One in the eye to plastic card fraud.

Introduction

In a previous article in IJR&DM, Gould[1] highlighted the growth in the use of plastic cards and the rise in associated crime. In the UK, the number of plastic card transactions is expected to reach 4 billion per year by the millennium. This represents a tenfold increase since 1985 and, if the estimates are accurate, is likely to result in a similar growth in plastic card fraud. In 1993 in the UK, 5,000 cards were lost or stolen every day, resulting in a loss of [pounds]129.8 million. Each stolen card results in an average of [pounds]580 being charged fraudulently against the account before the card is recovered [2].

In the UK, 78 per cent of fraudulent transactions passed through the tills of retailers [1]. This is costing retailers, financial institutions and their customers a great deal of money and effort to rectify the problem. While the problem is not unique to the UK, the dominance of plastic in the pockets and purses of British consumers makes the search for effective preventive measures all the more urgent. In the short term, the introduction of photocards and personal identification numbers (PINs) may deter fraudulent use. However, in the long term, smart cards and more advanced technology will be required to verify the cardholder's identify[1]. This article aims to explore the potential of biometric technology in the battle against plastic card fraud.

Cardholder verification

Gould[1] identified that to prevent "...plastic card fraud depends on being able to establish that...the person presenting the card is the genuine cardholder". This can be partially achieved by requiring the cardholder to demonstrate "knowledge", such as a PIN. However, the card can still be used if the PIN is known by the unauthorized user. This problem can be overcome by validating the cardholder using "possession", "knowledge" and unique "characteristics" of the human body as shown in Figure 1.

Biometric technology

The intersection between possession of the card, knowledge and a unique human characteristic using biometric technology enables the card holder's identity to be reliably verified. Perhaps the earliest examples of biometric techniques were the use of fingerprints and voice recognition in 1944 [4,5] positively to identify individuals. While novel in the retail environment, biometric devices have been used to prevent unauthorized access to areas such as computer rooms, vaults, research laboratories, blood banks, jails and airports for several years. Table I shows the two main classes of biometrics used to verify a person's identity.

Devices based on behavioural technology attempt to measure actions over which an individual has a degree of control and, as such, must be capable of dealing with intra-personal variation in performance. Behavioural biometrics work best when used regularly to build up a pattern of that variation. On the other hand, physiological techniques use constant human characteristics such as fingerprints These devices do not have to deal with intra-personal variations which occur each time the customer uses the machine.

Application of biometrics

For biometric techniques to be successful in the retail or banking environment, there are four factors which have to be considered:

(1) recognition performance;

(2) speed of use, usability and customer acceptance;

(3) device and card cost;

(4) industry standards.

 
Table I 
Classes of biometric technology 

 
Biometrics class              Example 

 
Behavioural                   Signature 
                              Voice pattern 

 
Physiological                 Fingerprints 
                              Retinal and iris scans 
                              Hand and finger geometry 
                              Face prints 

Recognition performance

Biometrics use statistical techniques to verify the cardholder's identity and, as such, they can make two types of error. The device may reject a valid …