The Next SIMNET? Unlocking the Future of Military Readiness Through Synthetic Environments
The soldiers prepared for the day’s training exercise — a force-on-force armored battle against their Soviet adversary. Yet, instead of maneuvering through the sands of the Mojave Desert — the location of the U.S. military’s Fort Irwin training ground — the soldiers prepared to enter a virtual battlespace, called SIMNET (short for “simulator networking”). Ethernet cables snaked out from their M1 Abrams tank simulators, plugging the First Company of the 12th Armored Cavalry Unit into one collective virtual training ground. As the instructor uploaded Soviet computer-generated tanks and armored vehicles into the scenario, emulating Soviet doctrine to the best of his ability, the U.S. troops ranged across the virtual desert, in an attempt to outwit and out-fire their adversary. The Soviet and U.S. forces engaged, clashing in a fierce melee of man and materiel, but U.S. command and control quickly began to disintegrate. American troops began to virtually die at the hands of the adversary or, in the midst of the confusion, via fratricide. The troop screens in the tank simulators turned blank, as if to signal the onset of their mass extinction. The battle had ended, but for these soldiers the training had not yet finished. SIMNET allowed the soldiers to replay aspects of the battle “Groundhog Day-style” and identify their mistakes and failings. They could experientially learn before the crucible of combat, and not via a prohibitively costly, one-off exercise.
SIMNET was sponsored by the Advanced Research Projects Agency (the precursor to the U.S. Defense Advanced Research Projects Agency or DARPA) in partnership with the U.S. Army between 1983 and 1990. Conceptualized by Jack Thorpe in 1978, SIMNET was the first demonstration of an extensive simulator network for collective team training and mission rehearsal. Prior to the 1980s, simulators were developed as stand-alone systems designed for platform-specific training needs, like pilot instrumentation training. Thorpe, however, felt that simulation could augment live training, not just act as a substitute. He believed that a synthetic environment could be used to teach necessary combat skills — like large-scale collective coordination — that were challenging to learn in peacetime. SIMNET sought to solve that problem by developing a scalable and cost-effective virtual architecture that networked simulators together into one collective synthetic (i.e., virtual and constructive) training exercise.
SIMNET is now considered a clunky piece of technological history. In the 1980s, however, its development and use were considered a technological “revolution,” changing “the way the military does business” and also, as a result, reforming “the simulation industry.” Today, the United States appears to be poised, once more, on the cusp of what could be another revolution in the way the defense establishment deploys synthetic environments — not solely for training, but across the full spectrum of military readiness. Indeed, just as SIMNET linked disparate simulators into one virtual world, newer synthetic environments provide that same possibility across differing facets of military readiness, integrating environments that enable education, training, maintenance, and decision support, as well as force structure and modernization. As the United States works to overhaul its readiness reporting system, the time is ripe to employ these future synthetic worlds to unlock new analytic insights, while also improving overall military effectiveness. Indeed, such an ecosystem should allow the military to better answer that wicked question: Are they prepared for the future fight?
All but War Is Simulation
In 1992, the U.S. Army stood up their Simulation Training and Instrumentation Command the precursor to today’s Program Executive Office for Simulation, Training, and Instrumentation. The mission was to usher in the future of military simulation, and its website reflected that ’90s science fiction style aspiration — complete with a spinning logo that, on one side, depicted a soldier in a futuristic space-style suit equipped with a laser gun and, on the other, a traditionally clad soldier wielding a lightning bolt. The rim of the logo highlighted the command’s mantra: “All But War is Simulation.”
While Simulation Training and Instrumentation Command’s maxim may have reflected the importance of training simulations within the Army, it also pointed to a broader truism — the growing influence of synthetic environments, and modeling and simulation more generally, on military readiness. Indeed, synthetic environments underlie many facets of what has been characterized as “structural readiness” or “institutional readiness” — the ability of military forces to meet the demands of their assigned missions, to include decisions on force structure and modernization.
