Article Preview
Top1. Introduction
Since the evolution of wireless technologies and mobile computing hardware, a mobile ad hoc network (MANET) finds extensive usage in many distributed applications. Specifically, many military and disaster response applications deploy a MANET owing to its intrinsic features of no infrastructure requirement, deployment ease and self-organizing nature which contribute to its applicability. Certain other characteristics which include open communication medium involving wire-less links, dynamic signal strength changes causing data transmission fluctuations, dynamic changes in network topology causing broken links due to node mobility and constrained battery power of mobile nodes act as limitations to its applicability and contribute to the security vulnerabilities. Hence the importance associated with its applications along with the vulnerabilities demand a lot of research towards security in MANET and a considerable amount of research has been done already in this field.
A classification of attacks upon MANET based upon the location of the adversary is: Insider attacks and Outsider attacks. The former category of attacks comes up due to the fact that, communication in a MANET happening in multi-hop fashion and latter category arises due to the existence of open wireless medium.
Another possible classification of attacks upon MANET based upon the network elements which experience the impact is: Control plane attacks and Data plane attacks. The former category of attacks has the routing protocol operation as their target whereas the latter category of attacks has the forwarding decisions employed during data transmission as their target. Accordingly, the focus areas of the security mechanisms designed for MANET can be addressing either the control plane or the data plane.
The focus of proposed security mechanism is over the data plane. It is accomplished through a novel uncertainty reasoning-based trust management framework (TMF) and the application of the TMF in the design of a routing protocol to determine a secure route intending for a reliable data transmission from the source to destination. The route established involves the intermediate nodes with best packet forwarding behaviour which is represented by a quantitative metric known as path allegiance metric (PAM). A reinforcement learning approach is employed wherein the non-malicious nodes perform forwarding actions so as to maximize cumulative reward in the form of improved trust throughout the network. Each forwarding action causes a trust update of a node by some of its peers and the updated trust values at the end of a session are utilized to select the path with the highest path allegiance metric. The path allegiance metric for each path is computed during the route discovery phase through the trust metric values of the nodes computed during the earlier sessions.
The paper is organized as follows to appropriately address the different aspects of the research problem. Section 1 presents the introduction including the objectives of the proposed research. Literature survey is presented by section 2 followed by section 3 which provides the contributions of the research work including the design of the trust management framework, trust components computation mechanisms, and the design of the path allegiance metric-based routing protocol. The performance evaluation is provided in section 4 and followed by the presentation of the conclusions and future enhancements in section 5.