Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.

Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.

Document Type

Open Access Dissertation

Degree Name

Doctor of Philosophy (PhD)

Degree Program

Electrical and Computer Engineering

Year Degree Awarded

2016

Month Degree Awarded

February

First Advisor

Dennis Goeckel

Second Advisor

Hossein Pishro-Nik

Subject Categories

Systems and Communications

Abstract

This dissertation focuses on secrecy, which is a primary concern in modern communication. Secrecy has traditionally been obtained by cryptography, which is based on assumptions on current and future computational capabilities of the eavesdropper. However, there are numerous examples of cryptographic schemes being broken that were supposedly secure, often when the signal was recorded by the adversary for later processing. This motivates seeking types of secrecy that are provably everlasting for sensitive applications. The desire for such everlasting security suggests considering information-theoretic approaches, where the eavesdropper cannot extract any information about the secret message from the received signal. However, since the location and channel state information of a passive eavesdropper is generally unknown, it is challenging to know whether the advantage required to achieve information-theoretic security for a given scenario is provided, and thus attempting to obtain information-theoretic security via commonly-envisioned approaches leads to a significant risk in wireless communication.

In this dissertation, we present a new perspective on how to generate the necessary information-theoretic advantage required for secret communication in the wireless environment. The proposed technique does not rely on the channel between the transmitter and the eavesdropper's receiver because we exploit receiver's processing effects for security. In particular, we attack the eavesdropper's analog-to-digital (A/D) converter to generate the advantage required to obtain information-theoretic secrecy, as follows. Based on a key pre-shared between the legitimate nodes that only needs to be kept secret during transmission (and we pessimistically assume it will be handed to the adversary immediately afterward) we insert intentional distortion on the transmitted signal. Since the intended recipient of the signal knows the key and hence the distortion, it can undo the distortion before his/her A/D, whereas the eavesdropper must store the signal in memory and try to compensate for the distortion after the A/D conversion. Since the A/D is necessarily a non-linear component of the receiver, the operations are not necessarily commutative and there is the potential for information-theoretic security. This dissertation studies two practical instantiations of this approach to obtain everlasting secrecy against eavesdroppers with different hardware capabilities. As a first step, the transmitted signal is modulated by two vastly different power levels at the transmitter based on the key. Since the intended recipient knows the key, he/she can undo the power modulation before the A/D, putting the signal in the appropriate range for analog-to-digital conversion. The eavesdropper, on the other hand, must compromise between larger quantization noise and more A/D overflows, and thus will lose information required to recover the message. Hence, information-theoretic security is obtained. We show that this method can provide information-theoretic secrecy even when the eavesdropper has perfect access to the output of the transmitter, and even when the eavesdropper has an A/D that has better quality than the legitimate receiver's A/D. A risk of the power modulation approach is a sophisticated eavesdropper with multiple A/Ds. In our second approach, in order to attack such an eavesdropper, we introduce the idea of adding random jamming (based on the ephemeral key) to the signal. In this case the intended recipient can simply subtract off the jamming signal and its signal will be well-matched to the span of its A/D converter, while the eavesdropper has difficulty because it does not know the key during transmission: if it does not change the span of the A/D, it will lose information due to A/D overflows, and, if it enlarges the span of the A/D to cover all possible received signal values, the width of each quantization level will be increased, and thus the eavesdropper will lose information due to high quantization noise. Hence, the desired advantage for information-theoretic secrecy is obtained. Finally, we study the combination of random jamming and frequency hopping in wideband channels, and show that considering the current fundamental limits of analog-to-digital conversion, this method can provide everlasting secrecy in wireless environments against any eavesdropper.

Share

COinS