For instance, today, synthetic environments undergird education delivery, allowing warfighters in disparate locations to access educational content at their point of need. The armed forces use simulators in a range of training tasks and scenarios. Much like the M1 Abrams tank simulators connected in SIMNET, simulators, but also virtual computer games, are employed to teach warfighters the instrumentation of their respective platforms. In other training scenarios, they allow warfighters to demonstrate individual and unit proficiency in key mission tasks, from maneuver to various medical procedures. Augmented reality applications are deployed across services to aid mechanics with platform and system repairs. A simple augmented reality headset can act as a hands-free recollection aid, allowing a mechanic to pull up military system information while the platform is in service. Synthetic environments support the design and testing of new technologies and later, modeling and simulation can also serve as a key aid in acquisition decisions. Simulated wargames — whether for education, training, or course of action analysis — enhance discovery and assessment, allowing commanders and decision-makers to grapple with those “known unknowns” and thorny “unknown unknowns” that influence future force structure and modernization decisions. Models of military readiness determine “resource inputs,” such as flying hours, and in the best of circumstances, track readiness outputs, like the ability to place a munition on target. In short, synthetic environments underlie many facets of military readiness today, helping to unlock the current and future fighting potential of the U.S. armed forces. Yet, despite their immense utility, these environments, as they are currently imagined, are fundamentally restrictive. They operate in silos, limiting the ability of the military to benefit from their full learning and analytic power.
Many modeling and simulation tools — whether used for education, training, or planning — provide features that are vertically integrated into one monolithic framework or engine. That one engine is responsible for the entirety of the simulated world including the simulation’s scale, the visualization layers, physics, pathfinding, and AI. This is a limitation, as various engines’ models may not represent the complexity or the changing character of the future battlespace, such as the cyber or information environment. Other models frequently lack the fidelity afforded by alternative options available in the marketplace or academia. Likewise, unless scalability has been designed into these solutions from the start, monolithic frameworks can be insufficient in size or complexity. They may be constrained in their ability to simulate, for example, a dense urban battlespace, a congested coastal environment, or an incoming missile saturation attack. As a result, the end-users’ learning outcomes are circumscribed by the monolithic engine’s offering. The end-users — whether that is the warfighter, training provider, or wargame adjudicator — cannot necessarily pick the optimal solution to meet their stated readiness needs.
When attempting to measure the readiness of the force, current synthetic environments are not being exploited to their full analytic potential. Recent advances in cloud computing, data storage, AI, and sensors have ushered in an era where rich information exists on the maintenance and utilization of equipment, capabilities of future platforms, and the battlefield effectiveness of personnel. These descriptive, diagnostic, and predictive indicators of “readiness,” can be collected, tracked, and analyzed via synthetic environments. This analytic power should be capitalized upon, helping the military to achieve its desired “data dominant” readiness ecosystem, replete with real-time and predictive readiness data.
While there are some notable exceptions, like the U.S. Army’s planned Synthetic Training Environment and the Defense Department’s Total Learning Architecture, many synthetic environments are constrained in their ability to capture the totality of actions or changes within a training event or simulation. Indeed, synthetic environments are often stitched together into a larger synthetic world or “federation.” While this does allow the end user to “play” in a more complex environment, the means, however, by which this is occurring are often dated and analytically limiting. The technical standards used to construct interoperable synthetic worlds, like Distributed Interactive Simulation or High Level Architecture, were developed in the ’80s and ’90s, respectively. While the synthetic environments interoperate, they hold separate data repositories — keeping the majority of their simulation state locally within their own environment. No canonical view of the entirety of the synthetic world exists, which is needed for true data exploitation and analytics.
Furthermore, outcomes from these various synthetic environments often remain siloed. For instance, lessons learned from one training event do not necessarily inform follow-on training, despite the best intentions of initiatives such as the Joint Lesson Learned Program. Training providers have bemoaned to these authors how despite writing detailed after action reports, key lessons are not always integrated into future training events. While some training providers point to a deliberate desire to ignore certain training findings, another likely cause is that key findings are buried within broader defense documentation and reporting and not easily accessible or retrievable. There is an urgent need for future systems to enhance information sharing across training events, services, and the total force.
In the absence of clear institutional processes and support technologies, such a situation risks making information sharing another checked box with no real follow-up. Much as how some wargame designers argue that their creations should be integrated into a broader “cycle of research,” synthetic environments should facilitate information sharing across various aspects of readiness, with minimal human effort involved. Indeed, findings from simulated training events — like the demonstrated readiness level of a unit — could be slotted into other synthetic environments, like those used for decision-support and concept development. Thus, information captured within one synthetic environment could support a broader ecosystem of learning and analysis — from education, to training, and decision-support.
The Department of Defense can do better. After all, technology already exists to support a different, and more powerful, military readiness ecosystem. Acquiring the “SIMNET of military readiness” is not solely a technical challenge. Rather, it requires organizational change and vision.
