Developing an Agile Space Architecture: Bringing the DoD into the Information Age
The U.S. military heavily relies on space-based capabilities, many of which have their origin during the Cold War. In an era defined by the access to and need for continually improved communications, the Department of Defense (DoD) lags behind both the private sector and more recently near-peer competitors. This can be attributed to the requirements and acquisition processes the DoD employs. This article examines the acquisition system, comparing it to the technology sector, then looks at using this technology sector development practices for space-based capabilities which the Department could develop into an agile and robust space architecture against peer adversaries.
The differences between DoD and other sectors are startling. First, we must understand how DoD acquires things. For an acquisition program, DoD produces requirements, or system parameters, through the Joint Capability Integration Development System (JCIDS) process. These documents most often articulate the art of the possible when it is written, with little or no thought about future expansion. After more than two years of coordination to produce a document – implying the technology is already out of date – DoD can finally begin to develop the new system. Repeat the process for production and the military finally gets the new system out to the field to service members. From concept inception through execution, this process can take ten or more years to complete, though the JCIDS process recently has provided some flexibility for producing information systems. The challenge with this development cycle is not just that it takes too long, but that the wrong requirement or idea may be identified, and the inflexibility of the system does not allow for the DoD to shift in a changing security environment.
Rather than trying to plan an ideal development path, consumer electronics and technology companies encourage experimentation of new ideas and concepts. This often leads to various ideas and teams competing, while also extensively collaborating and communicating in a loosely structured environment. This allows engineers to fail early and often in a project, learning critical lessons contributing towards the finished version. Once a set of teams has finished the development of a new product or technology, they begin working on improvements for the next model, up to several models ahead of the current production version. Even companies such as Apple, which use extensive networks of suppliers and contractors, continually invest in pushing beyond proven technology because their competitors do the same. These companies continually push their suppliers for greater performance, often co-opting the R&D and design of new technologies and devices at the same time as pitting several suppliers against each other.
To improve their understanding of the market and customer needs, many Silicon Valley companies engage in extensive beta testing and user studies throughout the development process. End-user and market feedback is critical; companies like Google rely extensively on user feedback throughout the life cycle of a product to guide subsequent improvements and development. Hundreds of people over millions of hours contribute to the development and production of the many finished, polished products and services that come from Silicon Valley. These products, however, cannot truly be separated from the many iterations of beta tests, prototypes, messy experiments, and the need for constant improvements to bring a product or company closer to perfection.
Rather than focusing on the large, exquisite systems, DoD should focus on smaller systems using incremental improvements. Space, particularly spy and communications satellites, can provide a good illustration developing this incremental approach. Spy satellites currently are behemoths launching thousands of pounds into low earth orbit (LEO) and beyond, depending on the required mission set. The developmental costs of these systems are millions of dollars alone, with many more millions spent on procurement. Using the Space Based Infrared System (SBIRS) as an example, the program encountered a Nunn-McCurdy breach because development costs exceeded $10 billion. Based on the Air Force’s most recent contract for GEO-5 and GEO-6 is validated at $1.86 billion, more than $900 million per satellite. Developing the spacecraft is only a single component in developing a constellation, the other significant contributor is the launch itself. Based on technical descriptions, a Delta IV rocket – the Air Force’s heavy lift rocket – is capable of launching more than 14,000 lbs into Geosynchronous transfer orbit, the most stressing scenario. While this is an extremely capable system, it costs approximately $435 million per launch based on the Air Force’s recent EELV contract. Atlas V, the other United Launch Alliance medium lift compatriot, costs $153 million. SpaceX, the newest entrant into the rocket business, has reported that a Falcon 9 rocket, their competitor the Atlas V, costs $62 million – significantly lower than others but still a substantial sum. All told a single satellite costs close to $1 billion. This does not include the cost of ground stations to control these assets, which can cost billions in their own right – such as FAB-T and OX-C.
