A Light weight Intrusion Detection System for Mobile Ad hoc Network

N. Ravi and Dr.G. Ramachandran

Mobile Adhoc Network (MANET) has a several unique characteristics when compared to other wireless networks. Recent advancement in MANET leads to various open vulnerabilities, which can be easily exploited. Further, this exploitation leads the entire network vulnerable and susceptible to malicious attacks. Security in MANET is absolute and it is near impossible to provide the complete security architecture for the MANET. Data security schemes, which involves a strong cryptography alone, are not sufficient to protect the attacks against MANET. Moreover, usage of cryptography in the MANET requires an effective infra and this may suffer due to wide usage of computational power among devices. In order to address the above-mentioned issues, this paper employs a novel lightweight intrusion detection system to detect only Layer 2 attacks in MANET. Layer 2 attacks are severe attacks, which focuses only on Layer 2 transactions. These attacks include DDoS, DoS, replay etc. The key idea of the proposed WIDS is to effectively utilize the hybrid genetic algorithm called Backward Q-Learning based State-Action-Reward-State-Action (SARSA) to model the entire state transition during attack and normal states and validate the same using Carl Pearson Correlation scheme. The proposed WIDS is experimentally tested using Kali Linux with 32 GB RAM, Intel Xeon, Octa-core processor.

Volume 11 | 04-Special Issue

Pages: 634-640