Southampton-based Neural Computer Sciences Ltd is developing a version of its neural network artificial intelligence system that can be embedded in a Smart Card chip. The work is the firm’s contribution to the Esprit Cascade project set up to develop technology for Smart Cards. Neural Computer Sciences is working to develop ultra small feed-forward neural networks that can work within the limited computational and memory capabilities of a low cost chip. The Advanced RISC Machines Ltd ARM 7 32-bit RISC processor is being incorporated into the chip to provide the processing power to enable the portable intelligence device to perform algorithms to enable improved security and data compression. The new technology will use biometric validation for complex security checks. These checks will enable the verification of the user’s identity through unique characteristics such as fingerprints, signature recognition, voice or facial characteristics. Telephone credit card purchase is an example of one area that is particularly open to fraud. With the use of a telephone that has an integral Smart Card reader – something that telephone companies such as British Telecommunications Plc are at present developing – and a neural network embedded within a Smart Card, the validation of transactions can be more securely carried out. For example, the purchaser could speak a random word over the phone and in real time the terminal would run a signal processing algorithm on the speech to generate a profile, amounting to a sequence of around a couple of hundred bytes of data, which is then sent back in encrypted form to the user. The integrated network within the Smart Card then analyses this profile to determine whether the caller is the rightful user of the card and enables the transaction to be accepted or rejected. Brian Kett, managing director of the company says that in large volumes the cards could cost between $5 and $10 each. By the 1996, he says the project should be able to demonstrate a working example of the technology. The Cascade project, or in longhand, Chip Architectures for Smart CArds and portable intelligent DEvices, is a two and a half year scheme which was set up at the start of this year and funded to the tune of $3.5m by the European Community to develop technology for Smart Cards. Other participants in the consortium include France’s Dassault SA, working on card reader and terminal development and Gemplus SA, the French manufacturer of Smart Cards.
