Basic biometric architecture

The idea of using biometric techniques to authenticate users in telecommunications has been around for many years as a theoretical, yet cumbersome to implement, possibility (Lapere and Johnson, 1997). Recent increases in computing power, data transmission speeds and the availability of affordable storage, now make viable the use of biometrics in enterprise and consumer-based telecommunications (Gafurov, 2010).

The definition of a telecommunications network derives from the concepts of computer networks and distributed systems. According to Tanenbaum and Wetherall (2011a), a computer network is a collection of autonomous computers (nodes) interconnected by a single technology and a distributed system is a collection of independent computers that appear to their users as a single coherent system. Consequently, a telecommunications network can be defined as a distributed system in which nodes are either computing entities or computer networks and offers services related to information sharing.

The above definition determines that the main objective of telecommunications is to share or allow access to information through or from any kind of voice or data network. Information can take any form, including documents, audio, voice and video. Some information is intended to be openly available, such as websites on the Internet. Other information is restricted to a single party like a personal bank account web site or phone line or to a specific group of people as with a corporate intranet. Different mechanisms have been developed to control access to shared resources, which use passive or active authentication by the user sharing the resources (Mallery, 2013).

When telephonic communications were controlled by public providers, security was maintained by their exclusive access to all hardware. As networks evolved to packet and mobilebased technologies access was controlled by a personal identification number (PIN) and, in the case of mobile telecommunications, a smart card in the form of a Subscriber Identification Module (SIM). These solutions are convenient but lack security (Lapere and Johnson, 1997). Any telecommunications access control measure must meet the following requirements:
a. be simple to use yet effective enough to provide a noticeable level of security;
b. be measurable, recorded and quantifiable (Eur, 1997).

These are also characteristics of a biometric security system.

Biometric characteristics for authentication in telecommunications environments have been explored since the 1990s when the European Telecommunications Standards Institute (ETSI), made it a priority to provide secure communications standards for UMTS. They stated, « without a reliable authentication service through the Telecommunications Management Networks (TMN), every other effort to secure the system is in vain » (Eur, 1997). In the 1990s, there were obstacles to use biometric authentication methods: sensors were costly, processing power was low, service provider charges for data transmission were very high, acceptance and use of the technology were challenging. In the particular case of voice biometrics, telephones already had audio capturing sensors and they were a familiar device, making them a viable and non-intrusive option.

Basic biometric architecture

Communication network biometric systems are known as remote biometric authentication systems (Syta et al., 2015).

• A subject from whom the biometric information is read.

• The biometric data capturing process. This is accomplished by one or more mechanical or electrical objects that capture raw data to be analysed. The usual objects are microphones, webcams, mobile phones (which have several capturing methods), fingerprint readers, keyboards, mouse devices. A capturing element does not have to be a device specifically designed for biometric operations. A finger print reader has been specially designed for biometric data capture whilst a telephone has not. In general, any device able to retrieve biometric data and retransmit it in a digital format can be considered a capturing element.

• Database: Once a template is created, it is stored in a database. During a verification process, the matcher retrieves the claimed user template from the database and compares it with the one obtained from the feature extractor.

• The extraction process of finding the digital representation biometric data. It is as a twostep process:

• Pre-processing: Before being digitized, the raw data set is pre-processed to assess its quality. It is then segmented and enhanced. A quality assurance process for the raw data set is necessary to determine if more sets of biometric data are to be collected. Segmentation is mainly the process of separating actual biometric information from the background model. Finally, the raw data set is enhanced to improve its quality and reduce signal noise.

• Feature extractor: Pre-processed biometric information is digitized to create a biometric template. This template is expected to have unique individual information. This step is functionally merged with the template creation and matcher processes. For a new biometric speaker, the product of the feature extractor is used to build their biometric template. In the case of an existing speaker, the product is used to create a temporary digital representation of the captured biometric payload to be compared against their existing stored template.