Abstract:
"
Authentication can be defined as the process of recognizing the user's identity and it is the most
important step in the access control process to safeguard data/resources from being accessed by
unauthorized users. At first single factor authentication was introduced to protect the data and
verify the identity. Due to the increase in the security vulnerabilities in the single factor
authentication two/multi factor authentication has been introduced. Multifactor authentication
prevents unauthorized users accessing the system because it involves various authentication
factors. But on the other hand, it introduces several other problems on the user experience side.
The additional authentication layer reduces the user friendliness/ user experience of a given
application and the user has to spend more time to verify the identity in the extra authentication
step. In order to overcome this problem adaptive authentication was introduced.
Adaptive Authentication dynamically selects the best authentication option for authenticating a
user based on the context factors such as behavioural factors, location, network and some other
user attributes. This technology has the capability to change the standard authentication (i.e
username/password) procedure and lead the authentication into a more secure and user-friendly
manner. Through this research, existing work will be analyzed and an improved adaptive
authentication design will be proposed and implemented.
Adaptive Auth provides a novel design of adaptive authentication. Compared to the existing work
this research work focuses on both the user behavioural factors and user context factors. Separate
machine learning models will be used to identify the user based on the user behavioural and user
context factors. A novel approach to improve the performance of the whole authentication process
is put forward in this research work. Based on the user’s risk profile the authentication mechanisms
will be selected. This enhances the user experience of the authentication process."
