UnifyID Passive Authentication

In situations where people use masks and gloves, or face coverings such as burqa or niqab, typical biometric authentication methods like face and fingerprint recognition do not work. Moreover, lower end phones do not have the necessary hardware for strong biometric authentication via face or fingerprint; face images are easy to steal and spoof, which is why Face ID requires a specialized high-end camera. However, all smartphones have an accelerometer, and we each move in a unique way that is very difficult to spoof. UnifyID solves re-identification and authentication challenges by leveraging the sensor data from smartphones to uniquely identify, validate and authenticate an individual. We have developed algorithms that can authenticate a user based on their unique gait that runs on any smartphone that includes a motion sensor, with accuracy that rivals a physical fingerprint.

UnifyID solves the identity verification and individual authentication problem by leveraging the sensor data from smartphones to uniquely identify, validate and authenticate an individual. In the middle of a pandemic when many people are sheltering in place, the digital transformation process is accelerating and millions of people still need to conduct business, make payments and execute transactions securely. UnifyID’s technology can enable these transactions in a secure way without adding friction to the flow by authenticating people based on their common behaviors they are already performing. This is a global scale project affecting millions of people. The technology is accessible, inclusive and user-friendly; as it does not require people to learn any new behavior or piece of technology for it to work, they just need to be themselves and the system works in the background. It even works in places with spotty internet connections as the model is retained on the device.

The target population is any end-user utilizing a mobile smartphone for financial, eCommerce transactions or login activities that need a layer of security without adding friction of complications to the user experience. For use cases where even low-end smartphones are not practical, we also have a design for an inexpensive device that can capture and validate human motion data for authentication purposes.

Our solutions align to the challenge countries have of digital authentication by offering an inclusive authentication solution via behavioral biometrics that work on a larger spectrum of smartphones (including low-end ones which still have sensor such as accelerometer and gyroscope) as well as in places with unreliable Internet connectivity because the authentication model stays on the device.

Out solution is a highly innovative approach to authentication. Most biometrics are static, easy to compromise, and require expensive specialized sensors. Our solution is dynamic and works with the billions of accelerometers already deployed in the world.

There are other methods for authentication, including biometrics such as face and fingerprint recognition. Moreover, there are other behavioral biometric solutions in the market. That said, there is no other behavioral biometric solution that can rival the physical fingerprint in terms of accuracy and can be appropriate used as an additional factor in user authentication without adding additional friction. It also works in cases where other biometrics like fingerprint and face recognition are not applicable or appropriate.

As seen in the videos as well as funding received by top investors such as NEA and Andreessen Horowitz, as well as winning of accolades such as being a TechCrunch Disrupt Runner Up, SXSW Security & Privacy Winner, MIT AI Idol winner, named a Gartner Cool Vendor and winner of the RSA Innovation Sandbox, our technology not only works but it is recognized by many as a top innovation.

Here is a sample of peer-reviewed publications covering different aspects of the core human kinematics technology:

- Vulnerability of deep learning-based gait biometric recognition to adversarial perturbations, CVPR 2017
- AcctionNet: A Dataset Of Human Activity Recognition Using On-phone Motion Sensors, ICML 2017
- SIMUni: Sampling Impostors from Misfit Universal Background Models in accelerometric gait biometric verification, BayLearn 2018
 - Classifying humans using deep time-series transfer learning : accelerometric gait-cycles to gyroscopic squats, KDD 2019

Activities: Leveraging sensory data from smartphones to uniquely identify an individual

Outputs: a model for authenticating individuals

Short-Term outcomes: Can be used as an additional identification and authentication factor for a large population in emerging economies even with lower end smartphones as they all have accelerometers

Long-Term outcomes: Could replace the need for passwords and become a more inclusive way of authenticating individuals without the need to have an expensive phone or remove traditional outfits for face recognition to work.

The UnifyID solution can play a role as an authentication factor in both 2FA (two factor authentication) and MFA (multi-factor authentication) solutions. One of the differentiating advantages of UnifyID’s solution is that it does not only add a layer of security by adding an authentication factor based on motion, but also that this authentication factor does not add friction to the end user experience since there is no additional step that the end user needs to perform for the authentication to work: The end user needs to move as they always move and the technology will authenticate them based on their movements.

UnifyID’s user friendly solution can be included as an additional authentication and identity validation layer in any current authentication workflow as an additional security factor without adding friction to the end user as it runs continuously in the background.

Our technology can be seen as an additional security factor in current multi-factor authentication solutions and, as such, it is interoperable with current solutions that allow for adding additional authentication factors into their workflows. It is compatible with OpenID Connect and OAuth 2.0, and is architected to be a FIDO-compliant authentication factor.

The beauty of our solution in this regard is its inclusiveness: While not all smartphones have “fancy and expensive” face recognition technology or physical fingerprint readers embedded in their hardware, even the lowest end smartphones have an accelerometer. Since UnifyID’s technology mostly relies on the accelerometer sensor readings, UnifyID’s technology is inclusive and works in environments where the latest technology is not available. 

Unlike some of the other authentication mechanisms, it does not require reading or literacy, or even eyesight. As long as the user is able to walk, they can authenticate.

Moreover, since the motion print stays on the device, the technology works well and is well suited for low connectivity environments.

Millions of devices served currently. Tens of millions in one year. Over 100 million in five years.

We are passionate about enabling strong authentication for everyone, in a way that takes human behavior into account. The current forms of authentication are too cumbersome, too insecure, and don't take the human element into account. We want to make authentication more seamless, more secure, and more human for everyone.

Demonstration and acceptance by the market that gait and motion is as accurate as a fingerprint in terms of authentication, as well as allowing sensor readings from the smartphone for this purpose by the end user.

By educating the market via white papers, conferences and other content and by partnering with other organizations which already have and need these end user permissions for other reasons, such as financial institutions who also need this additional level of authentication.

25 full-time employees

5 contractors

Key team members:

John Whaley is Founder and CEO of UnifyID. He is a renowned expert in digital security with over 15 years of experience in the field. Prior to UnifyID, he was the Founder and CTO of Moka5, a desktop virtualization company. Before that he worked at IBM’s T.J. Watson Research Center and Tokyo Research Lab on static program analysis and just-in-time compilation. He was also a Visiting Lecturer in Computer Science at Stanford University.

John received his bachelor's and master's degrees in Computer Science from MIT and his Ph.D. in Computer Science from Stanford University, where he won the Arthur L. Samuel Thesis Award for Best Doctoral Thesis.

Mike Morris has over 20 years of experience working in, building and leading world-class engineering teams creating visionary new products for enterprises and consumers.

Prior to UnifyID, Mike was VP of Engineering at Proofpoint (via acquisition of NetCitadel), where he led the Threat Research and Engineering group. Before that, Mike helped to grow VMware from a startup with 80 engineers and one product to an 18,000 person company delivering dozens of products to millions of enterprise, SMB and consumer users.

Vinay Prabhu is currently the Chief Scientist at UnifyID. He has more than 30 peer reviewed publications in areas spanning Physical layer wireless communications, Estimation theory, Information Theory, Network Sciences and Machine Learning. He holds a PhD in Electrical & Computer Engineering from Carnegie Mellon University and an MS in Electrical Engineering from Indian Institute of Technology-Madras.

Working with FinTech companies, Fortune 500 and public companies on their digital security needs to their digital applications by both employees and customers. In addition, working on some physical access use cases using the phone as a digital key to open doors, such as in retail and automotive.