Date of Award

Spring 5-31-2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Electrical Engineering - (Ph.D.)

Department

Electrical and Computer Engineering

First Advisor

Abdallah Khreishah

Second Advisor

Issa Khalil

Third Advisor

Nirwan Ansari

Fourth Advisor

Roberto Rojas-Cessa

Fifth Advisor

Reza Curtmola

Abstract

According to Wikipedia, authentication is the act of confirming the truth of an attribute of a single piece of a datum claimed true by an entity. Specifically, entity authentication is the process by which an agent in a distributed system gains confidence in the identity of a communicating partner (Bellare et al.). Legacy password authentication is still the most popular one, however, it suffers from many limitations, such as hacking through social engineering techniques, dictionary attack or database leak. To address the security concerns in legacy password-based authentication, many new authentication factors are introduced, such as PINs (Personal Identification Numbers) delivered through out-of-band channels, human biometrics and hardware tokens. However, each of these authentication factors has its own inherent weaknesses and security limitations. For example, phishing is still effective even when using out-of-band-channels to deliver PINs (Personal Identification Numbers). In this dissertation, three types of secure entity authentication schemes are developed to alleviate the weaknesses and limitations of existing authentication mechanisms: (1) End user authentication scheme based on Network Round-Trip Time (NRTT) to complement location based authentication mechanisms; (2) Apache Hadoop authentication mechanism based on Trusted Platform Module (TPM) technology; and (3) Web server authentication mechanism for phishing detection with a new detection factor NRTT. In the first work, a new authentication factor based on NRTT is presented. Two research challenges (i.e., the secure measurement of NRTT and the network instabilities) are addressed to show that NRTT can be used to uniquely and securely identify login locations and hence can support location-based web authentication mechanisms. The experiments and analysis show that NRTT has superior usability, deploy-ability, security, and performance properties compared to the state-of-the-art web authentication factors. In the second work, departing from the Kerb eros-centric approach, an authentication framework for Hadoop that utilizes Trusted Platform Module (TPM) technology is proposed. It is proven that pushing the security down to the hardware level in conjunction with software techniques provides better protection over software only solutions. The proposed approach provides significant security guarantees against insider threats, which manipulate the execution environment without the consent of legitimate clients. Extensive experiments are conducted to validate the performance and the security properties of the proposed approach. Moreover, the correctness and the security guarantees are formally proved via Burrows-Abadi-Needham (BAN) logic. In the third work, together with a phishing victim identification algorithm, NRTT is used as a new phishing detection feature to improve the detection accuracy of existing phishing detection approaches. The state-of-art phishing detection methods fall into two categories: heuristics and blacklist. The experiments show that the combination of NRTT with existing heuristics can improve the overall detection accuracy while maintaining a low false positive rate. In the future, to develop a more robust and efficient phishing detection scheme, it is paramount for phishing detection approaches to carefully select the features that strike the right balance between detection accuracy and robustness in the face of potential manipulations. In addition, leveraging Deep Learning (DL) algorithms to improve the performance of phishing detection schemes could be a viable alternative to traditional machine learning algorithms (e.g., SVM, LR), especially when handling complex and large scale datasets.

Share

COinS