China’s People’s Liberation Army Navy (PLAN) has overtaken the U.S. Navy in total ship count and in the speed with which new technology is being introduced in new ships. Since 2014, the PLAN has delivered two carriers, with an additional indigenously designed and much more capable one under construction, and 11 modern Type 052D destroyers, with another 9 being built. In the past 15 years, it has received 30 highly capable Type 054A frigates.1
By comparison, the U.S. Navy is moving slowly to deliver capability, both in new construction and in upgrades that ensure the relevance of the legacy force against a rapidly accelerating threat. To address this, the Navy needs ships that can be rapidly and inexpensively upgraded and not to be constrained by shipbuilding and maintenance timelines.
One approach to address this need is to separate platform from payload. Dr. Reuven Leopold argued, in his August 1975 Proceedings article, “U.S. Naval Ship Design: Platforms vs. Payloads,” that ships and their combat systems should be considered separately: “The characteristics of modern combatants indicate that the trend is back to the ancient role of warships as transport vehicles. Rather than transporting troops, however, they are carrying sophisticated collections of weapon systems.”2 In 2012, in his seminal Proceedings article, “Payloads over Platforms: Charting a New Course,” then–Chief of Naval Operations Admiral Jonathan Greenert argued for the “decoupling of payload development from platform development,” so that long-serving ships and aircraft could maintain relevancy throughout their service lives.3 The next step in the evolution of this thinking is to consider the payload as a platform, following the model of commercial cloud computing, and to define a standard software platform as the foundation for shipboard applications.
A software platform includes the development environment used to create an application, along with the “stack” of operating system, middleware, and virtual machine on which it runs. Software platforms are taking hold across industry, from commercial buildings to automobiles.4 They provide advantages that include faster software development, better cybersecurity through centrally managed vulnerability patches, and on-demand scalable computing resources. For the Navy, converging on a standard software platform also would enable greater sharing of data and faster response to fleet feedback using the high-velocity DevOps/enhanced security DevSecOps software model, while at the same time making software testing both more effective and efficient through modern testing tools.5 It also greatly simplifies the process for obtaining an authority to operate.6
Moving to a software platform can provide the horsepower and environment needed to employ the latest artificial intelligence and machine learning (AI/ML) algorithms. The 2018 Department of Defense Artificial Intelligence Strategy, Harnessing AI to Advance Our Security and Prosperity, notes, “AI is poised to transform every industry, and is expected to impact every corner of the Department, spanning operations, training, sustainment, force protection, recruiting, healthcare, and many others.”7 The dynamically scalable processing and storage capacity offered by cloud technology provide the foundation for the introduction of AI/ML to the Navy. This will improve the effectiveness and reliability of the full range of ship systems and accelerate the pace of capability delivery to the fleet. AI/ML applications that have clear utility to the Navy include automatic target recognition and sensor optimization; advanced cybersecurity using state-of-the-art intrusion-detection systems; predictive analytics to improve system reliability; automated situational-awareness augmentation systems that sift through vast quantities of data looking for patterns; improved user interfaces; highly realistic training systems that learn over time and can emulate threats; and fully autonomous unmanned vehicles.
Commercial cloud vendors offer two principal options for hosting their customers’ applications: infrastructure as a service (IaaS) and platform as a service (PaaS). In IaaS, computing hardware is furnished, usually with software virtual machines that enable applications to run in their own exclusive operating environment on shared servers, with processing resources and storage allocated on demand. In this approach, the user is responsible for the software stack.8 By contrast, in the PaaS model, in addition to supplying the hardware and virtual machines, the cloud provider delivers the operating system, middleware, and software tools and deploys version updates and patches for cybersecurity vulnerabilities.9
Too Many Networks
Today’s fleet has multiple server networks, including the Consolidated Afloat Networks and Enterprise Services (CANES) system, which provides the computing environment for command, control, intelligence, and logistics applications. There are other networks that support combat systems, such as the Common Processing System for Aegis, the Ship Self-Defense System, the Aircraft Carrier Tactical Support Center, the Surface Electronic Warfare Improvement Program, and another for the SQQ-89 shipboard sonar suite. In addition, the submarine force runs the Submarine Warfare Federated Tactical Systems on its Tech Insertion hardware. However, current technology makes it possible for the Navy to converge on a consolidated computing plant. Doing so would make the computing power, network storage, and graphics-processing units needed to support AI/ML available to all shipboard systems and, at the same time, provide shipwide redundancy to improve system availability. Moreover, the resulting computing platform could be scaled to support the smallest unmanned vehicle to the largest aircraft carrier, with resulting economies of scale in procurement.
