TOPIC CALL 3.pdf

CB BAA HDTRA122S0002 CALL 3

THE PROPOSAL SUBMISSION DEADLINE FOR THE FOLLOWING

TOPICS IS OCTOBER 26, 2023 AT 2:00PM EASTERN DAYLIGHT TIME

(EDT).

THE DEADLINE TO SUBMIT QUESTIONS RELATED TO THE

FOLLOWING TOPICS IS SEPTEMBER 18, 2023.

CBI-01: Next-Generation Transport and Dispersion Model for Chemical and Biological Defense Applications: From Training and Mission Planning to Real-World Decision Support

Background: Chemical and biological (CB) modeling and simulation (M&S) capabilities are critical to provide Warfighters with an understanding of the hazard environment, the operational impact, and potential mitigation strategies. M&S tools can be leveraged for planning, training exercises, and real-world event responses. Representative programs that leverage CB hazard models include the Hazard Prediction and Assessment Capability (HPAC), Joint Effects Model (JEM), and Hazard Estimation and Assessment Toolkit (HEAT), which functions within in the Tactical Assault Kit platform.

Recently, the Digital Battlespace Management Division within the Joint Science and Technology Office for Chemical and Biological Defense (JSTO CBI) has also developed CB M&S tools to support Warfighters at the tactical edge. This has expanded the application space from left-of-boom planning to right-of-boom responses. The end users have also grown from command and control and reachback cells to on-site operators. Therefore, CB M&S tools must be able to provide results at the speed of relevancy in various connectivity environments, with varying computational power, and at different levels of fidelity.

Accurately modeling and simulating CB agent releases often requires the coupling of multiple models at varying degrees of fidelity. Representative examples include source term models that characterize the initial dissemination; transport and dispersion (T&D) models that predict hazard area dosages and deposition; and models that account for secondary evaporation. Furthermore, human health effects models are necessary to translate the contamination into easily interpretable operational impacts. This topic is specifically focused on the T&D models.

Two commonly implemented, outdoor T&D methodologies include using Gaussian puffs to represent three-dimensional, time-dependent concentrations that provide an average representation of the concentration field, or using a Computational Fluid Dynamics (CFD) solution that solves diffusion equations along with the relevant fluid mechanics conservation equations for atmospheric flow and turbulence. The latter approach allows simulations of single realizations of the transport of material. These two modeling approaches are based on different physical principles and provide different advantages and disadvantages for the expanded application space and audience that JSTO CBI supports. Indoor modeling methods have included: box models for simple, quick modeling estimates; CFD approaches for complex environments and highly detailed analyses; and multi-zonal models, which take an intermediate approach in both complexity and modeling detail.

Objective: Proposals should describe the solution, scientific foundation, development steps, and methods to demonstrate the advantages realized through the adoption of the proposed next-generation T&D model over currently fielded approaches.

The desired features of the next-generation T&D model(s) include but are not limited to:

• Account for different properties of CB materials (e.g., decay, persistence, gas density)

• Ability to couple with existing models (e.g., source term, numerical weather prediction, secondary evaporation, health and human effects)

• Capable of modeling and simulating T&D of CB agents:

o In varying spatial domain sizes (i.e., different geographical sizes of interest with varying levels of detail about the area) o In indoor, outdoor, and indoor-outdoor coupled environments o At varying fidelities, ranging from limited or missing parameters to highly detailed inputs related to building features, agent properties, population, etc.

• Support applications deployed in different environments from the tactical edge to

C2/reachback using platforms with varying computational power

• Ability to recognize and account for terrain data from external sources (e.g., NGA GRiD, One World Terrain, common industry standard file formats such as.obj files)

• Allow real-time dynamic altering of model simulations for applications such as synthetic training environments or the possible future incorporation of real-world sensor data

• Model results at speeds and fidelities to support a variety of uses such as real-world operational decision support, training in extended reality-based synthetic environments, and synthetic test environments

The capabilities developed under this initiative should implement modular design to allow flexibility in maintenance and further development. The to-be-developed capability is expected to function as microservice and should work on Windows, Linux, and Android systems. The proposed approach should also allow flexibilities to adapt to/function with potential future AI/ML-based models. Agile development is expected.

