Hybrid Trust Model for Internet Routing

TL;DR

This paper proposes a novel trust model for BGP routing that combines voting and recommendation to improve detection of malicious or unreliable routers, enhancing Internet security.

Contribution

It introduces a new trust model integrating voting and recommendation mechanisms into BGP, with simulation results demonstrating its effectiveness in identifying distrusted paths.

Findings

Effective detection of distrusted paths beyond immediate neighbors

Identification of malicious neighbors advertising trusted paths

Simulation shows improved trust assessment in autonomous system networks

Abstract

The current Internet is based on a fundamental assumption of reliability and good intent among actors in the network. Unfortunately, unreliable and malicious behaviour is becoming a major obstacle for Internet communication. In order to improve the trustworthiness and reliability of the network infrastructure, we propose a novel trust model to be incorporated into BGP routing. In our approach, trust model is defined by combining voting and recommendation to direct trust estimation for neighbour routers located in different autonomous systems. We illustrate the impact of our approach with cases that demonstrate the indication of distrusted paths beyond the nearest neighbours and the detection of a distrusted neighbour advertising a trusted path. We simulated the impact of weighting voted and direct trust in a rectangular grid of 15*15 nodes (autonomous systems) with a randomly connected topology.