Machine learning, GPS alternatives key for navigating future jammed environments

Story

November 30, 2022

Ask anyone involved in the high-conflict areas of Ukraine, and they will tell you just how big of a problem it is when a major military power unleashes its jamming capabilities on GPS access over a large area.

Early reports during the war indicate that Russia was quick to use its electronic warfare (EW) technologies to disrupt GPS, which the United States and allies rely on for navigation of military assets. Because of this vulnerability to jamming, many in the industry are taking a hard look at machine learning to mitigate the problem, and also examining alternatives to GPS.

Sean O’Hara, director of machine intelligence and autonomy and a fellow at defense research company SRC (North Syracuse, New York), says that a major contributor to the challenges of operating in GPS-denied environments is understanding the tactical situation and responding accordingly.

“Solutions for reestablishing navigation and timing are highly situational, depending on the operational mission context; the local operating environment; how GPS is being denied, degraded, manipulated, or otherwise affected; and the tactical and strategic resources available to support reestablishment of position, navigation, and timing capabilities at the edge sensor/platform,” he says.

O’Hara notes that today’s autonomous and semi-autonomous systems already do a good job when it comes to leveraging multiple subsystems in numerous domains to deal with this challenge. That level of success doesn’t mean they’ve solved the problem, however.

“This is both a blessing and a curse, as over the last several decades our adversaries have consistently and quickly weaponized advanced commercial technology,” he says. “This trend is holding today, especially with regards to robust autonomous navigation systems and low-cost loitering munitions.” (Figure 1.)

[Figure 1 | SRC’s PROTEAN system is a multimission RF suite of systems. (Photo courtesy SRC)]

Leveraging machine intelligence

Machine intelligence (MI) plays a critical role in addressing this challenge, and innovation in this area is rapidly increasing with new capabilities emerging sometimes in as little as a few months, O’Hara says.

“One foundational area supported by machine intelligence is in the application of deep, or high dimensional, sensing to assist with situational understanding,” he says. “This involves using deep learning approaches, often across multiple sensing domains (radio frequency, computer vision domain, and others), to assist in determining the current operational ‘state.’”

This state, he asserts, consists of the situational understanding elements that estimate where the platform or sensor is, and whether the current operational environment is challenged or unchallenged.

A second area where MI methods come in handy is to adapt and optimize the operational capabilities of sensors, platforms, and weapons in GPS-denied environments. This could happen locally within the platform or sensor itself if it is disconnected from communications, or it could be orchestrated by “assets at a higher level tactical or strategic echelon,” O’Hara adds.

“This approach of situational understanding and state estimation to dynamically adapt the solutions can be robust and effective,” he says. “It may be prohibitive to provide perfect/full capabilities everywhere all the time. However, it is very possible to deploy solutions that provide the level of mission capabilities that are needed, where they are needed, and when they are needed.”

New possibilities from Starlink and 5G

One interesting recent development is SpaceX’s Starlink, a constellation of satellites that provides internet access to 40 countries – including Ukraine, an area where GPS navigation is virtually impossible due to the vast amount of jammers from the ongoing war. New research as reported in the MIT Technology Review suggests that these satellites could be used to create a useful navigation system to rival GPS.

Starlink operates more than 3,000 satellites orbiting about 340 miles above the surface of the Earth, and researchers determined that by using synchronization sequences, they could exploit Starlink signals for positioning, navigation, and timing (PNT). Basically, a receiver on the ground could analyze the signals beamed down by Starlink satellites, calculate the distance to the satellite, and then pinpoint a location.

In a similar analysis, electronics company Rohde & Schwarz recently penned an abstract discussing the use of 5G broadcasting as an alternative PNT method to GPS. Author Stefan Maier notes that an extended outage of GPS could cause “multi-billion [dollar] losses for the economy” and therefore it is useful to search for backup solutions.

The abstract notes that Rohde & Schwarz is working on a prototype of an alternative PNT system using the existing terrestrial TV broadcasting infrastructure in the UHF band by: 1) improving the synchronization within and between transmitters, 2) adding special positioning reference signals, 3) adding UTC time stamps, and 4) adding the transmitter locations. (Figure 2.)

[Figure 2 | Rohde & Schwarz’s SMW200A system, which the company says can create GNSS scenarios and emulate wireless communication signals such as 5G broadcast. (Photo courtesy Rohde & Schwarz.)]

“A passive, receive-only mode device (ROM) can calculate its position and precise time without having [to] transmit hardware,” Maier writes. “The capacity of such a system is unlimited in terms of users. Since only a few percent of the 5G broadcast signal are needed for positioning at a time, the vast majority of the signal could still be used for video and audio broadcasting or other data transfer.”

Because television transmitters operate at power levels of up to 100 kW, the signal would be much harder to jam compared to low-power space-based signals like GPS, he adds.

MOSA and SOSA are helping

Until such a GPS alternative emerges, the industry must improve the sensors it has. A key challenge in GPS-denied environments is power consumption, which is a concern with any mobile system. O’Hara says there’s no one-size-fits-all solution here, as “the power required to support re-establishment of position and timing can vary greatly.”

But there are encouraging advancements in technology, particularly when it comes to microelectronics, that could help in this area. And standards developed through the Modular Open Systems Approach (MOSA) – and more specifically hewing to Sensor Open Systems Architecture (SOSA) criteria – is creating momentum when it comes to reuse and modularity for commercial sensing and microelectronics components that lead to breakthroughs in size, weight, and power (SWaP)-constrained conditions, O’Hara says.

“The combination of DARPA’s ERI [Electronics Resurgence Initiative] program portfolio and recent U.S. strategic investments targeting domestic microelectronics (for example, the CHIPS Act) are placing us on a good path toward an enhanced national security posture,” he says.

One of the best things that industry can do when it comes to improving sensor navigation in the future is to reduce single-point vulnerabilities within core systems, O’Hara says.

“Moving to distributed and collaborative architectures that degrade together with ‘soft failure’ modalities is critical,” he says. “I think our military is on the right path in this area. I am getting a stronger sense of buy-in from the services over the last several years with regards to mosaic warfare approaches – using larger numbers of smaller distributed and collaborative elements. Legacy approaches have used smaller numbers of large/exquisite elements that are less resilient and more easily targeted, creating challenges to their operational lethality and survivability.”