Presentation Title: Advanced Learning Mechanism for SAC (Smart Access Control) System

Committee:

• Dr. Sitharama Iyengar (Committee Chair)
• Dr. Jianhua Chen
• Dr. Konstantin Busch

Date: Sept 29, 2011
Time: 10:00 AM
Location:164 Coates Hall, Robotics Lab

Abstract:
Security is the major concern in the society. The resources must be safely protected and in order to protect the resources, proper control must be provided to access the resources. Different methods have been proposed to provide secure access to the legitimate users like traditional lock/key, pin access, smart cards and biometric techniques such as voice recognition, face recognition and many others. The access control systems map the resources to the appropriate users and prevent any kind of loss.

The SAC system is a security system, which is used to provide secure access to the legitimate users by recording the users daily activities and using those activity patterns for providing adaptive security. Analysis of the collected data is performed, where the system carefully chooses users for additional layer of authentication procedure and isolates the legitimate users. The SAC control system also includes the concept of dynamic non-physical key, which uses individual users memory and brainpower to generate and update his security key dynamically. This helps in changing the key regularly with time and making the key more secure and preventing inappropriate users from securing access to the system.

Now that the SAC system is ready to provide secure access to the system, the performance is the major characteristic to be considered. The performance mainly depends on reducing genuine user rejections and bad user acceptance. The faster the system recognizes the user in all situations the better is the system. This project mainly deals with the learning mechanism for the system to adapt and learn the behaviors of the users. The learning mechanism allows the system to select the best match, make system faster, and provide most secure access.

The learning mechanism this project mainly deals with uses a set of valid secure user access; the system performs the respective learning from these valid users and learns to adapt to deviant situations to be more user friendly, secure and accurate. From the final result obtained, the system also corrects the wrong entry by updating the respective weights.

The goal of this project is to make the SAC system more secure and faster making it user friendly. With this learning technique the systems adaptability to different situations increases and making the decision making of the system more reliable.




All are invited.