U.S. military researchers are asking for industry’s help in developing a combination hybrid analog, digital, photonic, and electronic processor to help analyze RF and optical signals for situational awareness. The project seeks to enable U.S. and allied warfighters to understand in near-real-time all the waveform details, source type and class, signal format, and geolocation of detected RF signals, as well as the information these signals carry and whether or not the information is encrypted.
The Department of Defense (DoD) conducts air, land, maritime, space and cyberspace operations which increasingly depend on access and control of full electromagnetic (EM) spectrum. Many layers of EM signals are leveraged for communication, command, control, intelligence, sensing and attack.
The EM signal environment is becoming crowded and cluttered, and DARPA anticipates it to grow exponentially more so as new technologies such as 5G wireless, drones, autonomous vehicles, and millimeter wave radar enter service. The EM environment is also used by adversaries to carry out ever-increasing electronic and cyber attacks to degrade US DOD’s ability to use electromagnetic spectrum resources in tactical and strategic military operations.
Real time Signal intelligence provide critical capability to DoD to know in real time exactly who is operating in the EM environment around them, what the adversaries are doing in it, what information adversaries are exchanging and, what the adversaries are learning about U.S. forces. The EM signal environment contains valuable information about the strategic and tactical order of battle of potential adversaries, their maneuvers and actions, and early indications of potential threats.
In order to have this superior knowledge and situational awareness, it is not sufficient to simply collect radio-frequency (RF) or optical signals—it is necessary to understand in near-real-time all the “externals” of the signals (waveform details, source type and class, signal format, geolocation, etc.) and much of the “internals” of them (the information carried on the signal, possibly encrypted or hidden).
The DARPA All-Signal Tactical Realtime Analyser (ASTRAL) program is interested in ensuring understanding of and access to the congested and contested EM environment of the battlefield, by exploiting innovative developments in hybrid analog/digital photonic/electronic processor technologies of wideband real-time signal processing in order to detect and exploit hidden EM signals in real time and accomplish high-value military signal intelligence, surveillance and reconnaissance (ISR) applications.
The analog photonic elements could take advantage of the wide bandwidth, wide optical dynamic range, ease of parrelization, and ability to implement multiplications by square-law detection that photonic technology offers. the digital elements, meanwhile, could implement general algorithms, low costs, and programming flexibility of electronics.
Photonic processors also have unmatched speeds and latencies, which make them well suited for specialized applications requiring either real-time response times or fast signals. One example is a front-end processor in radio-frequency transceivers. As the wireless spectrum becomes increasingly overcrowded, the use of large, adaptive phased-array antennas that receive many more radio waves simultaneously may soon become the norm. Photonic neural networks could perform complex statistical operations to extract important data, including the separation of mixed signals or the classification of recognizable radiofrequency signatures.
By greatly increasing real-time signal speed and spectrum coverage by 1000x or more, the ASTRAL program will enable superior EM signal awareness at the tactical edge with new technology suitable for tactical mobile units. The program will also greatly improve the capability of current computing and communication resources applied to EM awareness.
ASTRAL technology will enable U.S. tactical forces, in both traditional military environments and nontraditional environments (e.g. urban and “hybrid” conflict), to understand what adversaries are doing around them, anticipate adversaries’ future actions, and recognize potential threats. ASTRAL will provide an asymmetric advantage, by making adversaries’ EM sensing and communications transparent at the tactical edge.
ASTRAL’s real-time wideband hybrid photonic/digital processing can advance all fronts of DoD’s Electromagnetic Maneuver Warfare (EMW) strategy, with applicability to a broad range of DoD Joint functions – movement and maneuver, fires, network defense, command and control, intelligence, protection, and sustainment. The successful demonstration of ASTRAL technology and the identification of ASTRAL application architectures can potentially advance current U.S. capabilities to the level of multi-domain theater scale real-time EMW tactical operations.
Hybrid Optical technology
Application of analog photonics for high nonlinear processing gain is made possible by the recent development of very stable optical sources with very high coherence, and by the development of nonlinear fiber optics for mixing signals together. Optical frequency combs with hundreds or thousands of mutually-coherent “lines” propagating in an optical waveguide (e.g. silica fiber, silicon photonic circuit) can channelize wideband optical and RF signals into thousands of parallel narrowband channels.
For example, researchers have demonstrated channelization and analog Fourier transformation of 110 GHz of RF bandwidth, and demonstrated high cyclostationary (CS) processing gain on its output (digitally, in a narrow subband scanned over the 110 GHz in non-real time).
Nonlinear optical transformations of these encoded combs in a waveguide can implement higher-order nonlinear algorithms, such as moment and cumulant analysis, at the speed of light. The waveguided photonics also enable compact and rugged systems for use in the field.
