School of Computer and Information Science, Edith Cowan University, Perth, Western Australia
The increasing capability and functionality of mobile devices is leading to a corresponding increase in the need for security to prevent unauthorised access. Indeed, as the data and services accessed via mobile devices become more sensitive, the existing method of user authentication (predominately based upon Personal Identification Numbers) appears increasingly insufficient. An alternative basis for authentication is offered by biometric approaches; which have the potential to be implemented in a non-intrusive manner and also enable authentication to be applied in an ongoing manner, beyond initial point-of-entry. However, the implementation of any authentication mechanism, particularly biometric approaches, introduces considerations of where the main elements of functionality (such as the processing of authentication data, decisions making, and storing user templates/profiles) should reside. At the extremes, there are two alternatives: a device-centric paradigm, in which the aforementioned aspects are handled locally; or a network-centric paradigm, in which the actions occur remotely and under the jurisdiction of the network operator. This paper examines the alternatives and determines that each context introduces considerations in relation to the privacy of user data, the processing and storage of authentication data, network bandwidth demands, and service availability. In view of the various advantages and disadvantages, it is concluded that a hybrid approach represents the most feasible solution; enabling data storage and processing to be split between the two locations depending upon individual circumstances. This represents the most flexible approach, and will enable an authentication architecture to be more adaptable to the needs of different users, devices and security requirements.