No abstract

Machine learning techniques are being widely used to develop an intrusion detection system (IDS) for detecting and classifying cyber attacks at the network-level and the host-level in a timely and automatic manner. However, Traditional Intrusion Detection Systems (IDS), based on traditional machine learning methods, lacks reliability and accuracy. Instead of the traditional machine learning used in previous researches, we think deep learning has the potential to perform better in extracting features of massive data considering the massive cyber traffic in real life. Generally Mobile Ad Hoc Networks have given the low physical security for mobile devices, because of the properties such as node mobility, lack of centralized management and limited bandwidth. To tackle these security issues, traditional cryptography schemes can-not completely safeguard MANETs in terms of novel threats and vulnerabilities, thus by applying Deep learning methods techniques in IDS are capable of adapting the dynamic environments of MANETs and enables the system to make decisions on intrusion while continuing to learn about their mobile environment. An IDS in MANET is a sensoring mechanism that monitors nodes and network activities in order to detect malicious actions and malicious attempt performed by Intruders. Recently, multiple deep learning approaches have been proposed to enhance the performance of intrusion detection system. In this paper, we made a systematic comparison of three models, Inceprtion architecture convolutional neural network Inception-CNN, Bidirectional long short-term memory (BLSTM) and deep belief network (DBN) on the deep learning-based intrusion detection systems, using the NSL-KDD dataset containing information about intrusion and regular network connections, the goal is to provide basic guidance on the choice of deep learning methods in MANET.

Preserving the consumed energy of each node for increasing the network lifetime is an important goal that must be considered when developing a routing protocol for wireless sensor networks. The main objective of cluster-based routing protocol is to efficiently maintain the energy consumption of sensor nodes by involving them in multi-hop communication within a cluster and by performing data aggregation and fusion in order to reduce the number of transmitted messages to the base station (sink) and transmission distance of sensor nodes. In this paper we propose a new approach called (DB-SEP) which cluster heads are selected on the basis of their initial energy and their distances between them and the sink. Experimental results show that our approach performs better than the other distributed clustering protocols such as SEP in terms of energy efficiency and lifetime of the network.

In Intelligent Transport Systems (ITS), Vehicular Ad-hoc Networks (VANET) play an essential role in improving road safety and traffic efficiency. Nevertheless, due to its special characteristics like high mobility, large size of the network and dynamic topology make routing of data in the vehicular ad hoc network more challenging. The problem in these networks is to determine the routing protocol best suited to this environment, and then secure it to provide optimal and secure routing for the data. Recently, position-based routing protocol has been developed by researchers to be a very interesting routing technique for communication between vehicles. In this paper, we propose an secured and enhanced version of the Greedy Perimeter Stateless Routing (GPSR) protocol. This protocol consists of two modules: (i) To implement an improvement of GPSR routing protocol which minimizes transfer delays and control messages. (ii) To deal with security issues, we have proposed a solution that combines between an improved Diffie-Hellman algorithm for reliable key exchange and the hash function based Message Authentication Code (MAC) for the verification of the authentication and integrity of GPSR packet. The proposed solution was checked by the security protocol verification tool, Automated Validation of Internet Security Protocols and Applications (AVISPA), which indicated that it is a very secure level. Simulation results showed that our proposed compared to the original GPSR offers better performances.

No abstract