Files in this item



application/pdfSP20-ECE499-Thesis-Sanyal, Debopam.pdf (634kB)
(no description provided)PDF


Title:Optimizing network approaches using P4 programming
Author(s):Sanyal, Debopam
Contributor(s):Mohan, Sibin
Subject(s):software-defined networking
data plane
Abstract:Forwarding data packets between switches is the basis of propagation of information in computer networks of all types. Hence, in order to ensure optimal utilization of hardware and computing resources, it is imperative to make the forwarding process fast, efficient and resilient. Decentralizing intelligence in networks often makes packet transmission time-consuming as there is no separation between the forwarding and routing processes. This project focuses specifically on reducing this time by implementing the forwarding decisions of the switch in the data plane and leaving the control plane to handle the routing process. This is a well-known feature of software-defined networks (SDNs). In most cases, switches forward packets depending on their characteristics. Therefore, the focus is on this conditional basis of forwarding at the switch level by implementing three different forwarding paradigms for two topologies: (a) utilizing unused, but optimal topology paths, (b) placing importance on the urgency of data flows and (c) guaranteeing connectivity. The P4 programming language was chosen over the OpenFlow protocol since the latter does not provide complete protocol independency. The evaluation process involved testing the programs on the mininet network emulator first and then on the SDNet and SUME simulations of the NetFPGA-SUME. The goal of this project is to provide end-to-end deadline guarantees for real-time networks using an SDN by programming the switch data plane. This consists of developing and evaluating (i) deadline-driven (end-to-end delay guarantees) and (ii) budget-driven packet scheduling algorithms and (iii) providing resilience in case of link failures. The broader mission is to forward packets in an optimal fashion.
Issue Date:2020-05
Date Available in IDEALS:2020-06-11

This item appears in the following Collection(s)

Item Statistics