Artificial Intelligence/Machine Learning, ASA(ALT), Direct to Phase II
Artificial Intelligence/Machine Learning (AI/ML) Ready Synthetic Radio Frequency (RF) Data
Objectives
The objective of this SBIR topic is to advance methods for generating and labeling synthetic data representing various classes of Radio Frequency (RF) signals. This synthetic data will support the training of Electronic Support and Signals Intelligence (SIGINT) models aimed at enhancing automated detection, characterization, and identification (DCI) of Signals of Interest (SoI).
By leveraging artificial intelligence (AI) and machine learning (ML), this initiative aims to address the challenge of managing the increasing volume and diversity of RF signals, which traditional techniques struggle to keep pace with. This innovation seeks to reduce operator workload and improve battlespace awareness and decision-making capabilities across Army strategic and tactical operations.
Description
The increasing volume and variety of Radio Frequency (RF) signal propagation presents a significant challenge to maintain situational awareness of unit and system surroundings. Traditional techniques for identifying Signals of Interest (SoI) in the environment corresponding to potential threats or targets, and for maintaining awareness of blue force or civilian activity in the area, are unable to keep pace.
Recent rapid growth within the RF technology space is driven largely by the affordability and proliferation of software defined radios (SDR) and modern communication protocols enabling the Internet of Things (IoT) and associated networking infrastructure. To outpace the ballooning signal space, automated detection and characterization is required.
Artificial Intelligence (AI) and Machine Learning (ML) are the key to this automation, along with a large volume of AI-ready data to train and develop the models that will perform these tasks. Because measured data collections can have high cost, high schedule requirements due to complex coordination of saturated battle spaces, and high risk due to many moving pieces, synthetic data is an important component of the Army’s data strategy.
Phase I
This topic is only accepting Direct to Phase II (DP2) proposals for a cost up to $2,000,000 for an 18-month period of performance.
Proposers interested in submitting a DP2 proposal must provide documentation to substantiate that the scientific and technical merit and feasibility equivalent to a Phase I project has been met. Documentation can include data, reports, specific measurements, success criteria of a prototype, etc.
Phase II
The focus of this SBIR topic is generating and labeling synthetic data of RF Signals of Interest (SoI) which could then be used for training AI/ML RF SoI detection models. During DP2, firms should (1) develop new or novel method(s) for generation, labeling, and testing of relevant data sets for training of signal detection models, (2) implement the developed method(s) in Project Linchpin’s AI Unclassified Operations Environment for DOD use cases.
Phase III
- Research has proven the efficacy of using AI/ML to automate radio frequency identification (RFI), especially as software defined radios (SDRs) proliferate [1,2].
- AI/ML neural networks are the enabling technology behind the detection, namely Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs).
- CNNs leverage radio spectrograms to detect RFIs by estimating noise distribution in wideband radio spectrograms. RNNs detect RFIs by being adept at detecting time-series data and finding minute changes in temporal data; RNNs have seen more efficacy by being combined with CNNs.
- Potential dual-use cases for AI/ML-enabled RFI include:
- Healthcare diagnosis sector that uses AI/ML to detect changes in radio waves that penetrate the body.
- Telecommunications – the bandwidth 5G/6G spectrum is becoming crowded and AI/ML can help ameliorate constraints by automating bandwidth allocations. Note: This is a similar, but not direct, application.
- Civil defense-tech applications, namely counter drone technology.
Submission Information
For more information, and to submit your full proposal package, visit the DSIP Portal.
SBIR|STTR Help Desk: usarmy.sbirsttr@army.mil
References:
- https://doi.org/10.1109/percomw.2017.7917555
- https://arxiv.org/pdf/1909.11512.pdf
- https://surface.syr.edu/cgi/viewcontent.cgi?article=2242&context=etd
- Radio Frequency, Signals of Interest, RF, SoI, Automation, Synthetic, Data, AI/ML
Objectives
The objective of this SBIR topic is to advance methods for generating and labeling synthetic data representing various classes of Radio Frequency (RF) signals. This synthetic data will support the training of Electronic Support and Signals Intelligence (SIGINT) models aimed at enhancing automated detection, characterization, and identification (DCI) of Signals of Interest (SoI).
By leveraging artificial intelligence (AI) and machine learning (ML), this initiative aims to address the challenge of managing the increasing volume and diversity of RF signals, which traditional techniques struggle to keep pace with. This innovation seeks to reduce operator workload and improve battlespace awareness and decision-making capabilities across Army strategic and tactical operations.
Description
The increasing volume and variety of Radio Frequency (RF) signal propagation presents a significant challenge to maintain situational awareness of unit and system surroundings. Traditional techniques for identifying Signals of Interest (SoI) in the environment corresponding to potential threats or targets, and for maintaining awareness of blue force or civilian activity in the area, are unable to keep pace.
Recent rapid growth within the RF technology space is driven largely by the affordability and proliferation of software defined radios (SDR) and modern communication protocols enabling the Internet of Things (IoT) and associated networking infrastructure. To outpace the ballooning signal space, automated detection and characterization is required.
Artificial Intelligence (AI) and Machine Learning (ML) are the key to this automation, along with a large volume of AI-ready data to train and develop the models that will perform these tasks. Because measured data collections can have high cost, high schedule requirements due to complex coordination of saturated battle spaces, and high risk due to many moving pieces, synthetic data is an important component of the Army’s data strategy.
Phase I
This topic is only accepting Direct to Phase II (DP2) proposals for a cost up to $2,000,000 for an 18-month period of performance.
Proposers interested in submitting a DP2 proposal must provide documentation to substantiate that the scientific and technical merit and feasibility equivalent to a Phase I project has been met. Documentation can include data, reports, specific measurements, success criteria of a prototype, etc.
Phase II
The focus of this SBIR topic is generating and labeling synthetic data of RF Signals of Interest (SoI) which could then be used for training AI/ML RF SoI detection models. During DP2, firms should (1) develop new or novel method(s) for generation, labeling, and testing of relevant data sets for training of signal detection models, (2) implement the developed method(s) in Project Linchpin’s AI Unclassified Operations Environment for DOD use cases.
Phase III
- Research has proven the efficacy of using AI/ML to automate radio frequency identification (RFI), especially as software defined radios (SDRs) proliferate [1,2].
- AI/ML neural networks are the enabling technology behind the detection, namely Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs).
- CNNs leverage radio spectrograms to detect RFIs by estimating noise distribution in wideband radio spectrograms. RNNs detect RFIs by being adept at detecting time-series data and finding minute changes in temporal data; RNNs have seen more efficacy by being combined with CNNs.
- Potential dual-use cases for AI/ML-enabled RFI include:
- Healthcare diagnosis sector that uses AI/ML to detect changes in radio waves that penetrate the body.
- Telecommunications – the bandwidth 5G/6G spectrum is becoming crowded and AI/ML can help ameliorate constraints by automating bandwidth allocations. Note: This is a similar, but not direct, application.
- Civil defense-tech applications, namely counter drone technology.
Submission Information
For more information, and to submit your full proposal package, visit the DSIP Portal.
SBIR|STTR Help Desk: usarmy.sbirsttr@army.mil
References:
- https://doi.org/10.1109/percomw.2017.7917555
- https://arxiv.org/pdf/1909.11512.pdf
- https://surface.syr.edu/cgi/viewcontent.cgi?article=2242&context=etd
- Radio Frequency, Signals of Interest, RF, SoI, Automation, Synthetic, Data, AI/ML