User authentication is the process of verifying whether the identity of a user is genuine prior to granting him access to resources or services in a secured environment. Traditionally authentication is performed
statically at the point of entry of the system (e.g. login); this is referred to as
static authentication. A popular form of a static authentication technique widely used in computer networks is password-based authentication. It is a well-established fact that traditional passwords are unsafe. Passwords may be stolen or may be cracked using the so-called dictionary attack. Generally speaking there are at least two main issues with static authentication techniques. Firstly, in the case where the authentication process fails to genuinely verify the identity of a user — as may happen, for instance, with password-based authentication schemes — there is no other opportunity to get things right in the rest of the login session or to establish after the session that some malicious activity occurred. Secondly, a successful authentication at the beginning of a session does not provide any remedy against the session being hijacked later by some malicious user.
One of the solutions proposed to address these shortcomings is
continuous authentication (CA). CA consists of the process of positively verifying the identity of a user in a repeated manner throughout a login session. CA departs from the traditional (static) authentication schemes by repeating several times the authentication process dynamically throughout the entire login session; the main objectives being to detect masqueraders, ensure session security, and combat insider threat. Although CA can be effective in detecting session hijacking, it requires special data sources to detect masqueraders. Such data source should allow the system to discriminate reliably legal users from imposters. Although non-biometric data sources such as user commands sequences and RFID may be used, biometrics technologies are the most suitable for this purpose. This book presents a selection of 11 chapters on continuous authentication using biometrics contributed by the leading experts in this recent, fast growing research area. These chapters provide collectively a thorough and concise introduction to the field of biometric-based continuous authentication. The book covers the conceptual framework underlying CA and presents in detail processing models for various types of practical CA applications.
This book is primarily intended for researchers, developers, and managers who are interested or are currently working in the emerging field of continuous authentication. The book is also of relevance to students of computer science and engineering, in particular, graduate students at the Master’s or PhD levels.
The book consists of 11 chapters organized in three parts as follows:
- Part 1: Fundamentals
- Part 2: Continuous Authentication based on Physiological and Cognitive Biometrics
- Part 3: Continuous Authentication based on Behavioral Biometrics
A brief outline of the book is as follows.
As in any emerging area, the field of continuous authentication requires establishing some common ground regarding the basic aspects and entities as well the characteristics of a standard continuous authentication system. Part 1 of the book involves three chapters (i.e. chapters 1-3) covering fundamental concepts and generic models for CA systems design, implementation, and evaluation.
Traore and Ahmed presents in Chapter 1, a conceptual framework outlining basic concepts and entities underlying CA systems. They also introduce a generic architecture for CA systems and discuss various applications of the technology.
In chapter 2, the same authors identify a set of metrics and models for the design and evaluation of CA systems. The chapter introduces several empirical guidelines that can be used to design high-performance CA systems.
In Chapter 3, Jagadeesan and Hsiao make a significant contribution in outlining CA fundamentals by identifying and discussing the factors and steps involved in building a good continuous authentication system. They illustrate the introduced factors and steps by presenting the case study of a multimodal CA system based on mouse and keystroke dynamics biometrics technologies.
Part 2 involves 4 chapters (i.e. chapters 4-7) covering the use of physiological and cognitive biometrics for continuous authentication.
The typical trend in continuous authentication research is to establish a clear separation line between static log-on phase and the repeated authentication occurring in the rest of the session. However, the static log-on can influence the rest of the verification activities taking place in the rest of the session to a large extent.
In this regard, Tsatsoullis and colleagues propose, in Chapter 4, to use a robust biometric technology for static log-on combined with a less computationally expensive biometric technology for the continuous verification.
While Tsatsoullis and colleagues are proposing a multi-biometric system involving two biometrics technologies operating independently, Kisku and colleagues discuss the merits of intra-modal biometric systems over uni-modal systems. Likewise, in Chapter 5, they propose a low-level multispectral palmprint biometric fusion scheme, which enhances the accuracy of the palmprint biometric technology by using Gabor wavelet transformation. The authors argue that the proposed multispectral palmprint biometric system can be used to continuously authenticate individuals (when required) in high security zones such as border control as well as low security ones such as home computing environments.
In chapter 6, Revett discusses a novel approach for continuous authentication based on cognitive biometrics which uses signals based on the response(s) of nervous tissue such as the electrocardiogram (ECG) or the electroencephalogram (EEG). Although the discussion remains at a high level, it provides interesting and useful insight into the background work and foundational technologies underlying cognitive biometrics.
In chapter 7, Riener discusses the need for continuous authentication for drivers, and then present a system that uses wireless ECG to continuously verify user identity while they are sitting in the car.
Part 3 involves 4 chapters (i.e. chapters 8-11) covering the use of behavioural biometrics technologies for continuous authentication.
In chapter 8, Derawi and colleagues discuss the use of gait biometric for continuous authentication and then introduces a continuous authentication system based on wearable gait recognition technology.
The remaining 3 chapters of part 3 cover continuous authentication based on keystroke dynamics. Keystroke dynamics, along with a few other biometric technologies, exhibit a number of key characteristics ideal in continuous authentication: the data comes as stream which is readily available throughout the session; data collection and analysis can be conducted unobtrusively and do not require any special purpose hardware.
In chapter 9, Gunetti and Picardi discuss and investigate the use of keystroke analysis for computer intrusion detection, which is one of the key applications of continuous authentication. They also briefly discuss other interesting applications of keystroke analysis relevant to continuous authentication such as identity tracing in the cyberspace.
In chapter 10, using a hybrid keystroke analysis model based on the combination of the array disorder distance metrics (introduced by Gunetti and Picardi), and the Euclidian distance, Samura and Nishimura establish that the effectiveness of keystroke analysis might be proportional to the skill levels of the typists. In the same vein, in chapter 11, Silva Filho and Roisenberg discuss the efficiency and cost-effectiveness of using keystroke analysis for continuous authentication, and introduce an approach based on neural networks committee machines.
The above chapters represent a broad coverage of the state of the art which we believe will lead the readers to a deeper understanding of the art and science underlying the emerging field of continuous authentication.