| PhD Seminar

Name of the Speaker: Mr. Krishnakumar G (EE19D410)
Guide: Dr. Venkatesh R
Co-Guide: Dr. Abhishek Sinha
Online meeting link: https://meet.google.com/pct-vwdt-jd
Date/Time: 5th August 2024 (Monday), 3:00 PM
Title: Tandem Queue Decomposition Plus: A Throughput-Optimal Routing Policy for Partially-Trusted-Node Quantum Key Distribution Networks

Abstract :

Quantum Key Distribution (QKD) networks have made significant strides in today's quantum technology landscape, particularly with the commercial implementation of point-to-point QKD networks. However, for long-range QKD networks, the prevailing approach still relies on the traditional "Trusted Node/Relay" setup. Huttner et al. (2022) demonstrated that, given current technological constraints, it is impossible to relax the trusted node assumption. Even our recent work, the Tandem Queue Decomposition Policy (TQD), which addresses secure and throughput-optimal routing in multi-hop QKD networks with generalized flows, adheres to this assumption.

QKD networks ensure information-theoretic security by leveraging the fundamental principles of quantum mechanics. However, trusted nodes within a QKD network, responsible for holding classical messages, represent vulnerabilities that eavesdroppers can exploit if not adequately secured. In multi-hop QKD networks, where nodes cannot be fully trusted, providing an unconditional security guarantee becomes infeasible. Our recent study acknowledges that all network nodes are susceptible to potential attacks by eavesdroppers. We partially trust the nodes and represent their vulnerability by associating a compromise probability to them. An oblivious eavesdropper tries to gain access to the network nodes in an independent and identical manner. Even in this simple setting, the literature lacks throughput-optimal routing policies that can guarantee some form of security.

We propose an online dynamic policy, Tandem Queue Decomposition Plus (TQD+), that offers maximum security guarantees during routing while simultaneously maintaining stable network queues. The key idea is to translate the notion of safety to link weights of the network, which is then jointly minimized along with the Lyapunov drift. Using techniques from the stochastic Lyapunov Drift-Plus-Penalty framework, we establish the bounded average queue occupancy and optimal security achievability. The theoretical results are supported with extensive simulations conducted on a comprehensive network simulator, TQD+SimModule, built on top of OMNeT++, and the simulator is made publicly available.