Preventing child abuse on digital computer based system through keystroke dynamics.
Paper Presenter: NARAINSAMY PAVADAY, ACM
Author(s):
- NARAINSAMY PAVADAY, ACM
- K.M.Sanjeev SOYJAUDAH, IEEE
The past years has witnessed an unprecedented development and exponential growth in the use of electronic, computer-based communication and information sharing devices. At the forefront is the computers and to a lesser degree, the mobile phones. Inherently there are numerous benefits in using the Internet for information retrieval and communication but until recently there has been little or no recognition of the dangers that may also result from the use of such technology. There is now growing evidence that the Internet is a new medium through which some commonly recognized forms of child maltreatment, sexual and emotional abuses may be pursued. Unfortunately the scars can be deep and long-lasting, affecting not just abused children but society. Moreover the raw fact is that for every incidence of child abuse that gets reported many more goes undetected.
This paper reports on preliminary investigation of the application of keystroke dynamics techniques in securing user interaction by controlling who access what. In brief values obtained from previous interactions are compared to the actual input and a match, within a certain precision will grant access to the requested data.
INTRODUCTION
Millions of internet users access servers each day and it is expected to rise further with advances in technology and decrease in cost. The vast majority of people use a computer to go online although nowadays mobile phones and TVs can also offer that possibility. The internet is a huge technological achievement, promising great benefits to society in general and to children and young people in particular, especially in education and entertainment. On the education side it symbolizes a vast and ever expanding library that users can bring into their home and is very easy to use with little or no knowledge or previous experience of computers. As with any emerging technology the internet is not different in that is represents a source of potential dangers for children, as seen by the numbers of incidents making the headlines of newspapers. Unfortunately no proper means exist, to reliably and irrefutably differentiate between an adult and children who is not authorized to access these illicit material on the web. Authentication is defined as the process by which a computer program or another user attempts to confirm that the computer, computer program, or user from whom the second party has received some communication is, or is not, the claimed first party [1]. Biometric mechanisms represent the strongest means to authenticate people [2, 3, 4] as they cannot be lost, shared, copied and are unique. As human beings we have characteristics that help identify us from others. Our genetic code, fingerprints, handwriting, and ocular retinal pattern are examples of biometric features that make us unique and distinguishable as individuals.
The keystroke frequency of a user is a distinctive feature that, even when it is not as precise as others in terms of entropy and classification power [2], has the advantage of not requiring costly equipment and software be implemented. Therefore it can be used to strengthen password-based authentication. The mechanism feasibility is based on the hypothesis that every individual follows different guide lines when interacting with the keyboard. In other words, when an individual is writing a character sequence he maintains certain rhythm that is different of the one of other individuals. It is thus possible to identify a computer user based on this feature and restrict what s/he is authorized to access or not. As the input device is the existing device, its implementation costs less. Moreover neither the enrollment nor the verification disturbs the regular flow as the user would be typing on the keyboard anyway. In addition to being non obtrusive no specific training is required by the user before interacting with the system. Biometrics being none perfect, acceptances rate are important factors used in evaluating a particular solution. False acceptance rate (FAR) is the percentage of imposters wrongly matched and authorized in the system. A false reject rate (FRR) identifies the percentage of valid users who are wrongly rejected.
In this paper, we detailed an efficient and seamless authentication scheme based on keystroke dynamics tested on two different algorithms. This system identifies the legitimate user in real time using pattern classification technique, based on neural and k-nearest neighbor, which is applicable to any keyboard based input. Before proceeding with the proposed scheme, we shall briefly review the related works on keystroke dynamics. Following this review, our proposed scheme is presented. The experimental results and the selection of an applicable structure are discussed thereafter. Finally, our conclusions and future enhancements are presented in the last section of this paper.