You are on page 1of 1

Hybrid Intrusion Detection System for Internet of Things

(IoT)
Dr. S. Smys, Dr. Abul Basar, Dr. Haoxiang Wang
Sarah Salameh , Reem Al-Sahli , Nouf essa

Abstract Methodology Evaluation

The Internet of things (IoT) is a promising This paper suggested that the security mechanism of The proposed intrusion detection system is
solution to connect and access every device the Internet of Things environment should focus on experimentally verified and compared with RNN
through the internet. IoT provides numerous preventing unauthorized access to the data source. It based intrusion detection system. By comparing the
services and applications, however, it faces also focuses on data confidentiality and integrity. In proposed model and RNN model-based intrusion
severe security issues and is vulnerable to addition to identifying security threats in the shortest detection systems. It is observed from the values, the
possible time. The paper also suggested the use of a proposed model attains better detection performance
attacks. Intrusion detection system is used to
centralized intrusion detection system in order to over RNN model. Few parameters such as recall and
detect such attacks when the network security is reduce the complexity of the intrusion detection
breached. This research work proposed an precision are similar to RNN however the ration of
process. This system monitors the network and remote true positive and false positive is much better than
intrusion detection system for IoT networks to nodes to identify intrusions. So, an alert is triggered to RNN model. By comparing detection accuracy of
detect different types of attacks based on hybrid the network administrator to respond to the security the proposed mode with the RNN model, the
convolutional neural network model. Proposed issues. Figure 1 depicts an illustration of intrusion proposed hybrid convolutional neural network
model is suitable for wide range of IoT detection system. model achieves a better efficiency of 98% which is
applications. Proposed research work is 3% greater than the traditional recurrent neural
validated and compared with the conventional network model.
model. Experimental result demonstrate that
proposed hybrid model is more sensitive to Conclusion
attacks in the IoT network.

Motivation Intrusion detection systems are an inevitable


Fig. 1 Intrusion detection system
processing unit in recent wireless networks. The
Hybrid IDS was used in this paper as it provides a proposed model in this paper has been
Since the IoT environment is not limited to one better trade-off between storage and computing cost experimentally verified and compared with the
specific goal, it is used for multiple goals and with fewer false positives. This proposed system is traditional recurrent neural network, and the results
services, and it is necessary to provide a safe built on the analysis of problems in existing intrusion show that it achieves a better detection accuracy of
and convenient environment for the user. It has detection systems. The proposed model is divided into 98%, which makes the application suitable for
become necessary to provide a separate module four stages: data collection, data pre-processing, different IoT environments.
to ensure the IoT network security this is network training, and attack identification. During
because the Internet of Things (IoT) these phases, trained data and information actually References
environment is vulnerable to various types of collected are processed together, and the training
attacks. Which leads to the conclusion that it is period is used to detect attacks. Figure 2 depicts the overall 1. Snehal Deshmukh-Bhosale, Santosh S. Sonavane (2019). A Real-Time Intrusion
Detection System for Wormhole Attack in the RPL based Internet of Things.
process flow of proposed intrusion detection system. Procedia Manufacturing. 32:840-847.
necessary to we have to use an Intrusion 2. Rup Kumar Deka, Dhruba Kumar Bhattacharyya, Jugal Kumar Kalita (2019).
Active learning to detect DDoS attack using ranked features. Computer
detection system that is used to detect attacks Communications. 145:209-222
when the network security is breached. 3. Shailendra Rathore, Jong Hyuk Park (2018). Semi-supervised learning based
distributed attack detection framework for IoT. Applied Soft Computing. 72:79-
Therefore, this paper proposed a model that is 89.
4. Alekha Kumar Mishra, Asis Kumar Tripathy, Deepak Puthal, Laurence T. Yang
more sensitive to attacks and can be applied to a (2019). Analytical Model for Sybil Attack Phases in Internet of Things. IEEE
Internet of Things Journal. 6(1): 379-387
wide range of IoT applications. 5. Guangjie Han, Xun Li, Jinfang Jiang, Lei Shu, Jaime Lloret (2015). Intrusion
Detection Algorithm Based on Neighbor Information Against Sinkhole Attack in
Fig. 2 Proposed Intrusion Detection System Wireless Sensor Networks. The Computer Journal, 58(6): 1280-1292

You might also like