Why is IDEX TrustedBio the only fingerprint sensor that can work directly with existing SEs while our competitor’s silicon sensors require a BioSE or an extra MCU and separate PMU?
Why is IDEX TrustedBio the only fingerprint sensor that can work directly with existing SEs while our competitor’s silicon sensors require a BioSE or an extra MCU and separate PMU? Let’s investigate why a BioSE is needed by silicon sensors, but not needed when paired with TrustedBio. There are three major capabilities that distinguish a BioSE from a standard SE – memory, energy harvesting/sharing, and processor speed.
Silicon sensors capture a fingerprint image which is then transferred to another MCU for biometric processing. Depending on the sensor size and resolution, the 8-bit greyscale image is in the range of 14K to 20K bytes. SEs have two types of memory, flash memory which is a relatively slow permanent type of storage for the software program and fingerprint templates and RAM which is used for data during program execution. Standard SEs have plenty of flash, but only about 10K to 20K bytes of RAM.
For an SE to work directly with a silicon sensor there needs to be enough RAM to accept the entire fingerprint image AND the biometric matcher AND the normal SE functions such as the card operating system and POS terminal communication. For standard SEs there is typically much less than half of the RAM available for biometric processing. As one can see from the math, getting it all to fit is a problem and explains why all Gen 2 cards have a separate MCU.
I manage a software engineering department and I know how ingenuous software engineers are at optimizing memory usage. However, no executive wants to start a new biometric card project and be told, it “might” fit. Particularly when there will be new features and requirements such as anti-spoof detection that will undoubtedly require more software and memory. Moreover, the techniques used to reduce memory usage often result in execution speed penalties.
On the other hand, IDEX TrustedBio needs very little RAM on the SE. TrustedBio runs biometric feature extraction using its own MCU with its own RAM creating a data template that is less than 750 bytes, more than 20 times smaller than the original image. These few hundred bytes are encrypted and transferred to the SE for a secure match against enrolled templates. Current generation SEs have enough RAM to accept encrypted feature data, but not the entire fingerprint image. One extra benefit, data transfer is much faster, 1 ms vs 20 ms since only a few hundred bytes are transferred. A second extra benefit is that this method is even more secure since the fingerprint image never leaves the sensor, only a small encrypted data template which makes it impossible to recreate the original fingerprint.
The 2nd important capability is energy harvesting and energy sharing. Contactless cards have an antenna that is used to harvest power when near a card reader. Energy harvesting and sharing is done by a PMU, Power Management Unit. There are 4 options to power a biometric card.
- Option 1: A separate PMU chip that connects to the antenna and powers both the SE and sensor. This is a typical configuration for Gen 2 biometric cards but adds cost and complexity.
- Option 2: SE connects to the antenna, harvests the energy and then provides power to the sensor through a power pin. Contactless SEs can harvest power for themselves, but few can also share power with the sensor. Full BioSEs support this option, an important difference with standard SEs.
- Option 3: SE and the sensor both connect to the antenna and harvest their own energy. They coordinate and take turns using the power. TrustedBio is the only sensor that can do this.
- Option 4: Sensor connects to the antenna, harvests the energy and provides power back to the SE via a power pin. TrustedBio is the only sensor that can do this.
Silicon sensors are unable to harvest their own energy so options 1 and 2 are the only viable options and just like the problem with memory, there are very few SEs that can harvest energy and share it with the sensor. On the other hand, TrustedBio has its own integrated PMU and supports all the options without any additional components.
The 3rd topic is MCU speed. The time to run the biometric matcher algorithm is directly correlated to processor speed and is one of the biggest contributors to overall biometric card performance. The goal is to complete an entire contactless card operation in less than a second and preferably much less than a second.
The SE MCU and even low power external MCUs have maximum processor speeds in the range of 60 to 100 MHz, however, max processor speed is only part of issue. The real question is how fast is the MCU in low field strengths when a card is approaching a card reader. BioSEs are designed to be more power efficient, i.e. faster at low power levels. TrustedBio takes this capability to yet another level. Because it is highly optimized for biometric cards, TrustedBio MCU can operate at substantially higher speeds across the entire field strength range.
Let’s summarize, BioSEs are designed with more memory, ability to harvest and share energy, and faster MCUs at lower field strengths. Today there is only one announced BioSE card, the Idemia/Zwipe/IDEX card. Until a new generation of BioSEs are available, our competition will require “other stuff”.
In contrast, TrustedBio has its own very fast and power efficient ARM MCU with a lot of flash and RAM. It also has its own very flexible PMU. These are the features that distinguish a BioSE from a regular SE and are required by silicon sensors. Because TrustedBio is implemented in an advanced 40 nm silicon process, even with all these capabilities, our ASIC is much smaller, less than 25% of our competitor’s silicon sensor ASIC and less expensive. As an additional benefit our ASIC uses a polymer sensor that is 250% larger than our competition, providing better biometric performance.
There have been many pilots with Gen 2 cards that have provided invaluable insight into the market and technology. Some, including IDEX and competitor’s sensor Gen 2 cards, will ramp up into volume production. However, it is clear that low-cost Gen 3 cards will be the ones that enable mass market adoption.
This past year has seen two highly innovative and disruptive solutions. Idemia/Zwipe have accelerated the market by focusing on one end of the card and developed a single chip BioSE. IDEX has focused on the other end of the card and developed a sensor that works equally well with BioSEs and regular SEs. Our two technologies work extremely well together as we leverage each other’s strengths and optimize a high-performance low-cost card for their customers.
At the same time IDEX is accelerating adoption with other customers and the rest of the world. When we introduce customers to the technology and performance of TrustedBio they instantly understand that TrustedBio is unlike any other sensor. They immediately understand that it is much more than a fingerprint sensor and that it will greatly simplify the card design and drastically reduce cost.
While it might seem that IDEX is a small company with little experience. The reality is that our 100 person engineering team is highly experienced with decades of work in mixed signal analog and digital ASIC design, sensor design, system engineering, modeling and simulation, software and algorithm development, biometrics, encryption, material science, high volume production testing, supply chain management and much more. Most have joined IDEX in the last few years as we recruited world class talent around the globe. We have major design centers in the US, UK and Europe with technical support throughout the world including Europe, US, and Asia.
Several years ago, when IDEX refocused 100% on biometric cards, we recognized that accelerating the biometric card market required substantially reducing card costs by eliminating the “other stuff” on a card. We did not want to wait for a new generation of SEs. We intentionally set out to completely change the technology for biometric cards and not just make another fingerprint sensor.
TrustedBio is our 3rd generation sensor developed in an advanced 40 nm mixed signal silicon process. We use multiple Tier 1 suppliers, for example, TSMC for the ASIC and Amkor for sensor assembly and test, so we can rapidly scale to very high volumes. TrustedBio is a disruptive sensor that is several years ahead of anything else in the market, dramatically cutting costs, simplifying manufacturing and providing outstanding biometric performance. No other fingerprint sensor can accomplish these goals.
All of us in the biometric card business should welcome these rapid developments. The biometric card market is huge. By enabling low cost, high performance and easy to manufacture biometric cards, the market will start accelerating in 2021.