Abstract

E-mail inspection and mitigation systems are necessary in today's world due to frequent bombardment of adversarial attacks leverage phishing techniques. The process and accuracy in identifying a phishing attack present significant challenges due to data encryption hindering the ability to conduct signature matching, context analysis of a message, and synchronization of alerts in distributed detection systems. The author recognizes a grand challenge that the increase in the number of data analysis systems corresponds to an overall increase in the delivery time delay of an e-mail message. This work enhances PhishLimiter as a solution to combat phishing attacks using machine learning techniques to analyze 27 e-mail features and Software-Defined Networking (SDN) to optimize network transactions. PhishLimiter uses a two-lane inspection approach of Store-and-Forward (SF) and Forward-and-Inspect (FI) to distinguish whether traffic is held for analysis or immediately forwarded to the destination. The results of the work demonstrated PhishLimiter as a viable solution to combat Phishing attacks while minimizing delivery time of e-mail messages.

Library of Congress Subject Headings

Phishing--Prevention; Electronic mail systems--Security measures; Machine learning

Publication Date

8-5-2020

Document Type

Thesis

Student Type

Graduate

Degree Name

Computing Security (MS)

Department, Program, or Center

Department of Computing Security (GCCIS)

Advisor

Sumita Mishra

Advisor/Committee Member

Yin Pan

Advisor/Committee Member

Kaiqi Xiong

Comments

This thesis has been embargoed. The full-text will be available on or around 8/18/2021.

Campus

RIT – Main Campus

Plan Codes

COMPSEC-MS

Available for download on Wednesday, August 18, 2021

Share

COinS