therefore does not rely on a computationally heavy approach such as neural networks for predicting the expected values. This approach can be easily modelled into the firmware of the system and can help make cloud services more reliable by cutting off rogue nodes that try to attack the cloud at any given point of time.

ADSBAN: Anomaly detection system for body area networks utilizing IoT and machine learning

“…The number of sensors depends upon the vital signs of the human body to be monitored and is generally proportional to the complexity of the monitoring system. 6 Possible anomalies in body area networks: A typical BAN may contain various types of anomalies 7,8 as mentioned below.…”

“…Possible anomalies in body area networks : A typical BAN may contain various types of anomalies 7,8 as mentioned below. Point or global anomaly: It can be defined as an individual anomaly that may originate from a single source of data points generated from the sensors or devices.…”