This PDF file contains the front matter associated with SPIE Proceedings Volume 12119, including the Title Page, Copyright Information, Table of Contents and Conference Committee list.

The purpose of this paper is to develop a systems dynamics model to probe the impacts of investment management decisions on early vs. late stage research. The model attempts to capture the dynamics associated with project completion time and cost as a function of inherent project risk and risk mitigation measures. The investment management decisions (i.e., independent variables) are the individual project expected impact which is assumed to be proportional to technical risk involved and the fraction of budget diverted to early technology maturation prior to project commencement. This notion of early technology maturation is intended to represent the aggregate effects of refining and solidifying detailed requirements, exposing and addressing areas of low technology maturity, and evolving the underlying technology to best meet user needs. This systems dynamics model is first run for a single project to understand its operation and to glean insights concerning how cost and schedule are impacted by investment management decisions. The model is then applied to multiple communities executing multiple projects simultaneously. This reveals insights into managing a portfolio of projects selected to have different technical risk levels and commensurately manage how much budget to devote to early technology maturation. Specific insights are uncovered for managing early technology maturation as a function of inherent project risk; for given levels of project cost, work involved, and early maturation impact; and when looking to optimize both project risk and early technology maturation simultaneously.

Open architectures promise interoperability, rapid prototyping, and affordability. Software defined radios (SDRs) require high performance embedded signal processing with precision timing, synchronization, and security. Users want flexibility and adaptability. Is it possible to develop an open architecture (OA) approach that provides all these attributes? Current and emerging generations of general purpose processors (GPPs) provide powerful new choices for implementing SDR waveforms. Waveform software that is optimized for vector processors on GPPs can compete in performance with Field Programmable Gate Array (FPGA) implementations, but the software has much faster development schedules. SDRs can now be designed with better modularity using high-speed OA interfaces between components. A waveform experiment was conducted to assess these new tools and techniques that can accelerate waveform development. This paper explores past and current trends in open architectures for SDRs, discusses lessons learned from case study results, and provides recommendations for creating a waveform development kit (WDK) based on OA SDR.

Designing sensors using Open Architecture (OA) principles enables interoperability and reduces the cost of procuring sensors by permitting a sensor be built by multiple vendors instead of only one. The Sensor Open Sensor Architecture™ (SOSA) Consortium is a consensus-based community of government and industry partners working together to develop the next generation of sensors that have a well-defined set of interfaces for software, hardware, and electro-mechanical components. In total, the SOSA Technical Standard is applicable to any one of the five sensor types (or combination): radar (RADAR), electro-optical (EO/IR), signal intelligence (SIGINT), communications (COMMS), and electronic warfare (EW). The ability of the SOSA Technical Standard to be a pointer to five different sensor types is what sets it apart from other standards available today. Many standards suffer from being application/platform specific, relevant to only one sensor type, and locked to a particular vendor. The SOSA Consortium combats this problem by focusing on being platform, vendor, and sensor agnostic. This paper provides examples of how a sensor system can be designed using the concepts and technologies defined in the SOSA Technical Standard in order to escape the proverbial vendor event horizon

This presentation concentrates on our efforts building the Digital Engineering, and M&S infrastructure that is being developed for Air Force Research Laboratory including, M&S tools, GRA, ASoT, MBSE, DevSecOps, Digital Twins, and Kill Web M&S, a to increase speed and flow of traditional and non-traditional vendor technologies to the acquisition process.

The open source software (OSS) movement has enabled an incredible list of innovations and underpins much of the current technology utilized in a broad array of applications. A strength of many of the current OSS solutions is in the modularity of these libraries and their interoperability which allows complex solutions to be created through combinations of powerful software components with well defined interfaces. The hardware world is slowly embracing this open philosophy of modular components and we are seeing a rise in open source hardware solutions, with the open source RISC-V instruction set architecture (ISA) acting as a leading technology platform and driver. Currently, the open source hardware movement boasts multiple hardware point designs for CPUs, GPUs, accelerators, Networks-on-Chip as well as various bus interface standards to connect disparate pieces of IP together. A crucial piece of this open source hardware ecosystem is the inclusion of a highly capable, open source software stack that includes production-ready compilers, operating systems, performance libraries, etc. This well supported software stack relieves the developers of custom processors of the significant burden of writing this software from scratch while enabling application developers to more readily port to any custom design. While there is broad availability of open source hardware components it is unlikely that a currently available design, or even a collection of designs, will exactly suit the often niche requirements of military systems. To take full advantage of this open source hardware ecosystem a new hardware design methodology based on hardware generators is required. Mimicking the modularity of the OSS ecosystem, basic hardware building blocks currently available in the open source ecosystem can be utilized in design environments capable of rapid generation of custom designs through extension of existing open implementations. Going further, this hardware generation methodology is capable of starting from an architecture level specification and creating a fully functional processor. This design methodology automates the creation of custom processors, accelerators and systems through the use of new hardware description languages, such as Chisel, as well as emerging integrated hardware development flows. We present OpenSoC System Architect an open-source, hardware generation, development environment which enables an arbitrary ISA specification to be translated into a processor implementation (RTL) complete with a LLVM-based compiler and cycle accurate simulator for the specified design. OpenSoC System Architect also enables integration with external IP blocks - including custom accelerators, caches, etc. to create a complete system-on-chip.

