Biometrics, measures habitual tendencies of individuals dueBiometrics, measures habitual tendencies of individuals due

Biometrics, an automated identification system 1, 2, is based on
an individual’s physiological or behavioral characteristics to make positive
identification. These systems identify the uniqueness in physiology and
behavior attributes of an individual for biometric authentication. Biometric
identifiers are distinctive in nature but physically measurable characteristics
to recognize an individual. However, physiological and behavioral both can be
separated as two subsets of biometric identification system. Physiological
characteristics accounts recognition of natural attributes, such as
fingerprint, iris, retina, face, palm veins etc. 3, whereas behavioral
authentication measures habitual tendencies of individuals due to psychological
and physiological differences. Writing style, typing rhythm and unconscious
body movements are used in behavioral biometric system. Among multiple
authentication technologies, behavioral biometric system is inherently more
reliable as a consequence of natural uniqueness of each individual,
psychologically as well as physiologically. Physiologically identified
authentication systems are successfully implemented, though with certain
recognition errors; the iris scanner, with an Equal Error Rate (EER) of 0.01%,
performed the best 3.

In security system, authentication is a process of verifying the
digital reference identity stored in the system, matching with the identity of
the individual asking for access. It is expected that, the reference
psychological and behavioral characteristics may reduce the error rate in
identification of an individual.  These
methods include keystroke dynamics 4, 5, mouse dynamics 6, 7, 8,
signature verification 3 and Graphical User Interface (GUI) analysis 9.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

In this study,
25 individuals’ typing behavioral characteristics were studied through key
stroke dynamics. Ten valid samples of each user were recorded for a common
password ‘welcome’ and features were optimized by using Hybrid Firefly
Algorithm (HDFA). The results of HDFA were compared with those obtained from
various advanced algorithms such as Particle Swarm Optimization (PSO), Genetic
Algorithm (GA), Artificial Bee Colony (ABC) and Ant Colony Optimization (ACO).
It was found that, the iterative generations and processing time required for HDFA
is about 18 % and 41% less respectively, as compared to those of other


Keystroke dynamics (KD) 10, is a
behavioral authentication tool, which extracts and analyzes how (pattern and
way) an individual type rather than what the individual types. In KD, no any
extra hardware is required to be added as in other biometric systems. A typical
keystroke dynamic authentication system may consist of several components like
data acquisition, feature extraction, feature selection, classification,
decision and retraining. However, the following three major steps are involved
in keystroke dynamics analysis 11:

A time vector is
registered for an individual, through typing a specific password by the

The specific features of the registered
password are extracted.

A feature selection is made by filtering
redundant or irrelevant features from large scale
data sets, to store the reference pattern.

In KD based
authentication system, reducing computational speed and improving prediction
accuracy, in feature selection, decides the accuracy and reliability of the

In the following lines, the
terminology and their specific significance is mentioned in accordance with KD

A. Data

Data collection is the preliminary and essential step of
keystroke dynamics 12. It is the process of gathering and measuring information
on targeted variables in an established systematic fashion. The goal of data
collection is to capture quality evidence. To capture keystroke dynamics, it is
necessary for users to type their own password a number of times during

B. Feature

There are number of different aspects
of keystroke characteristics for feature extraction that can be used for
identification such as cumulative typing speed, the time elapsed between
consecutive keys, the time that each key held down, the frequency of the
individual in using  other keys on the
keyboard like the number pad or function keys, the sequence utilized by the
individual when attempting to type a capital letter (for example, does the user
release the shift key or the letter key first), pressure applied for a
keystroke, typographical errors made etc. Initially the statistical measures of
feature characteristics such as latency, duration and digraph are computed 12,

a)     Duration: Keystroke duration (Dwell/Held time) is the interval time
in milliseconds (ms), between a key press and a key release of the same key.
Figure 1 shows, how the duration time of the key ‘T’ is determined. Duration
time is strictly greater than zero.

Latency: Latency is defined as
the time in ms between two consecutive keystrokes. In Figure 1, the time
between the key release of ‘H’ and key press of ‘E’. The latency times can be
negative, i.e. the second key is depressed before the first key is released.

c)     Digraph: Digraph is the time interval between the down key event
of the first keystroke and the up key event of the second keystroke. In Fig. 1,
the time between the key press of ‘T’ and key press of ‘H’.