Ikendi Fingerprints in Place of Car Keys

Biometric recognition systems are destined to displace PIN codes in many applications (smart credit cards and mobile phones immediately come to mind). The advantages of biometrics are compelling — enhanced security, easy and convenient handling, and unambiguous verification of identity. These methods of recognition create entirely new opportunities for automotive applications.

To protect mobile telephones and credit cards from unauthorized use, people have relied for many years on anumber generally composed of four digits, the PIN code. This system has its drawbacks. With just 10,000 possible combinations, it offers relatively low security. Beyond that, the user is often compelled to memorize a multitude of different numeric codes for different devices and/or electronic cash and credit cards. For these reasons, engineers have been looking for alternative systems offering failsafe protection against unauthorized access, unambiguous identity verification, and greater convenience. Biometric technology can store and retrieve the distinguishing traits of physiological features that enable unambiguous identification.

EASE OF USE IS DECISIVE
However, for biometric technology to make inroads, it must satisfy other criteria, for example, the need for low cost, ease of use, and microelectronic space constraints. And of course the method of identification must win widespread acceptance among consumers. There is much to be said for fingerprint recognition — next to its popular appeal, it is easy to use. The finger is simply moved over or placed on a sensor, and a scanner records the papillary lines in a fraction of a second. Several manufacturers offer sensors that use different methods to transform the values for papillary lines detected on the skin by the sensor’s surface into gray shade images. Fingerprint sensors  enerally have a surface area of between 150 and 250 mm2. Line scanners (thermal scanners, for example) have a much smaller area (approximately 30 mm2). All scanning processes are able to achieve resolutions up to 500 dpi. Complex mathematical formulae extract the distinguishing features from the image recorded by the sensor.

MINIATURIZATION OF ELECTRONIC EVALUATION UNITS
Highly integrated, low-cost solutions must be made available to effect a wide-scale breakthrough of this identification technology in automotive engineering, mobile telephones and banking cards. For this reason, IKENDI partnered with Atmel to develop an ASIC based on ARM technology and Atmel’s AT91 microcontroller standard product. It features all of the functional units required for secure, fast and affordable fingerprint recognition.

ARM MICROCONTROLLER AS THE POINT OF DEPARTURE
IKENDI chose ARM technology to provide the architectural underpinning for the ASIC with forethought: It boasts remarkably powerful processor cores. Offering various levels of performance, they are highly efficient when it comes to silicon area, memory usage and power consumption. After settling on ARM for its core technology, IKENDI adapted and optimized its IP components to suit the architecture. These efforts focused on implementing complex image processing functions in low-cost embedded systems (system-on-chip).

As far as IP goes, there is a range of bus peripherals and operating systems available. With the extremely compact hardware and software by IKENDI, it was possible to integrate a complete, highly flexible fingerprint recognition solution on a single chip offering high-speed processing and very low power consumption.

A programmable interface is available to connect different manufacturers’ proprietary fingerprint sensors. An ARM7TDMI® processor writes the image data captured by the fingerprint sensor to the integrated RAM. From there the lines are transmitted via a special multiplexer that converts this 32-bit data into 128 bits and sends it to the IKENDI ConvTree-III image co-processor. This component executes the basic tasks involved in evaluating images (encoding). Comprising less than 50,000 gates, this co-processor  was developed specifically for image processing. It features extremely effective algorithms, some of which are hardwareencoded. With the benefit of these algorithms and 128-bit processing, this solution encodes images 10 to 20 times faster and with greater energy efficiency than other solutions at the same clock speed (DSP-powered solutions, for example). The encoded result is called a template. A data record with just 200 bytes, it can be transmitted via a failsafe circuit to a host system, chip card or control unit for further processing. Diverse interfaces such as USB, serial or CAN buses are available for this purpose. Data is coded by the ASIC’s on-board RSA encryption unit. In a second step, the scanned data containing the characteristics of the fingerprint is matched with fingerprints stored in the system.

The entire fingerprint check takes less than 500 ms, with image capture and processing taking up the lion’s share of this time. Since the algorithms employ integer operations rather than floating point calculations, the system does not require a floating point co-processor. Moreover, the entire application has been optimized to conserve chip area and cut costs with regard to memory requirements. The ARM7 and the image coprocessor require less than 32 kilobytes of RAM for image processing, and three kilobytes of RAM for fingerprint matching.

Also, program code takes up very little room in the ROM. The matcher makes do with just 12 kilobytes of ROM, while 128 kilobytes suffice for all the fingerprint software, including the application. In total, the ASIC has a chip area of only27 mm2. Atmel has used its 0.25µm CMOS process to produce the ASIC. It is available as a wafer, single die, or in different packages such as TQFP, BGA and CSP. For mobile applications, the chip’s already very low power consumption can be decreased further still. To this end, in standby mode the entire circuitry of the chip is switched off except for a low-power area. Located in this area is a tiny supervisory circuit timed by a 32-kHz clock quartz. It is able to wake up the ASIC whenever its services are required. Particularly in automotive applications where many powered components are on permanent standby, power efficiency is a primary concern. A contemporary standard lithium battery is able to provide sufficient power for 10,000 fingerprint recognition operations. The fingerprint scanner can be integrated into a car key.

