Alpesh Saraiya — Senior Director, Product Management — Intrinsic ID |
“A billion here, a billion there, and pretty soon you’re talking about real money”
Everett Dirksen, a U.S. politician noted for a florid oratory style, commenting on the need to control government spending
A hundred million. That’s a big number. And for Intrinsic ID a milestone marking progress in multiple ways.
As we announced recently, more than 100 million devices have been deployed and are protected by Intrinsic ID hardware root of trust technology. It’s a great indicator of the confidence our customers have entrusted in our company that they have used Intrinsic ID to protect so many products. It’s also an indicator of just how robust our foundational security technology is.
Failure Rate Analysis
In the lifecycle of any device, the failure rate over time typically follows the bathtub curve (see Figure 1). At the initial stages of a product’s lifecycle, you typically see an increased failure rate due to early “infant mortality” failure. This is essentially the stage where you get the kinks out of the system. You’ll see some random, and some not-so-random, phenomena that cause devices to fail. Devices that survive the infant mortality stage typically will operate with a predictable reliability at a low, and more or less constant, failure rate. When products approach the end of their useful life we see failure rates tend to increase due to wearout. This reflects the other end of the bathtub curve.
Let’s apply this model to reaching deployment in 100 million devices. We designed and developed the hardware root of trust IP using a two-pronged approach. First, we analyzed the mathematics, physics, semiconductor and real device behavior. Second, we modeled the IP by way of empirical analysis in the lab and through accelerated aging and testing. We have also worked closely with our customers on their own end products’ accelerated aging to come to the level of confidence we have in terms of reliability and resilience. And we conduct ongoing monitoring of customer-deployed devices to gauge performance in the field.
SRAM PUF behavior is resilient and has been tested on a wide variety of process nodes and under a wide variety of externally applied conditions such as temperature, voltage variation, radiation and aging. For example, we have measured temperature variations covering military and automotive temperature ranges (-55 Celsius to +150 Celsius). Supply voltage variations of plus and minus 20 percent of Vdd have been covered. Key extractor algorithms inside our products using SRAM PUF have been designed to provide an error rate of less than 1 in a billion in worst-case conditions, i.e. extreme temperatures.
This all speaks to the quality and reliability that we are standing behind these 100 million units. This number is, of course, a snapshot in time, and is growing at an exponential rate. Also significant is that these 100 million devices are implementations from a variety of manufacturing nodes and fabs.
Passed Early Mortality
In the context of the failure model, we have clearly passed the early mortality phenomenon and are well into the deployment stage of the active useful life of the technology. At Intrinsic ID we have characterized the useful life of our technology to be at least 25 years – likely exceeding the expected useful life of the end product itself.
Proven in Top Manufacturing Environments
IoT products utilizing microprocessors based on a range of technologies, from older technology at 350 nanometers to newer designs at 7 nanometers – the most aggressive geometry currently in production – utilize Intrinsic ID security. The top semiconductor fabs are represented and we have customers shipping in a variety of configurations.
Other vectors of confidence that contributed to deployment in 100 million devices are the variety of certifications and applications. Intrinsic ID security has been deployed in products certified by Common Criteria, EMVCo and Visa. Products in which we’ve shipped have been in the payments arena, as well as government and military – meeting the very high security qualification bars required “from banks to tanks.” We’ve also worked with independent labs to qualify the claims that we make here.
These accolades are coupled with the fact we are designed in at three different points in the supply chain, where each has its own qualification criteria, and in many cases their own internal security center of competency. We’ve been assessed by four of the Top 5 MCU vendors in the world as meeting their standards for security. And we have met the needs of module manufacturers such that they are able to supply OEMs with strong hardware root of trust security on a variety of semiconductor platforms in relatively quick fashion, thereby enabling a versatile multisource vendor strategy. For direct engagement with OEMs we have been able to introduce the benefit of quick time to market with the software implementation of our hardware root of trust.
All these intersection points of our product to the marketplace have helped to contribute and accelerate achieving this milestone of 100 million shipped.
This early maturity stage of the mortality fallout leads us to conclude that, to put it bluntly, if we were going to have problems we’d have had them by now. So now we’re in this long stretch of the deployment period where the connected product would fail for any of a thousand different reasons before it would fail because of our IP.
We’ve gotten through this stage in any product deployment where early mortality would be expected to occur at a higher rate. We’ve gotten through this with very low mortality – actually zero. It doesn’t degrade. And in fact just the opposite: it gets better over time, despite the fact the end device as originally designed may be degrading in pure function or form. You can see results of our experiments on anti-aging technology that showed SRAM PUF actually grew more reliable over time.
And so this aspect of our root of trust, is like fine wine – with time, it only gets better.
Alpesh Saraiya has served in senior product roles at some of the world’s top electronics companies. His experience managing secure, connected products includes leading the webOS Core for LG Electronics’ Smart TVs and other connected IoT applications such as smart home, smart car, wearables and mobile. He held senior marketing roles at Broadcom and C-Cube Microsystems, as well as R&D positions at IBM. Alpesh holds a Master of Science degree in Computer Engineering from Syracuse University and a Bachelor of Science degree in Electrical and Computer Engineering from the University of Tennessee, Knoxville.