Securing unfamiliar system entry points against faulty user authentication via electromagnetic side channel attacks

This research presents concepts intended to reduce a system's vulnerability to a password sniffing technique that exploits differential current usage in hardware, known as an electromagnetic side channel attack. The first countermeasure discussed is an augmentation to the standard register design that significantly reduces the amount of electromagnetic interference produced by these circuits. Furthermore, this paper outlines a novel approach to user authentication based on keystroke dynamics that, in the event a user's password is compromised by this kind of attack, renders this information insufficient as a standalone credential for system login.;One well documented security concern when using a keyboard as an input device to a system is the capability of an individual to measure electromagnetic emanations produced by the changes in current and register states associated with a user's keystroke activity. By capturing and analyzing these signals an attacker can discover the user's exact keystroke pattern, possibly revealing a password or other sensitive information. Background research was conducted into this form of computer system attack and causes as well as potential countermeasures identified. Techniques known as signal strength reduction and signal information reduction are described as traditional corrective methods. However, because of limited commercial interest, these approaches have not been applied to vulnerable system entry point devices such as the standard keyboard. To address this potential security concern a new model of register was created with an emphasis on equalizing current flow to the various nets within a device. Using this design significantly reduces the electromagnetic interference produced by the keyboard device, making it considerably harder to isolate the information bearing signal from standard atmospheric noise, thus hardening the system against this type of a side channel attack.;Despite significantly reducing the possibility of an electromagnetic side channel attack, the use of these secure registers does not entirely eliminate the potential for hardware signal interception. Consequently, an improved keystroke dynamics algorithm is also presented which when used in combination with interference reducing registers provides near total protection from hardware password sniffing techniques. Keystroke dynamics is a field of study that leverages an individual's consistently demonstrated tendencies when typing commonly used words as a means of behavioral biometric identification. Prior to the presentation of a novel augmentation to this concept, an investigation into traditional methods for identifying and isolating the unique characteristics of user's typing pattern was performed. Concepts of "flight" and "down" time were recognized as established means for these purposes and the outcome of research implementing these mechanisms produced and analyzed. Close examination of these results showed inconsistencies in the ability of such routines to consistently and accurately determine user identification based on these two heuristics alone. Consequently, new capabilities which incorporated the concept of pressure were modeled and added to expand upon the original timing based characteristics of this biometric technique. These enhancements resulted in substantial increases in the accuracy of the overall authentication routine such that this method of user recognition now displays the potential for deployment as a standalone system that further could diminish the threat posed to keyboard entry devices by electromagnetic side channel attacks.