EXTENSIVE DEVELOPMENT TOOLS
Offering the fingerprint application for licensing only, IKENDI provides an end-to-end IP solution featuring recognition and matching functionality, various fingerprint sensors,  and secure data transfer via different industry-standard interfaces to facilitate development of custom applications. For this purpose, the company offers a comprehensive design kit featuring a base board (Figure 1), sensor plug-in board, fingerprint sensor, software development environment, extensive documentation, training, and support. The design kit is available for different models and fingerprint sensors.

By virtue of the ASIC’s compatibility with Atmel’s AT91 family, the design kit is already available with the Atmel AT91FR4081 for prototyping. Once the ASIC has been approved for volume production, all of the software can be adapted from this environment. For the operating system, IKENDI offers eCos (from Cygnus, Red Hat), a royalty-free multitasking operating system environment, with a corresponding AT91 port. All software was generated in ANSI-C, and may easily be adapted to different operating systems. IKENDI also offers services for this purpose.

IMAGE PROCESSING IN THE AUTOMOTIVE FIELD
Fingerprint recognition and encoding systems for identity verification and access control will see widespread use in the coming years, and this certainly will not be the last image processing application in the field of automotive engineering. The IP solution by IKENDI satisfies all of the automotive industry’s salient requirements: extremely fast response, energy efficiency, extended temperature range, and availability of interfaces prevalent in automotive applications (CAN, for example). On account of the IKENDI algorithms’ image processing capability, this solution may also be adapted for use in other applications, for example, dynamic track sensing, interior monitoring, and traffic sign recognition.

Fingerprint identification and verification technology has the advantage of higher security with regard to vehicle access control and ignition immobilizers. It can also be dapted for use in driver comfort/convenience applications. The fingerprint of course makes the person’s identity known to the vehicle, so it can adjust the driver's seat and rearview mirror position, dial in his or her favorite radio station and even call up his or her personal mobile phone directory automatically. To authorize individuals for access, biometric data is entered to the system via standard communication and information systems in use today. In other words, the underlying infrastructure is already available.

By virtue of its miniaturized components and low power consumption, the IKENDI fingerprint solution may also be integrated into the car «key», which then could be used for other identification applications (for example, to unlock the front door at home). This means fingerprint recognition systems could in the very near future take the place of the ubiquitous key ring by converging many keys into a single universal «key». The danger of lost keys being misused for theft and the like would also be averted for good. The possibilities are many, and the technology is there for the asking.

©IKENDI® 2001. All rights reserved.

EDITOR'S NOTES IKENDI®
…the identification company

having the roots in image processing technology, IKENDI® develops fingerprint recognition software and distributes fingerprint development tools for embedded systems and Smart-card integration in a combination with training, consulting and support. IKENDI® only licensees software and Hardware IP and does not have an end product development. The main focus is on OEM, System supplier and Integrator for mass applications in consumer and security markets wherever fingerprint is a solution for more comfort and security.

IKENDI Software AG
Argelsrieder Feld 11
D-82234 Wessling
Germany
Email: info@ikendi.com

ATMEL
Founded in 1984, Atmel Corporation is headquartered in San Jose, California with principal manufacturing facilities in Colorado Springs and Irving, USA, Nantes and Rousset, France, Newcastle in the UK and Heilbronn, Germany. Atmel designs, develops, manufactures and markets on a worldwide basis semiconductors, including nonvolatile and embedded memory, logic, analog, microcontroller products and system-level integration (SLI) solutions using advanced CMOS, BiCMOS, BiPolar and SiGe process technologies.

Further information can be obtained from Atmel’s Web site at www.atmel.com .
Contact: Peter Bishop, Communications Manager, Atmel
Rousset, France, Tel: (+33) (0)4 42 53 61 50
Email: pbishop@atmel.com

Atmel, the Atmel logo and combinations thereof are trademarks of Atmel Corporation.ARM, ARM7TDMI, ARM720T, Thumb and ARM Powered are registered trademarks of ARM Limited. ARM922T, AMBA and Embedded Trace Macrocell are trademarks of ARM Limited. All other trademarks or product names are the property of their respective owners. “ARM” is used to represent ARM Holdings plc (LSE: ARM and NASDAQ: ARMHY), its operating company ARM Limited, and the regional subsidiaries ARM, INC.; ARM KK; ARM Korea Ltd.; ARM Taiwan.

Back to ARM Powered Products in Embedded Solutions