Signal Processing: Finding Weakness in Strength

Extracting faint signal data amid powerful measurement interferences may save both time and, in many tech spaces, lives

3 min read

Whether you're piloting an aircraft through thick fog, searching for a lost child at the mall, or just trying to find that key presentation room in a convention center, emerging signal processing technologies can mean the difference between success and failure. And for IEEE Fellow and Stevens Institute of Technology professor Dr. Hongbin Li, failure is not an option.


Li is developing innovative solutions to detect signals of interest and reject interference, exposing and extracting more accurate information from noisy, limited measurements. His research focuses on mathematical issues of knowledge-aided weak signal detection. It's a narrow focus, but the applications are wide-ranging—not only in obvious areas such as radar, but also in wireless communications, bioinformatics, hyperspectral imaging, radio astronomy, border control, nuclear power plant monitoring, and more.

For example, you may not think of oceans, trees, or buildings as clutter, but on airplane radar these static surfaces can produce interference that overwhelms the signal of interest, jeopardizing the aircraft's safety as it navigates through dense fog, over water, or near potential ground threats. 

"In an airborne environment, when you have a sensor looking down for objects such as vehicles, your sensor echoes with reflections – not just from the target, but also from the ground, trees and man-made objects like roads and bridges," Li explains, "and they’re often much stronger. It's called clutter. To expose and detect weak signals in the clutter, you have to first reject the extraneous interference. And that's just what we're doing – developing signal processing algorithms and exploiting resources such as digital maps and databases of the environment, as well as statistical and computational tools." 


Not surprisingly, Li's work involves a lot of trial and error – and trial again. "Sometimes we use parametric models to influence model-based solutions," he says. "Sometimes we look at inherent structures on the ground to complement the limitations of data. On a plane we have limited data because the sensors are moving, so from the sensors’ perspective everything is moving and they get flooded by the clutter. But, by using certain geometry, we can distinguish moving objects from stationary ones. Combining joint spatial recognition and Doppler processing helps us differentiate the target — even though the reflection from the target is much, much weaker."

Li was able to hone these methods courtesy of two high profile awards: a $350,000 research grant from the U.S. Air Force Office of Scientific Research to research adaptive signal radar detection in 2016, and a $330,000 grant from the National Science Foundation to explore passive radio frequency sensing technologies for civilian applications in 2018.

"For radar, you need an active high-power transmitter. But those emit radiation and no one wants the environment to have too much radio frequency radiation," Li says. "For our NSF project, we're not using any active transmission. Instead, we're using existing wireless sources such as radio, TV, cellular towers, satellites and GPS. Just as you use a flashlight to illuminate a room, we're using existing sources to illuminate the object we want. It's there – you just have to listen."

Of course, it's a little more – OK, a lot more – complicated than that.

"The challenge is to identify the target and differentiate it from everything else," Li says. "Passive signal processing for our considered set-up is especially challenging, because the signals are designed to directly solve communication challenges, not indirectly support finding exhibits in a convention center or reuniting a lost child with their family. In order to do those tasks, you have to suppress direct path transmissions. One solution to suppressing those transmissions is using an antenna array to perform selective receiving. Selective receiving enables a search of a surveillance area while forming nulls – think of it like putting earplugs on the signal – toward the directions of direct path transmissions. More sophisticated techniques than this are needed in complex environments, however."

"Radar is often expensive," Li adds, "but these passive technologies, which can be implemented in mobile devices such as cellphones are far more economical. For example, wireless service providers can offer wireless devices with built-in passive sensing functions, such as home security, indoor localization, health monitoring (i.e., heart/breathing rate monitoring by measuring the micro-Doppler effects of the chest movement, fall detection, etc.) for babies and elderly people, etc., and make them broadly available to any user."

"The potential economic and societal impacts are huge."

Professor Li joined Stevens Institute of Technology in 1999. His research interests include statistical signal processing, wireless communications and radar. He is an elected member on the IEEE Signal Processing Theory and Methods Technical Committee and Sensor Array and Multichannel Technical Committee, and an associate editor for IEEE Transactions on Signal Processing and Elsevier Signal Processing. His honors include the 2013 IEEE Jack Neubauer Memorial Award for the best systems paper published in IEEE Transactions on Vehicular Technology.

The Conversation (0)

How Police Exploited the Capitol Riot’s Digital Records

Forensic technology is powerful, but is it worth the privacy trade-offs?

11 min read
 Illustration of the silhouette of a person with upraised arm holding a cellphone in front of the U.S. Capitol building. Superimposed on the head is a green matrix, which represents data points used for facial recognition
Gabriel Zimmer

The group of well-dressed young men who gathered on the outskirts of Baltimore on the night of 5 January 2021 hardly looked like extremists. But the next day, prosecutors allege, they would all breach the United States Capitol during the deadly insurrection. Several would loot and destroy media equipment, and one would assault a policeman.

No strangers to protest, the men, members of the America First movement, diligently donned masks to obscure their faces. None boasted of their exploits on social media, and none of their friends or family would come forward to denounce them. But on 5 January, they made one piping hot, family-size mistake: They shared a pizza.

Keep Reading ↓Show less