Cancellations of DoD acquisition programs have resulted in billions of dollars of losses annually, which reduces resources for new capabilities. One area identified is a lack of proper requirement scoping, which is prohibitively complex when tracing architectures for large systems. MBSE was developed to address these issues and provide a common framework to rapidly represent, convey, and synchronize information. This research explores the maturity of NLP and ML methodology needed to automatically generate and trace requirements in MBSE models, thus providing the tools to rapidly generate models from requirement documents and reducing the risk of program cancellation due to requirement scoping problems.

Nowadays, new services and current IT applications require an increasingly dynamic telecommunication network capable of adapting to the changing needs of users and network managers in a short time. So, in the recent years a new technological paradigm is born called Software Defined Networking (SDN) able to change strongly the way to interact with the network devices. These changes have been possible also thanks to the rapid development of a new technological concept: the Virtualization. With the introduction of the Network Function Virtualization (NFV) and the SDN architectures the software component is playing a key role in a lot of applications of the Information and Communications Technology (ICT). Now, thanks to these new paradigms, the networks are became ”smart” and ”programmable”. Then, the possibility of using adequate tools for performing test on networks and devices represents an important way to operate without the need of buying expensive hardware. Mininet is a software emulator able to manage a set of network terminals (hosts), switches, routers and also the various connections on a single Linux environment and it is able to simulate an entire network. In this paper, an overview on Mininet network emulator is provided showing its potential and scalability with different network topologies

The Aberdeen Architecture is a high assurance microprocessor architecture concept which implements Saltzer and Schroeder’s 1975 security principles in hardware. Current microprocessors execute instructions without any verification or authentication. Saltzer and Schroeder defined fundamental security principles: complete mediation, and open design. Complete mediation means to verify access rights and authority for every operation. Protection mechanisms should be based on open design principles: protecting keys, not design secrecy. In 2002, Mann describes how secrecy makes systems brittle and subject to catastrophic failure. The Aberdeen Architecture is high assurance computer architecture based on ‘open design’ principles, complete mediation, and RISC-V instruction set architecture. Aberdeen Architecture uses several hardware state machine monitors to enforce hardware security policies for the execution pipeline. The state machines’ security policies cover instruction execution, control flow integrity, data flow integrity, and memory access integrity. The individual security policies provide overlapping coverage. The security of the whole architecture is greater than the sum of the individual parts. The Aberdeen Architecture provides near complete mediation for instruction execution. This paper presents an introduction to the Aberdeen Architecture

Control in multi-agent hybrid dynamical systems – systems in which the state contains both discrete and continuous elements – is poorly understood. Theoretical results on state reachability and avoidablility typically rely on restrictive assumptions which do not hold in many important cases, hampering results in both trust and optimization of such systems. We introduce a flexible framework to enable control in multi-agent hybrid dynamical systems. We present an agent-based finite horizon temporal logic (FHTL) framework that enables mission monitoring and improves agent to agent collaboration in multi-agent hybrid systems under significantly lighter assumptions than required for similar infinite horizon temporal logic (IHTL) applications. We demonstrate our framework in an example scenario and provide both quantitative and qualitative analyses of the performance gains and mission trust monitoring enabled by our tools for this example.

We implement online deep learning for target behavior prediction. Our online deep learning algorithm provides an autonomous agent the ability to train in real-time while also shaping the frequency of training based on its current performance level. The benefits of our algorithm are twofold: (1) to enable an autonomous agent to train in real-time and continue to learn to accurately predict target behavior even while its target changes the strategy guiding its behavior, and (2) to achieve more efficient usage of its computational resources by managing its training frequency. This trained predictive capability is leveraged in autonomous decision-making to influence a target’s behavior by selecting those actions that produce a predicted response from the target that supports the end goal of the autonomous agent. In our scenario, the goal of the autonomous agent is to influence its target to circle the perimeter of the environment. We test our online deep learning algorithm in environments of varying sizes to demonstrate that the time it takes for an autonomous agent to achieve the target level of accuracy is directly proportional to the size of the environment.