The SIMNET of Future Military Readiness
SIMNET was considered a revolution by some, not solely because it was a successful technological gamble, but also because it led to a fundamental transformation of the military’s approach to training. It altered the status quo within the simulation industry. It created the impetus for the military to commit significant resources to the development and deployment of large-scale networked virtual environments, and it forced a change resistant simulation industry to adapt, from bespoke and expensive solutions to more commercial-off-the-shelf options.
Synthetic environments are at a similar inflection point today. Much as how SIMNET’s distributed architecture benefited from emerging technologies in its day (e.g., microprocessors, computer image generators, and communication technologies), today’s technologies could catalyze a much more advanced military readiness ecosystem. Indeed, progress in distributed and edge computing now allow users to access richer and more immersive worlds without latency and throughput challenges. Dispersing computer processing across hundreds, if not tens of thousands of machines, can allow multiple simulation engines and models to run concurrently and in seamless coordination — thereby, simulating a synthetic world at near-limitless scale. Hybrid clouds allow the military to leverage private cloud or on-premises infrastructure, while simultaneously utilizing the benefits of the public cloud for their synthetic environments, as appropriate for security and point-of-need requirements.
These technological solutions, if properly leveraged through a modular open systems architecture, allow the military user, training provider, or wargame adjudicator to seamlessly integrate the models and engines that best meet their learning needs. Indeed, much like how Tetris required players to interlink differing shaped blocks into one playing field, a modular open systems architecture, with well-documented interfaces governing connection, should allow different models and engines to be interlinked, like a puzzle.
The need for such a modular approach to acquisitions, in general, was highlighted in the 2015 National Defense Authorization Act as an important goal across crucial mission areas. However, many military requests for future synthetic environments continue to pay only lip service to the concept, failing to define requirements that would truly implement such an adaptable architecture. Instead, all too often, requests call for a single engine to support the entirety of the synthetic environment, without also requiring that same engine be broken down into interoperable components, or composable puzzle pieces, that can be changed as the military’s needs evolve. Requests for proposals that focus on synthetic environments should seek to genuinely cultivate this sort of modularity across military simulation and serious gaming capabilities, thus breaking free of the more limited solutions tied to legacy monolithic simulation engines.
Adopting such a flexible architecture will also unlock deeper analytic value. To the extent multiple engines are hosted and synchronized, on one simulation plane, the military will have access to an extraordinary amount of information. Much like in immersive video games, a synthetic world could track and capture all player moves, achievements, failures, decisions, and much more. This data, when combined with the analytic power of AI, could better inform readiness outputs. For example, today’s simulators for individual pilot training can track and assess maneuvers in real-time, thereby providing tailor-made feedback to trainees. Wearable devices create opportunities to measure individuals’ physiological responses, allowing for wargames or training demands to be adjusted in real-time to meet learning needs. And the wargame community has started to explore the use of AI to identify player biases in games, provide real-time planning support, or rate player performance. Each of these capabilities should be built upon as AI algorithms and data sets are further refined and developed. The key, however, is to ensure that this information does not remain in silos, but instead informs a wider military readiness ecosystem.
Implementing such a solution, however, isn’t merely a technology problem — it’s also an organizational issue. Just as multiple engines and models can be hosted within a single system, each operating in seamless coordination, a similar approach could be taken with the diverse range of synthetic environments. These different systems, each made for distinct purposes, could collectively support and inform education, training, maintenance, or mission planning — all in harmony. Such an ecosystem naturally requires leadership and change within the Department of Defense, as shared standards or “common plumbing” would need to be set and enforced across differing synthetic environment programs. Without common standards baked into synthetic environment requirements, the integrative potential of these environments will be left to defense contractors, picking and choosing how to assemble the various pieces of the puzzle.
SIMNET met a discrete training need — it plunged warfighters into a collective training environment, allowing them to thrive in the chaos of a simulated battlespace before deployment. Technologies are available today to do much more than that. With some organizational reform, the military can durably transform the way it approaches and measures readiness, allowing it to preserve its warfighting edge in the near and long term. A readiness ecosystem scaffolded around synthetic environments is essential to ensuring that end.
Jennifer McArdle is a product strategist at Improbable LLC and a fellow in defense studies at the American Foreign Policy Council.
Caitlin Dohrman is the president of Improbable LLC and general manager of Improbable’s U.S. defense business.
Image: U.S. Army