GUARD: an intrusion detection framework for routing protocols in multi-hop wireless networks

• Published in: Wireless networks Vol. 25; no. 2; pp. 819 - 836
• Main Authors: Thivakaran, T. K., Sakthivel, T.
• Format: Journal Article
• Language: English
• Published: New York Springer US 15.02.2019
• Subjects: Communications Engineering; Computer Communication Networks; Electrical Engineering; Engineering; IT in Business; Networks; Fuzzy q-learning; Watchdogs; Multi-hop wireless networks; Intrusion detection system; Routing attacks; Non-cooperative game theory
• ISSN: 1022-0038; 1572-8196
• DOI: 10.1007/s11276-017-1594-y

The Multihop Wireless Networks have received great attention in recent years, owing to the rapid proliferation of wireless devices. The wireless routing protocols assume that the nodes are cooperating and well-behaving. However, such networks are subject to several active routing attacks such as dropping, flooding, and modification. The primary intention of such attack is to thwart the objectives of routing protocols and cause network malfunction. This state of affairs motivates the recent research towards the development of a sophisticated security framework that works well against active routing attacks. This paper proposes GUARD, an intrusion detection framework that aims at detecting the active routing attacks efficiently with a considerable reduction in energy consumption. The GUARD incorporates three mechanisms such as a Restricted Directional Watchdog Selection, a Game Design, and an Incentive Estimation. Applying the Restricted Directional Watchdog Selection makes the GUARD a lightweight intrusion detection system (IDS) model, where only a limited number of nodes turn on IDS. The GUARD utilizes non-cooperative game design and exploits the advantage of Fuzzy q-learning to determine the malicious activity. The notion of Fuzzy q-learning determines the results of the player strategies of game design to measure the appropriate reward. The GUARD framework is incorporated into the popular protocols such as AODV and LEACH to validate the effectiveness of the defense mechanism. The simulation results show that these extended protocols outperform the existing protocols in terms of attack detection accuracy, throughput, delay, and network lifetime.