By Dave DeFusco
A team of researchers that includes Dr. Yucheng Xie, assistant professor in the Katz School’s Graduate Computer Science and Engineering, has developed a new technology called “mmPalm,” which uses millimeter wave signals to create an ubiquitous, low-effort authentication method through palm recognition, aiming to secure environments like hotel entryways, apartment buildings and even vehicle settings.
This system, detailed in the paper, “mmPalm: Unlocking Ubiquitous User Authentication Through Palm Recognition with mmWave Signals,” allows for user authentication by scanning a person’s palm with millimeter wave signals. Unlike traditional biometrics that rely on fingerprints or facial recognition, mmPalm is a low-cost alternative that could revolutionize security access in public spaces like smart cities and smart homes.
The study recently received the Best Paper Follow-Ups award at the IEEE Conference on Communications and Network Security (CNS), a prestigious IEEE conference in cybersecurity. IEEE CNS attracts leading researchers, industry experts and academics from around the world who are contributing cutting-edge work in protecting communications networks and systems.
Traditional biometric methods, such as fingerprint and face recognition, often require expensive hardware to install and maintain, limiting their widespread use. The mmPalm approach takes advantage of the mmWave technology widely used in WiGig and 5G networks to identify individuals based on the unique pattern of their palms. This technique offers a low-cost solution that can be easily deployed in public and private spaces.
“The technology is promising because of the fine detail that mmWave can capture,” said Dr. Xie, first author of the paper who collaborated with researchers from George Mason University, Rutgers University, Temple University and New York Institute of Technology. “Each individual’s palm has a unique combination of shapes, skin thickness and texture, making it as distinct as a fingerprint.”
The mmPalm system captures these characteristics by sending and analyzing reflected signals, creating a distinctive “palm print” for each user. This advanced technology also builds virtual antennas to further increase the spatial resolution of a commercial mmWave device, capturing subtle differences in each palm print.
For users, the process is as simple as showing a hand to authenticate their identity. The device transmits frequency-modulated waves that interact with the palm, with mmPalm analyzing the reflected waves for specific traits. This reflection data is processed to form a unique biometric profile, which is compared to stored profiles to verify identity.
Beyond its cost-effective nature, mmPalm also addresses challenges that often arise in authentication technology, such as distance and hand orientation. The system includes a method to “learn” various palm orientations and distances using a type of artificial intelligence called a Conditional Generative Adversarial Network (cGAN), which generates virtual profiles to fill in gaps. Furthermore, the system employs a transfer learning framework to adapt to different environments, so mmPalm works reliably in various settings.
Testing mmPalm with 30 participants over six months showed a remarkable 99% accuracy rate, with high resistance to impersonation, spoofing and other potential breaches. As cities and homes increasingly adopt smart technology, mmPalm could pave the way for secure, contactless user authentication—whether to enter a hotel room or adjust settings in a connected car.
“By harnessing high-resolution mmWave signals to extract detailed palm characteristics,” said Dr. Xie, “mmPalm presents an ubiquitous, convenient and cost-efficient option to meet the growing needs for secure access in a smart, interconnected world.”
