Biometric-Based Authentication and Key Management in WBANs for M-Health

Abstract

The development of technology and communication was behind using sensors to measure the biometrics and movements of human body which resulted in the design of Wireless Body Area Networks (WBANs). This type of networks deals with very important data (biomedical data). However, the existence of such networks in places where it is possible to be maliciously accessible threatens the security of their information that may be exposed to attack. Any unauthorized manipulation or change in this data may cause medical fatal errors that threaten the life of patient. Hence, security issues are crucial in this type of networks, whereas key management represents the basic building block for achieving most of other security services.

In this thesis, we have proposed WBAN key management scheme based on symmetric cryptography using (RC4) and (AES-CFB) algorithms which depends on biometric keys that are generated from biomedical data. Every sensor is responsible for generating four keys specialized to it by measuring or sensing the same biometric more than once. In the protocol, every sensor takes its turn according to key refreshment schedule that determines the turn and time allocated to every sensor. The Personal Server (PS) launches key refreshment schedule periodically and participates in key management by update (Kcomm) in order to transmit the data in a secure manner on the network in normal operation. Our scheme takes into account deleting and/or adding new sensors to the WBAN. It also contains a simple solution to compromised sensor or PS.

A simulated prototype system has been implemented and simulation results have been obtained. From these results, it is obvious that the followed distributed key management style is suitable for WBANs because it's small size. Indeed, depending on biometric data to generate keys can achieve long, random and time variant keys, this looks so practical for WBANs. After computing execution time for every phase for this scheme with different options; one might conclude that the RC4 algorithm which achieved the least execution time. Also, achieved least complexity is very suitable choice for such applications and services