All-Signal Tactical Realtime Analyser (ASTRAL) program
The objectives of the ASTRAL program are to (1) develop a hybrid analog/digital photonic/electronic processor demonstrating real-time nonlinear CS (4th order) and convolutional (2nd order) processing with high processing gain over input EM signals filling an instantaneous bandwidth of 1 to 10 GHz, and (2) identify architectures of wideband hybrid analog/digital photonic/electronic processors that are well-suited to implement high-value tactical military signal processing applications.
Some of the strategic and tactical spectrum awareness applications of interest include (a) optical communication real-time internet protocol (IP) packet identification and exploitation for physical layer network defense, (b) city-wide wireless device geolocation, (c) X-Ku-Ka band low probability of intercept (LPI) radar warning, and (d) theater wide spread-spectrum radio geolocation.
Many of the signals U.S. forces need to observe and understand are hidden—either intentionally transmitted below the noise or clutter floor, or unintentionally buried in the clutter and interference of other irrelevant signals (“needle in a haystack”). In order to have all-signal awareness of the EM environment, systems need the significant processing gain of high-order nonlinear algorithms.
An algorithm of “Nth order” uses the signal waveform(s) x(t) raised to the Nth power or Nth product, e.g. x(t)N, or x(t)N-mx(t’)m. Fourier transformation x(t) X(f) is 1st order; correlation <x(t)x(t+t’)> X(f)Y*(f) is 2nd order. High order nonlinear algorithms can have large processing gain, but it comes at the cost of many steps of digital computation.
Current real-time high-gain signal processing is limited to an instantaneous input rate of ~ 1-10 MHz in a single digital electronic processing system (sometimes applied across larger bandwidths of raw data, at the cost of undersampling which discards data). The tactical EM environment contains many military systems operating over multi-GHz of RF bandwidth, interfering civilian systems can occupy up to 100 GHz of RF spectrum, and backhaul networks use many 1000’s of GHz of optical spectrum.
ASTRAL’s approach to processing signals at 1000x higher data input speed and still achieve high processing gain is to combine new analog photonic technology with state-of-the-art digital electronics in a hybrid photonic/electronic processing system. The analog photonic elements take advantage of the strengths of photonics: wide optical bandwidth (> 100 GHz), wide optical dynamic range (>120 dB, shot noise limited), ease of massive parallelization (superposition of many non-interfering carriers), and ability to implement multiplications (N>2 processes) by square-law detection.
The digital elements take advantage of the strengths of electronics: ability to implement general algorithms (e.g. transcendental functions, recursive transformations), low cost of commercially-driven processors, and programming flexibility. Analog photonic elements can create Nth-order signal products at the speed of light over the full optical bandwidth and filter for signals of interest. Digital elements can apply more general logical and nonlinear algorithms for specific applications such as emitter identification, geolocation, waveform classification, etc.
Technical Area 1 (TA1): Real-Time Wideband Photonic/Electronic Processor.
In Technical Area 1, ASTRAL performers will develop a hybrid analog/digital photonic/electronic processor brassboard implementing the 4th order CS processing algorithm on wideband signal input approximately 1000x wider than current real-time all-digital processors.
The CS algorithm is generally applicable to all signals and it has high processing gain limited only by its frequency resolution and integration time. It is commonly applied (digitally, over a narrow bandwidth) to pull LPI signals uncooperatively out of noise and clutter. TA1 will conclude with the demonstration of a brassboard system running wideband signals through the
CS process at approximately 1000x higher real-time data input speed, detecting and recovering LPI signals of interest (hopping, intermittent, spread spectrum, etc.), passing these to further digital processing tailored to each signal, and demonstration of the brassboard running wideband signals through a general convolutional process. These demonstrations of ASTRAL-improved real-time EM signal awareness and EM-ISR are necessary milestones before proceeding to full development of military applications. The TA1 demonstrations also directly mirror several transitionable military applications including (1) X-Ku-Ka LPI threat radar detection and (2) spread-spectrum radio detection and geolocation.
Technical Area 2 (TA2): Flexible Photonic Processing Architectures and Algorithms.
In Technical Area 2, ASTRAL performers will design hybrid analog/digital photonic/electronic architectures matched to high-gain signal processing algorithms (beyond CS) addressing candidate military EM ISR applications. In a photonic processor, the architecture is the algorithm. Design trade studies will identify candidate architectures for future development of military applications. TA2 performers will focus on design of photonic architectures that implement specific high-gain and high-value algorithms of military applications, and that can potentially be packaged into small size, weight and power and cost (SWaP-C) tactical systems.
Potential DoD transition organizations will be invited to suggest critical EM ISR problems and applications, and to advise on the applicability and operability of ASTRAL architectures. TA2 will conclude with the recommendation of architectures for potential future development into prototype systems in follow-on DARPA programs.