Impact: This effort will deliver a simulation capability that provides the Warfighter with an ability to operate in and overcome a CB-contested environment through increased understanding of the threat and hazard area. The capability will be a next-generation T&D model that provides end-to-end, coupled, indoor-outdoor modeling for variably scaled scenarios.

CBI-02: Tactical Decision Aids for Chemical and Biological Defense

Background: The Digital Battlespace Management Division within the Joint Science and Technology Office for Chemical and Biological Defense (JSTO CBI) is charged with researching and developing technologies to support the Chemical and Biological Defense Program. JSTO CBI develops modeling and simulation tools to generate decision-oriented information and enable enhanced situational awareness with the goal of protecting military and civilian populations from intentional or incidental chemical or biological threats and Toxic Industrial Chemicals/Toxic Industrial Materials (TICs/TIMs).

Mobile applications and microservices are revolutionizing the way we approach new technologies for the Department of Defense (DoD). Lightweight and transportable devices such as smartphones and tablets have demonstrated their potential to change the way we approach decision support capabilities that enable the Warfighter to operate under the threat of Chemical, Biological, Radiological, and Nuclear (CBRN) threats. Systems such as Nett Warrior have deployed under wartime conditions, increasing situational awareness, decreasing reaction and decision times, and augmenting operational capability. The Tactical Assault Kit (TAK)—a mapping system with a plugin architecture—has seen similar success across the DoD, where operators must routinely execute missions in spectrum denied environments.

Situational awareness often develops from the edge of the force, while decisions are made furthest from that edge—the ability to provide a software solution that communicates with other architectures and promulgates awareness from the flagpole to the fighting hole and increases the rate at which decisions are made in CBRN environments is critical. Proposed solutions should speak to current and future DoD requirements with the individual Warfighter in mind. Given that our Warfighters must operate in environments with little to no network connectivity, proposed decision aids should be able to operate both within the cloud at lower Tactical Internet (TI) and while disconnected.

Objective: This topic seeks to develop decision support tools for CB mission planning, warning and reporting, hazard prediction, and consequence assessment. The development approach must leverage a modular, open systems architecture. If the proposed decision aids leverage or develop atmospheric transport and dispersion tools, the performer shall leverage a Government-supplied Common Chemical/Biological Modeling Framework (CCMF) software developer kit. Applications may span from simple, rules-based applications to tools that take advantage of heterogeneous computing to implement machine learning approaches that enable early warning of exposure to CB agents.

The desired features of decision support tools include but are not limited to:

• Integrating within multi-domain/echelon architectures

• Having a minimal footprint and logistics requirements (e.g., low size, weight, power and logistic requirements) for the host platform

• Fully documented software that is easy to access, modify, and extend (modular)

• Fully comply with DoD software standards for authorization to operate (ATO)

• Should not require recurring license fees

• Ability to run on personal computing devices and mobile devices in potentially disconnected environments

• Software must be tested, verified, and maintained

• Should be “turn-key,” so that an individual could operate the system with a limited amount of additional training

Impact: This topic will support the Warfighters’ ability to operate in and overcome a CBRN-contested environment through the use of superior tactical decision-making aids, the promulgation of situational awareness, and tactical applications.

CBI-03: Sensor Data Fusion and Algorithm Reform: Moving Toward Flexible Multi-echelon CBRN Sensor Analytics

Background: Chemical, Biological, Radiological, and Nuclear (CBRN) sensors are a critical component to provide the Joint Force with an understanding of the contaminants in the operational area. With advances in sensor technology based on a variety of analytical principles, CBRN sensors can perform powerful analysis in the field. However, a universal sensor capable for detecting and analyzing the expanding list of CBRN threats is not yet available.

The Joint Force relies on a collection of sensors to identify the presence and type of CBRN contamination, to understand the extent of the contamination, and to further analyze the hazard associated with the contamination. Due to the complexity and volume of data collected from the sensors, it is challenging for operators to understand and interpret the data without assistance.

