Fine Grained Analysis of Packet Loss in MANETs in NS2

Abstract:

Most existing trust-based security schemes for MANETs consider packet loss an indicator of possible attacks by malicious nodes. There may be several reasons for packet losses, such as interference, queue overflow, and node mobility. Identifying the real underlying cause of a packet loss event is important for any security solution. To detect truly malicious nodes, it is necessary to carry out a fine-grained analysis to determine the underlying cause of such loss. Without such analysis, the performance of any security solution may degrade, due to the punishment of innocent nodes while actual malicious nodes may remain undetected. Therefore, approaches are required that can correctly identify the reason of packet losses and can react accordingly. In this paper, we present a scheme able to correctly identify malicious nodes, using network parameters to determine whether packet losses are due to queue overflows or node mobility in MANETs. The contributions of this paper include a fine-grained analysis (FGA) scheme for packet loss and the development of a comprehensive trust model for malicious node identification and isolation. Our proposed FGA scheme is evaluated in terms of effectiveness and performance metrics under different network parameters and configurations. The experimental results show that our proposed trust model achieves a significant reduction in false positives rate and an increase in the rate of detection of truly malicious nodes as compared to traditional non-FGA schemes.