STITCHES
System of Systems Technology Integration Tool Chain for Heterogeneous Electronic Systems
Too often, the DoD thinks that the way to solve interoperability problems is to create new global standards. But the reality often looks more like this (as conveniently captured by https://xkcd.com/927/)
Even when the communities involved all agree (which is a long and arduous task unto itself), the result is a new standard that is now even harder to change than before (because there are even more adopters with sunk costs). To hedge against future changes, these global standards incorporate optional structures and flexible configurations – which adds enormous configuration and runtime complexity, making the implementations fragile and slow. At some point, when a new system requires data not supported by the standard, the dam breaks, and everyone shifts to a new version of the global standard (breaking interoperability with the past capabilities) – and the process starts again.
Current industry approaches to interoperability also don’t work in the DoD. For example, while the practice of forcing consumers, on a regular basis, to buy new hardware and upgrade to current software works in commercial settings, it is untenable for the military, where the costs of fielding a new jet far exceed the costs of purchasing a new version of an iPad. As a result, the military needs to simultaneously support many different generations of hardware and solve the general version of the interoperability problem – something that the commercial world simply avoids.
STITCHES lets systems interoperate together without requiring them to share a common standard by recognizing that interfaces can be related to one another. This set of transformations can be chained (George can talk with Alice, who can talk with Bob, who can talk with Fred), allowing automatic generation of interoperability code between systems not directly connected together. STITCHES leverages the system specification (without the systems themselves) and modern compilation techniques to replace the chain with a single optimized transform and autogenerate an executable binary that provides efficient interoperability at runtime.
We store this information a graph structure, which we call the Field and Transform Graph (FTG). Because the relationships within the FTG are used compositionally, as more interfaces are added, N^2 interoperability can be achieved through linear work. These improvements help avoid the need for big-bang integration efforts that handicap modernization of many large systems.
The result is the rapid generation of new mission capabilities that have the efficiency of hand-optimized bespoke solutions (often used in the military) with the maintainability and flexibility of abstracted designs used in the commercial world.
See this presentation to learn more about the STITCHES technology.
STITCHES has been used to achieve faster integration times across a number of domains, including distributed ISR, integrated airborne and distributed command & control, and integrated fires. While these sorts of integration efforts would normally take years to do, STITCHES has enabled this integration to occur within a few months (faster if there is good access to the core systems) – and this includes operational capabilities, not just lab demos.
Besides compressing the development timeline, by producing code tailored to the specific set of desired interactions, STITCHES also provides improvements in both runtime performance and security (because auto-generated code tailored to each use removes unnecessary code and interaction paths, decreasing the attack surfaces).
Looking forward, STITCHES is poised to change the way that the military integrates its systems. The 2021 National Defense Authorization Act, Sec. 804 requires that the DoD rethink the way that it integrates systems and names STITCHES as a key enabling technology, along with System of System Integration Technology and Experimentation (SoSITE, the DARPA program that funded STITCHES) (see page 348 here). In 2022, STITCHES became an Air Force program of record (more here). In addition, key industry leaders have recognized the value of STITCHES. According to Justin Taylor, Vice President of Artificial Intelligence at Lockheed Martin, “Seamless all-domain integration requires the ability for warfighters to bridge a data exchange gap across DoD and commercial standards. STITCHES plays an important role in ensuring our warfighters have the right data when they need it to successfully perform their mission.”