Distributed beamforming (DBF) has received significant attention in the Radio Frequency (RF) domain. There are also potential performance benefits in the Underwater (UW) Acoustic domain. Specifically, in this work we investigate the application of DBF to UW sonar for Unmanned Underwater Vehicles (UUVs). First, we utilize distributed coding applied across multiple UUVs to enable beamforming gain. Next, we develop a multi-vehicle UUV motion model that emulates movement of the mobile DBF UUV array. The motion model enables our simulation model to induce position errors of the DBF array that a UW DBF sonar system might experience in practice. To ensure DBF beamforming gain in the UW environment, these position errors must be estimated and corrected during DBF sonar UUV mobility. In addition, each UUV will experience unique synchronization offsets, which are also estimated and corrected. For an N-distributed UUV element array, we show that our distributed coding method provides N² -gain in signal-to-noise ratio (SNR) on initial sonar transmission and reception, and N³ -gain thereafter. While our simulation model demonstrates these results for direct target returns, it is also theoretically possible to further increase the received SNR using multipath combining.

The evolution of commercial software and hardware development is designed to enable rapid innovation and product improvement in order to meet consumers’ ever-changing and growing demand for new capabilities. In contrast, Department of Defense (DoD) software and hardware development is tightly coupled with the acquisitions process: a heavily managed, regimented systems development process. The typical DoD process of research and development (R&D), accreditation, fielding, and sustainment limits the agile delivery of capabilities to the warfighter. This paper examines Defense Advanced Research Projects Agency (DARPA) efforts to incorporate commercial Development, Security, and Operations (DevSecOps) models adapted for DoD R&D. We present the research, development, and integration of secure data sharing technology conducted in the DARPA Secure Handhelds on Assured Resilient networks at the tactical Edge (SHARE) program aligned with DevSecOps. Building on lessons learned from industry, as well as previous DARPA programs, the SHARE program is poised to rapidly transition capabilities for operational use.

The original problem was to design a software defined radio (SDR) transceiver that met the size, weight, and power (SWaP) constraints of a small unmanned air vehicle (UAV). The initial demonstration was in six months, so a rapid prototyping approach was used to design the SDR with open commercial components. As the project progressed, additional objectives were levied on the design that had to be accommodated. This paper will describe the design challenges in more detail, and provide lessons learned and successes achieved through the use of open standards.

In multinational defence operations, either EU or NATO driven, the exchange of surveillance and reconnaissance data and information is an essential aspect to provide the commander with the needed situational awareness. This improvement of situational awareness, especially in a maritime environment, may be achieved amongst others by extending the Intelligence, Surveillance, Target Acquisition and Reconnaissance (ISTAR) performance through using unmanned systems (UxS) and integrating them into the combat management system (CMS), ensuring interoperability between the deployed forces and building the overall system based on a solid architecture. Within this frame, the OCEAN2020 (Open Cooperation for European mAritime awareNess) project, funded by the European Union's Preparatory Action on Defence Research and implemented by the European Defence Agency, saw 43 partners from 15 EU countries working together to build future maritime surveillance by integrating drones, unmanned vessels and unmanned submarines into fleet operations. Data and information have been integrated in a comprehensive (maritime) picture of developing situations, enhancing the situational awareness, and thus supporting military commanders on different unit levels in their decision making. An integrated system of systems has been developed and demonstrated in both simulated and live trials. The aim of the trials has been to showcase the enhanced maritime situational awareness achievable by employing different types of unmanned systems. The live trials have been maritime demonstrations involving command, control and communication assets, ships, communication centres, space centres and unmanned vehicles operating in the air, on the surface and in the underwater domain. For de-risking the live trials and for demonstrating advanced technologies, which cannot be demonstrated in the live trials, OCEAN2020 performed a series of simulated trials. This paper focuses on the development of the system simulation architecture and design for the simulated trials. The simulation components, their relationships and the used communication infrastructure will be presented.