In order to achieve maximum utility of CBRN sensor data and obtain accurate identification, quantification, mapping, and advanced analyses of the CBRN threats, the Digital Battlespace Management Division within the Joint Science and Technology Office for Chemical and Biological Defense (JSTO CBI) seeks to combine artificial intelligence (AI), machine learning (ML), and data science (DS) approaches to develop robust sensor data fusion algorithms that are flexible enough to smartly adjust between environments supporting optimal computational horsepower and data transfer rates through to operations constrained by computational and data transfer limitations.

The focus of this topic is not to develop a single sensor data processing algorithm nor a fusion algorithm to support a specific subset of sensors. The focus is to develop flexible data analytics and fusion framework that can mitigate various challenges associated with CBRN sensor data processing and support multi-echelon CBRN sensor analytics. It should be emphasized that this effort focuses on CB sensor data analytics. However, it is desired for the fusion framework to be compatible with Radiological and Nuclear (RN) sensors for future incorporation of RN sensor data processing. Therefore, in the following sections of this solicitation, contaminant/hazard detectors are referred to as CBRN sensors to highlight the need for compatibility.

The challenges faced by CBRN sensor data analytics can be generally separated into the following categories:

• Data Challenges o Non-uniform data formats, as sensors can be provided by various manufacturers o Non-uniform data types, as some sensors generate categorical data while others yield time series data o Non-uniform sensor data collection rates

• Sensor Types o Various sensing modalities for CBRN hazard detection (e.g., ion mobility spectrometry, mass spectrometry, colorimetric) o Various sensor types, including environmental sensors and physiological monitoring devices that provide advance warning of exposure o Various sensor configurations (i.e., the operators may have multiple of the same sensors (homogeneous), a group of sensors with different modalities

(heterogeneous), or multiple different modality sensors (heterogeneous with varying numbers of each modality)

• Data Analytic Methods o Features in data from different sensor modalities may need to be extracted by different data analysis methods (e.g., random forest) o Sensor data analytics can be centralized, parallel fusion, parallel fusion with side information, or parallel without fusion depending on sensor configurations o Data fusion level (i.e., sensor data fusion may need to occur at the decision level, feature level, or raw data level, depending on operational environment)

• Connectivity and Computational Resources o Sensors can be operated in disconnected, locally connected, or fully connected environments o Sensors with some connectivity may or may not be able to continuously send data to the fusion cell o The computational resources available vary at different levels of the decision echelon o The impacts of connectivity and computational resources on data analytics can be further complicated when considering the scenario when a computation/fusion node fails and nearby nodes need to assume the functions of the failed node

• Timing of Data Fusion o CBRN sensors can be man-worn/human-handled or on unmanned platforms moving at different speeds o Natural progression of contamination transport and dispersion may result in delay in sensor responses when sensors are distributed in an area

Objective: CBI seeks to develop a flexible sensor data analytics and fusion framework that can overcome the above mentioned challenges. The desired characteristics/capabilities of the to-be-developed framework include but are not limited to:

• Perform multi-level fusion

• Support multi-modal sensors (including physiological monitoring sensors)

• Support analytics and fusion of data from sensors in varying connectivity environments

• Support analytics and fusion of data from sensors deployed in various configurations

• Support fusion when decision are made or features are extracted based on different data processing methods (e.g. working with random forest based model for sensor 1 and autoencoder based model for sensor 2)

• Complete the observe-orient-decide-act (OODA) loop to support CBRN response operations o For context, when sensors operate in various connectivity environments, there can be multiple sub-level OODA loops to be completed o The desired analytics and fusion framework should have the flexibility to integrate with future capabilities that can influence the analytics of CBRN and physiological data.

Two future capabilities in the horizon to consider are (1) microsensors, when they operate as a package (e.g., on a swarm of unmanned vehicles) and (2) autonomous threat mapping algorithms, which could have its own OODA loop.

• Flexible modular design to reduce model retrain efforts and allow scalability to support future addition of new sensors and future upgrade of sensors to perform quantitative analysis

Impact: This effort will deliver a sensor data analytics capability that provides the Warfighter with a comprehensive understanding and assessment of the CB-contested environment for improved situational awareness of the threat and hazard area.