| PhD Viva

Name of the Speaker: Ms. Jaswanthi Mandalapu (EE19D700)
Guide: Dr. Krishna Jagannathan
Online meeting link: https://meet.google.com/mad-qbon-vpe
Date/Time: 12th June 2025 (Thursday), 10:00 AM
Title: Capacity and Coding for Queuing Systems with Waiting-Time Dependent Noise

Abstract :

In this talk, we first consider a quantum network with Markovian queues and investigate the fundamental limits of classical communication using quantum states (or qubits) that decohere as they traverse through the network. Furthermore, we identify optimal pumping rates and routing probabilities to maximize capacity in simple topologies, such as tandem and parallel. We next consider a photonic communication system where photon detectors experience random dead time following each successful detection. If subsequent photon arrivals occur during the dead time, we assume the information to be deleted or erased. We observe that this results in a highly correlated channel model, where each correct detection is followed by a burst of deletions or erasures. For such a burst noise channel model, we derive the fundamental limits on the rate of classical information transmission under two settings: (i) systems with paralyzable detectors and (ii) systems with non-paralyzable detectors. We highlight that such channel models are also critical in communication systems such as IoT devices and multimedia streaming, where the servers are faulty, or the buffer sizes are limited. We then consider a classical queue-channel setting and propose capacity-approaching channel codes. Our contributions in this part are two-fold: (i) we propose a generic wrapper based on interleaving across renewal blocks of the queue to convert any capacity-achieving code for a memoryless erasure channel into a capacity-achieving code for the EQC, and (ii) we further analytically examine the performance of LDPC and Polar codes without interleaving, and theoretically prove that Arıkan’s Polar transform successfully polarizes the M/M/1 EQC, and hence approaches its capacity. Finally, we consider a quantum Markovian channel with countable state-space, and aim to propose capacity-approaching coding schemes for such channels. Focusing on erasure and unital noise, we theoretically prove that classical polar coding is sufficient to achieve the classical capacity of such channels. Notably, our ability to accommodate countable-state memory enables explicit code construction using classical polar codes for quantum queue-channels.