A long-standing barrier to marrying the separate computing environments that support information technology to those that power combat systems is the requirement to support real-time latency and determinism. Latency is the time it takes for a system to respond to an event. Determinism in real-time systems is the quality of responding to an event in a consistent amount of time, every time. For many years, information technology systems based on probabilistic ethernet protocols could not achieve these two critical real-time capability requirements. Today, however, cloud providers have demonstrated the capability to meet real-time performance requirements, and the Navy has successfully run its most advanced combat systems in a virtual environment consistent with a commercial cloud approach.10
While achieving this standard software platform vision would go a long way toward bridging the gap between innovation in the lab and capability in the fleet, the complex ecosystem that spans from basic research through design, development, and sustainment that feeds new systems in the fleet must also be optimized. A multipronged approach is required, including engaging industry, experimentation and modeling of innovative technologies, incremental development and fielding, and a comprehensive data strategy. Systems must be developed, tested, and fielded through an end-to-end process optimized to rapidly deliver capability in cycles limited only by technology readiness. The Navy should seek companies, including nontraditional vendors, that have advanced technology, understand naval requirements, and offer solutions using their unique technical capabilities through streamlined contract vehicles, to ensure their participation in forums such as the Naval Surface Technology and Innovation Consortium and the Naval Submarine League Technology Symposium.11
Fleet–Technical Community Dialogue
Through the integrated DevOps/DevSecOps software development approach, experimentation, modeling, and prototyping need to be at the heart of the generation of requirements, with an open dialogue between the fleet and the technical community. The technical community must encourage scientists and engineers to work with the fleet to understand the user environment and identify evolving technologies to apply to emerging warfighting requirements. Similarly, the fleet must have exposure to new systems in development and have an opportunity to test prototypes, to see where updates are required, and to provide a foundation for developing concepts of operation, standard operating procedures, and potential for employment. The Navy created the Surface Development Squadron in San Diego, California, and Submarine Development Squadron in Silverdale, Washington, to address these issues, and the acquisition community should embrace them.
While the first step is to create new capabilities, the rapid fielding of new technology is key to outpacing U.S. adversaries. Risk can be mitigated during the development process, but new systems will be unlikely to fully meet desired requirements, and should instead target the 70-to-80-percent solution, with future spiral developments planned to address additional capability.12 The incremental development approach enables technology to mature concurrently with fielding and gives the fleet an opportunity to refine requirements and determine which attributes of a new system are most important and which areas need improvement.
The iterative engineering refinement process should include exacting reliability and production engineering to ensure that systems can be built economically and continue to perform to requirements over their lifetimes. Systems should be assessed in terms of capability, cost, and maturity both in comparison with existing capability in the fleet and in contrast to alternative approaches. Finally, the power of the network should be employed to quickly push out software upgrades and fixes, mirroring the approach taken by companies such as Tesla, which can rapidly field a software change across its entire fleet of cars.13 (See “Diving Off the Platform-Centric Mind-set,” pp. 26–31).
The Navy also needs a comprehensive data strategy that uses the computing power and storage capacity that a modern cloud-architected PaaS system provides. Today, most data received by Navy sensors and systems is neither preserved nor analyzed. Going forward, program offices should have funded data-collection plans that feed AI/ML systems and provide the foundation for other updates throughout a system’s lifespan, with a focus on continual performance and reliability improvements and reduction of cost over time. This plan to collect, analyze, and disseminate data for Navy systems should clearly spell out roles and responsibilities for the use of the data by warfare centers, university-affiliated research centers, federally funded research and development centers, the Defense Advanced Research Projects Agency, the Office of Naval Research, and the responsible acquisition program office, including its industry partners. The Navy has a key advantage here: No other navy today can match its global operational tempo and resulting potential to collect data from theaters around the world. Defining the data to be collected and putting in place processes to take what is learned and turn it into improved capability to the fleet is critical. To respond quickly to new threats, the fleet must also push data in near–real time back to shore facilities that have the capability to analyze it and make changes.