An alternative to these large systems would then be using cube satellites. How might a cube satellite architecture work? Current technology would not provide the same high-definition picture quality today’s large satellites provide. This would be offset by cost and quantity; a single educational cube satellite is approximately $50,000 – approximately $10,000 for the spacecraft and $40,000 for the launch – so it is possible the military could develop a cube satellite with a camera for less than $100,000. At that price, the DoD would be able to launch perhaps 20 satellites at a cost of $2 million into a condensed area. There would be a distinct advantage having twenty or more sensors in an area, providing better situational awareness compared to a single large sensor. These constellations could also operate on multiple frequencies to prevent jamming. The small size and low cost of these miniature spy satellites would not have to carry any fuel (further reducing launch costs), allowing their orbits to decay after five or six months, at which point the latest cameras and sensors would be launched as replacements.
Due to the light weight and relatively small size, these rockets could be utilized by forward operating forces. First, a rocket with 20 satellites onboard would have a complement of capabilities including navigation, communication relays, and intelligence sensors across multiple platforms. The low weight and limited size of the payload would potentially allow rockets to be designed for the vertical launch system utilized by surface combatants and submarines – to further reduce development costs there is the potential of using standard missiles, such as the SM-3, to launch payloads into LEO. Aircraft might be able to launch five to 10 systems from fixed-wing aircraft operating at their high altitudes and speed. This approach would be able to utilize physical domains – subsurface, surface, land, and air – to resupply space forces. This would provide a significant improvement over the current two-site launch system in place, reducing operational risk in the space domain by drastically changing the targeting dilemma for adversaries – particularly when every fighter, surface combatant, submarine, and even air defense system becomes an enabler launching space sensors.
Automation would significantly improve the utility of these disaggregated sensors. Current systems on unmanned aircraft are currently able to identify ships by hull number, as well as track convoys and fighting positions. Combining many disaggregated sensors with a powerful automated processing center could reduce manpower to analyze information. Through Moore’s law, the processing and exploitation of intelligence data might be performed on the sensor itself, before disseminating the information to a command post, integrating individual sensors into a broader battlefield picture. Fewer people could make better decisions by combining small sensors performing intelligence analysis onboard the system itself.
There are several issues to consider if DoD were to use this approach. First, using a distributed architecture would enable more information, though at lower quality, to be analyzed. In an era of “big data,” access to more data points would allow better analysis of the battle space and potentially provide predictive capabilities. It is unclear, however, if current miniature technologies would be able to provide target-level data commanders’ desire. Second, this incremental improvement scheme would be more responsive to military needs, and most likely would cost less in the long run – for the cost of a single large satellite the DoD could launch 200 smaller constellations, if not more assuming prices were to decline due to economies of scale. Third, there is an inherent military, and commercial, advantage of having disaggregated sensors providing real-time information. This same approach could be used not just for spy satellites but also communications relays and potentially command and control assets.
On the negative side, it would most likely take years for overall constellations to provide the same decision quality data a single satellite currently provides. There is potential for significant increases in training costs using this approach, particularly if interfaces with new generations of cube satellites change drastically. Finally, this approach requires a relatively large industrial base constantly working to add new technology and capability to a system – this includes both companies and engineers. This larger industrial base, however, might be able to provide lower unit costs due to competition if DoD spread satellite block buys evenly across different competing entities. Overall it appears that the benefits would outweigh the risks.
It must be noted that cube satellites are not perfect substitutes for larger spacecraft. Rather this approach attempts to provide commanders and warfighters with more flexible response options should an adversary attempt to limit U.S. access to space. Cube satellites would not be sufficient for providing capabilities beyond LEO. These systems would not be sufficient to provide a global architecture for precision navigation and timing for instance, which require updates from ground systems to maintain accurate timing. Nor would they have the fidelity or longevity to serve in highly elliptical polar orbits which require longevity and persistence. Finally, these cube satellites would probably not be sufficient to serve as secure communications during a national emergency.
Mr. Hoehn is a national security consultant specializing in futures and capabilities development. He has extensive experience on the Air Staff and Headquarters Marine Corps. The views expressed in this article are explicitly the author’s and do not necessarily represent the views of his employer or the Department of Defense.