What does Leidos do? Part 6: Sensors, Collection, and Phenomenology
What does Leidos do? Today's interview explores our work in sensors, collection, and phenomenology (SCP). We interviewed Tim Cunningham, who leads this technology area for the company. Cunningham spent 20 years in the Air Force specializing in SCP capabilities, including space-based signal detection. In his six years with Leidos, he has directed business development and strategy for the company's sensors and phenomenology operation, and played a lead technical role in captures.
Here's what he had to say.
What is sensors, collection and phenomenology, and how do these domains work together?
In general, sensors, collection, and phenomenology (SCP) is the art of understanding the world through the physics of sensing, and how you might perceive those signals. Next is the mathematics of processing what you can actually measure with a sensor to eliminate the noise and background environmental fluctuations. Finally, it involves getting the pure signal and understanding the originating physical phenomena behind that signal. This competency area also extends to how Leidos implements multiple sensor modalities and integrates collection systems to meet demanding customer needs.
, Sensors, Collection and Phenomenology Lead
Often the signal we're trying to detect is so faint, and there's so much noise in the environment, that you just have to laugh. Yet the real problems our experts are able to solve often give the same sense of disbelief.
Give us a sense of how challenging this work is.
Imagine your job is to identify the species of a tree in a big grass field, blindfolded, by trying to detect the dew on the ground in the morning under that tree, in the middle of a monsoon. That's a paradigm for the tough problems that our signal processing experts are able to overcome. Often the signal we're trying to detect is so faint, and there's so much noise in the environment, that you just have to laugh. Yet the real problems our experts are able to solve often give the same sense of disbelief.
What exactly is 'phenomenology?'
Let's say I'm trying to detect the presence of a human being inside a building with no windows. What are all the phenomena that might indicate a person being somewhere? People breathe, so one phenomenon is that their chest might expand and contract at their breathing rate. That's at least worth considering as a signature of a person being in the building. Human beings also emit modest amounts of heat. They emit moisture through perspiration and respiration, so they typically add humidity to an environment. These are ways we might detect a person in a building remotely. When we consider the range of phenomena that an object of interest generates, we're engaging in phenomenology.
In this example, using certain frequencies or radar, our experts are able to "look" through walls by detecting signals. As you might imagine, there's a lot of interference in trying to collect signals in such an environment where there are other phenomena in the room, like air currents, that might mask a signal. But, even in the face of lots of noise, if we process all the data correctly, we can do "true" detections and suppress false alarm rates.
Why was this capability designated as a technical core competency of Leidos?
Our people are incredibly good at it. We do it in every sensing modality that I'm aware of. We do everything from electric and magnetic field sensing to seismic, acoustic, radar, light detection, and ranging (LIDAR), ultraviolet, hyperspectral imaging, and every kind of electro-optic and infrared imaging imaginable. We have absolutely world-class sonar capabilities, and we’re at the bleeding edge of radiological detection. We really span the map. Further, we are often able to bring multiple modalities to bear on a single problem, and by correlation, can draw conclusions that one signal may not enable. And we can integrate and field systems to address really difficult problems for a range of defense, intelligence, and civil customers. So when a customer has a mission problem and some component of that is a requirement to sense the environment, we can offer a best-in-class solution.
Give us one practical application of this capability.
Let's say you're a technical battlefield commander, and you're faced with an enemy who is positioned in heavy woods. You know they have combat ground vehicles, but you don't know exactly how many, what type, and where they are. If you were to fly over that wooded area, and you had a sensor that enabled you to detect all the big metal objects, you would look at that data and know where those objects are. But you'd also like to understand what class of object they are, whether each one of them is a tank, jeep, artillery piece, or command post.
Now let's say that I understand which of those big metal things are tanks, Humvees, command posts, and artillery pieces. Identification would go further and say, "This tank is a Russian T38," or "That command post is for an anti-air battery," or "This is a Humvee, and it's got some really interesting enclosure on the back of it." Now we're starting to identify for the commander exactly what threat he's facing in this environment so that he can respond to it appropriately in real-time.
That's a military context, but we do quite a bit of work in other markets. One of them is in a field called CODIS, or identifying DNA. There are vast numbers of other areas where our experts apply this art.
What makes Leidos so good in this area?
Leidos takes a disciplined, science-based approach that distinguishes what we can typically accomplish, even on really difficult SCP problems. We drive to understand the physics, or biochemistry, or whatever the underlying science may be of the object of interest, and of the environment. Whatever we're trying to sense, we try to understand everything about it at the most fundamental level possible. Then, with that same drive, we look at all the technology that is capable of sensing the phenomena of interest in that environment, and we understand the physics of each method of detection. This allows us to apply, or when needed develop and apply, the best detection tools to the problem. We model the entire process and assess how well we're doing. If it's not good enough, we will collect signals in multiple modalities, taking advantage of their joint contributions to improve performance at the system level. Finally, we drive to understand and apply all of the mathematical techniques that might be advantageously used against the signal(s) we have collected, to derive actionable insight.
Our key performance metric is the ratio of the sensitivity to a signal divided by the false alarm rate of our sensing process. You can always make a sensor more sensitive, but it usually comes at the penalty of perceiving more false alarms. A false alarm, or false detection, is when you think you've seen the signal you're looking for, but in reality, it was an artifact in the background, and not really your signal. We work very hard to minimize false detections. Our art is increasing the ratio of sensitivity divided by false alarm rate beyond what most of our competitors are capable of doing. This is a metric that ties together our capabilities, even across the many modalities and domains in which our experts focus.
How did the company make a name for itself in this domain?
Several years ago, working with a San Diego biotech firm, we developed a laboratory system that was able to analyze DNA from environmental or human samples using DNA amplification and mass spectrometry. It was and remains new technology. Our machine was able to not only detect and classify, but also identify the strain of anthrax that had been mailed to Congress in 2001. That identification was good enough to track the anthrax sample to the specific lab from which it originated. Leidos experts were pioneers in that art. It won a 2005 R&D 100 award, the 2006 Innovation Award from the Association for Laboratory Automation, and the Wall Street Journal’s 2009 Gold Medal Technology Innovation Award. Later, we developed a revolutionary acoustic sensor that sits on the ocean floor to detect different things in the water column. We also created a tsunami buoy detection system for monitoring worldwide seismic events.
More recently, we won a program to develop and integrate new sensors that the Army wanted to fly in their fixed-wing aircraft. In order to win that program, we had to know everything about the sensors—not just how they sensed, but their mechanical robustness and how they would operate and survive in an aircraft environment. And we had to know about the phenomenology of what we needed to detect on the ground.
We develop new signal processing architectures and algorithms that increase performance while reducing SWAP (size, weight, and power) of our systems. For example, Leidos was the first company to re-formulate Space-Time Adaptive Processing algorithms to optimize them to run on graphics processing units (GPUs) that do the computationally complex pipeline processing for high-end graphics cards used by gamers. That resulted in a factor of 100 improvement in radar processing density. We're currently working on an approach that will yield another factor of 10 improvement in many applications.
These are examples of what Leidos SCP practitioners do every day, in just about every application domain imaginable. When you do that consistently for a few decades, word gets out.