We Have Already Done This
There are both historical and recent examples in which an effective data collection methodology has been successfully employed. For instance, during World War II, ships were required to capture very specific gun-system performance data on standard record sheets and send it to the Fleet Readiness Division for analysis. From there it was shared with both the engineering and operational communities. Changes to systems and tactics were implemented quickly to address performance shortfalls. For example, the Mk 37 surface ship gunfire control system continually received updated tactical guidance along with 92 engineering upgrades during the war.14 In a more recent example, the Advanced Capability Build/Acoustics-Rapid commercial off-the-shelf (COTS) Insertion (A-RCI)/Submarine Warfare Federated Tactical Systems (SWFTS)/SQQ-89 Systems Engineering Measurement Program uses recorded data and assessment techniques to assess the performance of currently fielded capability and to define and prioritize future spiral updates. These have resulted in rapid upgrades to submarine and surface sonar systems that have more than paced the threat.15
The Navy can harness ongoing revolutions in cloud computing and software development to build a fleet that is both affordable and relevant throughout its service life and to provide the computing power and storage to support artificial intelligence. While Naval Sea Systems Command (NavSea) has recently sent out a request for information that outlines its desire to move to IaaS, and CANES already has an IaaS offering, NavSea and Naval Information Warfare Command and their associated program executive offices must come together on standardized, scalable IaaS hardware. Such hardware must support all computing requirements and be designed to be rapidly refreshed with minimal impact to the ship, an approach that has the potential for significant cost savings.16 Moreover, a common PaaS should be defined as the software platform that science and technology efforts target in their early designs. It should include industry standard development tools that support modern DevOps/DevSecOps software development, and it should be used across all ship development communities. This approach also provides the opportunity to tear down stovepipes and share resources across communities, such as using all communication paths to support an integrated and fully routable network and considering radar apertures and electronic support measure sensors holistically. Finally, the Navy must focus on the collection and use of data, fleet feedback, and prototyping to provide a constantly improving capability to the fleet that outpaces its adversaries.
1. “How Is China Modernizing Its Navy?” ChinaPower Project, 2018; Eric Wertheim, “China’s Luyang III/Type 052D Destroyer Is a Potent Adversary,” U.S. Naval Institute Proceedings 146, no. 1 (January 2020).
2. Reuven Leopold, “U.S. Naval Ship Design: Platforms vs. Payloads,” U.S. Naval Institute Proceedings 101, no. 8 (August 1975): 30–37.
3. ADM Jonathan Greenert, USN, “Payloads over Platforms: Charting A New Course,” U.S. Naval Institute Proceedings 138, no. 7 (July 2012): 16–23.
4. Ryan Fletcher, Nick Santhanam, and Shekhar Varanasi, “Laying a Foundation for Success in the Era of the Connected Building,” McKinsey & Company, 12 October 2018; Ryan Fletcher, Abihijit Mahindroo, Nick Santhanam, and Andreas Tschiesner, “The Case for an End-to-End Automotive-Software Platform,” McKinsey & Company, 16 January 2020.
5. CAPT Kurt Rothenhaus, USN, CDR Kris De Soto, USN, Emily Nguyen, and Jeff Millard, “Applying a DEVelopment OPerationS (DevOps) Reference Architecture to Accelerate Delivery of Emerging Technologies in Data Analytics, Deep Learning, and Artificial Intelligence to the Afloat U.S. Navy,” 9 May 2018 (Monterey, California: Naval Postgraduate School).
6. Timothy A. Chick, “Maintaining Your Authority to Operate (ATO) While Being Agile: Achieving Continuous Reauthorization with DevOps,” Carnegie Mellon University Software Engineering Institute, June 2018, 53.
7. Department of Defense, Summary of the 2018 Department of Defense Artificial Intelligence Strategy: Harnessing AI to Advance Our Security and Prosperity.
8. “What Is IaaS? Infrastructure as a Service,” Microsoft Azure.
9. “What Is PaaS? Platform as a Service,” Microsoft Azure.
10. Navy Program Executive Office Integrated Warfare Systems, “AEGIS Virtual Twin Successfully Intercepts First Cruise Missile Target,” Naval Sea Systems Command, 23 April 2019.
11. “NSTIC: Naval Surface Technology Innovation Consortium—Providing Innovative Technological Solutions of Complex Naval Warfare Systems”; “Naval Submarine League Submarine Technology Forum—Promoting the Importance of Submarines to the National Defense.”
12. Ronald O’Rourke, “Navy Force Structure and Shipbuilding Plans: Background and Issues for Congress,” Congressional Research Service, 10 June 2019, 65.
13. “Software Updates,” 2019, www.tesla.com/support/software-updates.
14. Buford Rowland, William B. Boyd, and William D. O’Neil, U.S. Navy Bureau of Ordnance in World War II (Washington: Navy Department, 1953), 377.
15. Michael Boudreau, “Acoustic Rapid COTS Insertion: A Case Study in Spiral Development,” 87, calhoun.nps.edu/handle/10945/592.
16. Andrew Eversden, “U.S. Navy Wants to Create a ‘Hardware Factory,’” C4ISRNET, 27 August 2020.