The comprehensive and efficient supply of information to the armed forces is an essential prerequisite for making the right decisions regarding the further course of action based on knowledge of the situation in the area of interest. Reconnaissance and surveillance assets are employed at the strategic, operational, and tactical levels to provide effective information. A well-thought-out use of the sensory information potential, which brings the most suitable sensors in the most favorable combination to the right place at the right time, ensures optimal success in the timely detection of situations with potential danger, given the usually scarce resources. The criteria for optimal deployment are the reconnaissance objectives, the prioritized information needs of the responsible force commander, and the capabilities of the sensor carriers and sensors and their readiness for deployment. Sensor deployment occurs in allied forces as a result of the IRM & CM (Information Requirements Management and Collection Management) process. In this process, starting from the "Commander's Critical Information Requirements (CCIR)" within the IRM part of the process, the "Prioritized Informatioin Requirements (PIR)" are derived by means of refinement, from these the "Specialized Information Requirements (SIR)" and from these again the "Essential Elements of Information (EEI)" are derived. The EEIs are then ultimately transformed into "Collection Requirements" and processed in the CC part of the process. The "Collection Requirements" specify what kind of information is needed for which "Area of Interest" at what time and in what quality. The established "Collection Requirements" must then be processed in an "ISR Management" process in such a way that the appropriate and available sensor carriers are used with the appropriate sensor technology to perform the required reconnaissance. The ISR manager responsible for this process must consider not only the requirements resulting from the "Collection Requirements", but also the capabilities and status of the available assets (sensor carriers, sensors, endurance, etc.). This includes responding to dynamic changes during the mission, e.g., failure of a sensor or sensor carrier during reconnaissance of a high-priority target, resulting in the need to redirect a suitable sensor carrier and sensor from a lower-priority reconnaissance target. The paper at hand presents an approach for an optimal planning of ISR asset deployment in order to satisfy the information needs of a commander. Based on the processes of information requirements management (IRM) and collection management (CM), a two-step approach has been developed. In the first step, a set of suitable assets is assigned to each target on which reconnaissance or surveillance has to be performed. The assignment of suitable assets may be done by an operator directly, based on his experience and knowledge, or supported by an asset selection assistant component of the application, or the operator may choose to delegate the asset selection to an automatic component comprised of intelligent software agents performing the selection task for him. In the second step an automatic asset assignment and execution order is computed. The SW-Architecture of the application is a layered architecture; each layer captures and groups application specific aspects and details the components of the layer.

Continuously Updating Reinforcement Learning (CURL) demonstrates the ability to rapidly maintain deployed ML models when there is a change in use case such as a denied target with minimal performance effects. Traditional Machine Learning (ML) lifecycle requires ML models to be retrained and redeployed in order to maintain performance of deployed models experiencing changes in underlying data such as data drift. Data drift can include a wide variety of changes in data such as the addition of a new class, operating in an entirely new environment, mislabeled data, or subtle changes in targets over time. (CURL) deviates from this traditional lifecycle with dynamic updates using Reinforcement Learning (RL) to identify and capture data changes, and then automatically retrain the model with data changes. CURL learns to identify changes in data through its RL policy that is designed to maximize the reward for identifying changes in data. Specifically, CURL’s RL approach includes an environment with both the model’s performance and current prediction confidence as the observation space for the agent to act on the discrete action space, and reward function of the model’s accuracy subtracted by the labeling cost to learn data changes. Our controlled experiment demonstrated that the same distribution of denied target data (3%) was found by our RL policy, and our retrained model exceeded the initial classifier performance. CURL can be considered a general purpose technology that could be applied to a wide spectrum of fielded ML systems.

In operational settings, where nations collaborate within an overall mission and multiple assets are used to gather information in an area of interest, the timely distribution of relevant information to the relevant person is key. To enable current situation awareness relevant information needs to be provided in (near) real time. This information distribution is especially challenging in an environment, where organization units are spread out in the field. Those mobile units operate in environments with hardware limitations, network disconnections, bandwidth constraints, under mental pressure and in a challenging overall situation. Depending on the type of unit, the mission, the military organization and the specific task, the information needs and essentials are varying. Given the limitations described above, information gathered from other units thus needs to be tailored and prioritized to the need of the consuming unit within its specific mission. This paper deals with requirements, derived functions and technological approaches to deal with these challenges. Therefore, the paper first introduces a selection of scenarios of interest, where mobile units receive information from connected operational centers through dynamic multi-hop networks. For these scenarios we analyze the requirements for data dissemination considering the need for filtered and prioritized information for the individual task of the unit. To guarantee the interoperability of multinational deployments, these requirements must be compatible with the requirements of relevant multinational standards (STANAGs). Further derivable functions and resulting technological approaches will be topics for future work.

The demand for space systems that leverage a proliferated constellation model is ever increasing. With this increase in demand, cost-effective interoperability testing of system components prior to deployment has become of significant interest. The solution discussed within this publication utilizes cloud technologies to create, configure, and control emulation environments that aid in the development of software and/or hardware for space-ground systems. Our solution provides realistic testing scenarios in which high numbers of terrestrial and/or spacefaring system components interact dynamically. Our digital twin environment includes emulation of link communication dynamics between earth-orbiting satellites during maneuver and non-maneuver periods, as well as the associated ground segment. Communication system parameters, as well as propagated satellite orbits inform the network link dynamics during constellation emulation scenarios. Additionally, the solution provides mesh network routing capabilities such that satellite software payloads deployed to a cloud-based, emulated space vehicle can be exercised in an accurately modeled scenario. Analysis of system performance data captured during emulation help identify interoperability problems prior to system deployment. Cloud adoption of an emulated satellite constellation test environment provides an agile, cost-effective, and repeatable capability to assist in development across satellite programs.