The Department of Defense, specifically the Navy, is seeking proposals for the Small Business Innovation Research (SBIR) topic titled "Cue Aggregation Algorithms for Multi-function Receivers." The objective is to develop new machine learning algorithms that can aggregate multiple radio frequency signals into distinct classes while distinguishing between similar signals from different sources. This initiative is crucial for enhancing the capabilities of wideband receivers used in military applications, allowing for improved signal detection and classification. The solicitation is currently in the pre-release phase, with the official opening date set for April 23, 2025, and proposals due by May 21, 2025. Interested parties can find more information and submit proposals through the official SBIR website at https://www.dodsbirsttr.mil/topics-app/.

## Description

When ML is used to locate new radio frequency (RF) energy associated with specific classes of signals in a wideband receiver architecture from within the digital representation of the electromagnetic (EM) spectrum collected by the antenna, the software module only responds to the signals it was trained for. While that is often a nuisance as locating appropriate training data and the time to retrain remain significant issues within the DoD, it can also be a virtue. Importantly, the software should alert (cue) only to the signals it was trained for.If two functionally distinct types of signals are closely matched in the normally used Pulse Description Word parameters, it is plausible to expect their feature space occupancies to be separable but also close together. Thus, the user should be free to train the software to cue as a single report when either is present in the current signal environment or reporting the two signals present separately. Two differently trained copies of the cue generator could first locate all copies of either waveform in the wideband data and then inform the user who needs to know which of the two signals was actually present. For the purposes of this SBIR topic, we call cue generators that aggregate multiple distinct signals into a single cue &ldquo;association or aggregation processors&rdquo;. A second example would be if the training data consists of the same waveform centered over a range of center frequencies. The cues ought to arise out of every occurrence of that waveform independent of frequency. Each cue will still report the actual frequency window of occurrence. Hence if a new association processor is devised that accepts multiple cues from such a frequency agnostics cue generator as inputs, it ought to be able to identify the data&rsquo;s commonality and discover the connections and similarities between the disjoint in time signals and potentially the guiding principles of their waveform alteration (e.g. their frequency hop patterns).Thirdly, note that the cue reports include the time of onset and end of a given transmission as well as identifying which trained signal classes appeared. If the separately received signals of the same signal class from two spatially separated antennae are supplied to a single cue generator and are confirmed to occupy the same feature space volume, a time difference of arrival measurement can be produced by aggregating the separate cue reports. Differences in the feature space points could indicate the signal is dominated by a corrupted transmission with a different effective antenna pattern compared to the other signals being compared. This mechanism of testing correlation can be used as the basis for determining the relative distance to transmitters and watching it evolve in time. Note that this application of cue aggregation processors does not need to be limited to two antenna outputs.Fourth, it may be desirable to train cue generators for each of the main types of radar signals and one aggregation processor that can recognize all as &ldquo;radar&rdquo; as a class so that users wishing to do only comms know to deprioritize all of them. Relatedly, if the signals associated with a single waveform are used to train an ML algorithm with a tightly defined feature space, it should be possible to train the same algorithm to recognize as related but distinct those signals with nearby feature space localizations, one class of anomaly. The Phase I proposal needs to define at least one new aggregation processor functionality and a means of demonstrating it in a lab setting within the base effort. The Phase I Option effort shall then begin the Phase II effort by defining and demonstrating additional types of processors. The Phase II may become classified depending on functionalities and GFE data sources provided.Work produced in Phase II may become classified. Note: The prospective contractor(s) must be U.S. owned and operated with no foreign influence as defined by 32 U.S.C. &sect; 2004.20 et seq., National Industrial Security Program Executive Agent and Operating Manual, unless acceptable mitigating procedures can and have been implemented and approved by the Defense Counterintelligence and Security Agency (DCSA) formerly Defense Security Service (DSS). The selected contractor must be able to acquire and maintain a secret level facility and Personnel Security Clearances. This will allow contractor personnel to perform on advanced phases of this project as set forth by DCSA and NAVAIR in order to gain access to classified information pertaining to the national defense of the United States and its allies; this will be an inherent requirement. The selected company will be required to safeguard classified material during the advanced phases of this contract IAW the National Industrial Security Program Operating Manual (NISPOM), which can be found at Title 32, Part 2004.20 of the Code of Federal Regulations.

## Objective

Invent and develop a set of new algorithms expressed in machine learning (ML) form running on commercial off-the-shelf (COTS) processors that can both aggregate multiple signals into classes of same pattern of emission and locate methods to distinguish look-alike signals from platforms with different spectral signatures or different intentions.

## Phase I

Realize the Digital Signal Processing technique. Demonstrate it in the lab on synthetic or collected data during the Phase I Base period. Prepare a Phase II proposal. The Phase I Option, if exercised, will discuss the signal classes to be focused on with the Navy TPOC and begin the preliminary work associated with the Phase II plan.

## Phase II

Work in a collaborative fashion with Navy personnel to develop and demonstrate a set of aggregative processors operating on COTS processors cards during the Phase II Base period. In the Phase II Option, if exercised, integrate these new processors into the DSP libraries and architecture of a software defined, ultrawideband receiver prototype, address any concerns over fielded operations, and demonstrate processor capacities to potential transition sponsors via an outdoor but in CONUS exercise. It is probable that the work under this effort will be classified under Phase II (see Description section for details).

## Phase III: Dual Use Applications

Transition the new processors into the soft/firmware libraries of a Program of Record. Support the Navy in transition efforts.In the civilian market, establish a set of low-cost software defined radios of the Internet of Things (IoT) sort based on the prioritized receiver concept and with a modest set of loaded DSP and specific cue generators. The IoT messages will set the class of signal to be processed at any given time and the cues active at any given time would define the channels offered to the user. An example might be the multi-functionality of modern television industries offering streaming services; broadcast, music, Kindle-like services; appliance control; and interaction with medical sensors and conveyance of information to clinicians, a one-stop shop for RF functions.

Topic

Agency

Program

Topic Link

Additional Information