SOW Rev1 for posting_Redacted.pdf

PARTICLE SOFTWARE DEVELOPMENT

STATEMENT OF WORK (SOW)

1.0 BACKGROUND

Naval Surface Warfare Center Dahlgren Division, Dam Neck Activity (DNA) has a requirement to procure software that is capable of processing RF data to enhance neural network machine learning models for integration into legacy and future US Navy Carry On Equipment (COE) systems. This capability has been identified as a requirement for future system architecture advancements and to support continued operations. NSWCDD DNA Intel Systems provides the Navy with systems for required to win decisively in major combat operations.

NSWCDD DNA develops and integrates emerging technologies into cohesive intelligence collection systems to provide actionable intelligence in the maritime environment.

2.0 OBJECTIVE

The objective of this task is to deliver the OmniSIG software solution with additional machine learning capabilities for signal collection in legacy and future COE.

3.0 REQUIREMENTS

This SOW provides requirements for the delivery of OmniSIG model file.

3.1 The Contractor shall generate a study for applying non-Radio Frequency (RF) data to commercial RF signal classifiers.

3.1.1 The Contractor shall generate test metrics for custom machine learning models, offering performance measurements of the trained models for probability of detection, probability of false alarm, and basic confusion matrix showing the probability of misclassification.

3.1.2 The Contractor shall retrieve government provided data, splitting it into model training and test sets, manipulate the data to ensure that the current COTS OmniSIG products are able to ingest both for model training and processing.

3.1.3 The Contractor shall create a custom OmniSIG model by using OmniSIG Studio tools to annotate /label and train a custom signal detection and classification model using the provided data.

NOTE: This task assumes that the task in 3.1.2 was successful in manipulating the provided data into a format that is usable by the OmniSIG Studio tools.

3.1.4 The Contractor shall generate test metrics for custom model development by gathering metrics regarding the performance of the trained models. Using the test set that was not used for training, and any other additional data provided by the government, performance metrics including probability of detection, probability of false alarm, and a basic confusion matrix showing the probability of misclassifications shall be generated.

3.1.5 The Contractor shall deliver a final report clarifying any issues with the curation of the data and detailing the performance of the commercial network on the non-RF data. The final report may also include future steps for any network modifications that may result in better performance for the provided data types.

3.2 The Contractor shall deliver one (1) OmniSIG Studio software license and documentation.

3.3 The Contractor will deliver two (2) OmniSIG Sensor license and associated user documentation.

***Note: “training” in these tasks denotes the training of machine learning models, not the training of personnel or any training for human educational purposes.

4.0 DATA DELIVERABLES

4.1 The Contractor shall provide data IAW CDRL A001 and CDRL A002.

5.0 GOVERNMENT FURNISHED PROPERTY (GFP)

5.1 The Government will provide the necessary data sets to the Contractor from which the Contractor is able to accomplish the tasking in section 3.1.

5.2 Due to the sensitive unclassified nature of the equipment, a complete list of the data set will be provided to the contractor upon award of contract and prior to preliminary developmental testing.

6.0 TECHNICAL POINT OF CONTACT

To be released at time of award