Degree Type

Dissertation

Date of Award

2014

Degree Name

Doctor of Philosophy

Department

Electrical and Computer Engineering

First Advisor

Sang Wu Kim

Abstract

Network coding can remarkably improve the network capacity by combining incoming packets at intermediate nodes. However, the packet combining also causes the network to be particularly vulnerable to the pollution attack that injects false data into the information flow of the network. This dissertation includes two studies on mitigating pollution attack in two-hop wireless relay network that employs random network coding.

First, we investigate how the finite field size affects the network coding performance in terms of the probability of symbol error and the throughput in adversarial networks where the false data is injected by the malicious attackers at source nodes and/or relay nodes. Also, we examine how the optimal field size that minimizes the probability of symbol error or that maximizes throughput changes as the trustworthiness of node or the number of combined packets changes.

Second, we propose two schemes for detecting the polluted packets and discarding them before decoding by exploiting physical layer information which is directly overheard from the source nodes. The proposed scheme I applies the threshold-based method to detect the presence of falsely injected data within each packet, while the proposed scheme II compares all received network-coded packets and selects the most trustable ones. Unlike many existing signature-based detection schemes, the proposed schemes do not require that additional information bits are attached into each packet.

DOI

https://doi.org/10.31274/etd-180810-3318

Copyright Owner

Duk Hee Yoon

Language

en

File Format

application/pdf

File Size

121 pages

Share

COinS