Jamming Classification Algorithms in Mobile Ad-Hoc Network

The latest technologies in Mobile Ad-Hoc Network (MANET) allows networking infrastructure to be set up quickly and easily. Every node in MANET plays a role as a router that creates a sustainability access regardless of time or location. On the other hand, a study found that eighty one percent of attacks occurs at the physical and MAC layers of the protocol stack. Knowing the types and characteristics of jammers will enable researchers to develop defense techniques that can prevent jammer attacks. Thus, this paper proposes the development of algorithms to classify jamming attacks using a set of metrics on the physical layer and MAC in MANET. The methodology consists of four main stages. The first stage is to apply reverse engineering method to obtain the specific patterns of individual jammers. This creates the jamming classification metrics with threshold values. The second stage is to classify individual jammers according to the specific pattern and characteristics as defined in jamming classification metrics. The third stage involves development of Max-Min Rule-Based Classification Algorithm. Final stage is testing stage where accuracy test was conducted to evaluate the effectiveness of proposed algorithms based on classier model using SVM and WEKA. Results show that data with normalization and scaling for polynomial kernel with cross validation presented the highest accuracy assessment which is 76.142% compared to other classifiers. Therefore, jammer attack classification study will be provided with better knowledge and technology to develop more robust de-fence techniques in MANET.

I. Introduction

There are many types of wireless networking and one of the fastest growing, easy to use and popular among users is the Mobile Ad-hoc NETwork (MANET). It is based on a self-organizing and rapidly deployed network. However, the popularity of MANET or IEEE802.11a/b/g/n poses potential security issues in which an attacker can exploit and jam networks. Researchers had worked hard to solve jamming related issues by proposing new techniques and models but most of their methodologies have several weaknesses such as wastage of bandwidth due to the need to broadcast the routing tables/updates, inability to monitor and detect jamming based protocol attack at MAC layer, and limited methods to classify intrusion and flexibility of the framework [1]–[3]. Steps such as jamming detection and mitigation have been proposed but an important technique has been skipped, which is the classification stage. The best technique is to detect and classify jammers based on jammer models to protect the network. However, there is limited study on classification of jammers and this knowledge helps mitigation developers to design better prevention methods. This is due to the fact that authors [1]–[3] concentrated on detection and mitigation instead of classification of attack. The best technique is to detect and classify jammers based on jammer models to protect the network. However, there is limited study on classification of jammers and this knowledge helps mitigation developers to design better prevention methods. This is because not many classification methods are found in the literature to identify specific types of jammers. In addition, the uniqueness parameter needs to be identified and tested to develop efficient prevention techniques.