The history of U.S. MILSATCOM has been one of aggregation of mission capabilities over time. The result has been todays limited number of large satellites of ever-increasing complexity. Core U.S. MILSATCOM will span just 14 satellites by 2020 [Advanced EHF (AEHF): 4, Wideband Global SATCOM (WGS): 6, Mobile User Objective System (MUOS): 4]. Though new capabilities have been fielded and proven valuable, the consequences of this approach have been profound, resulting in fewer assets, increased cost and schedule uncertainties, and increasing delays in the fielding of new capabilities to support the warfighter. Recent budgetary constraints combined with program performance issues have forced a reevaluation of future requirements, procurement approaches, and technology roadmaps.
Additionally, these systems are not tiered, with the only augmentation being polar EHF payloads for protected communications, and leased commercial SATCOM transponders. While there are benefits to an aggregated approach, significant drawbacks have also become apparent in the form of program overruns and delays in the deployment of new capabilities. Long cycle times delay technology refresh and modernization, and this can cause the development of vulnerabilities over time. A disaggregated approach avoids many of these drawbacks by using shorter, focused missions to create a more cost-certain and incrementally deployed architecture to meet mission needs.
The Rise Of Complex Systems
Early MILSATCOM programs were largely single-mission in nature, providing somewhat specific services. Examples of these programs include Marisat and IDSCP (DSCS Phase 1). Missions were often created around frequency bands best suited for these particular services: UHF for narrowband services, X-band for wideband trunking, and so on. Individual satellites were neither particularly complex nor costly.
The next waves of MILSATCOM included larger, more complex, and more costly satellites: DSCS (Phase II, III), UHF Follow-on (UFO), and Milstar. Capabilities expanded greatly. DSCS provided phased array antennas for robustness and Milstar added on-board processing, agile beam antennas, and nulling antennas, for example. As the satellites grew, so did the required launch vehicle size, and the cost to launch. While early satellites could be launched on small or medium launch vehicles, a highly aggregated satellite such as Milstar required a Titan IV.
As satellite mass is a common metric often used in parametric cost estimating relationships (CERs) , it is useful to examine the evolution of the mass of U.S. MILSATCOM satellites over time. Figure 1 is an illustration of the mass growth of core U.S. MILSATCOM systems  since their inception: narrowband, wideband, and protected EHF systems. Of note are the striking increases in moving between major program blocks, such as those in the late 1980s and early 1990s, a trend that continues today. Systems that increased in mass often added missions, in addition to improved performance and added functionality for the existing missions. WGS aggregated heritage wideband X-band missions and the Ka-band Global Broadcast System (GBS). MUOS includes a new wideband code division multiple access (WCDMA) payload to the legacy UHF payload, creating a hybrid UHF/Ka-band system.
The latest wave of MILSATCOM programs began experiencing significant delays and cost overruns early in their development periods. Figures 2 and 3 show a comparison of cost growth and schedule delay of U.S. space programs, including MILSATCOM programs, over a 15-year period. A conclusion is that with increased complexity, often brought on through the aggregation of mission requirements and resulting capabilities, has come difficulties in adequately assessing cost and schedule risk due to underestimation of the level of technical and programmatic effort required to meet the required milestones as planned.
In a RAND Corporation report to the U.S. Air Force in 2008, this dependence on complexity for space programs was noted as a driving factor in cost and schedule overruns of DoD space programs:
The much-increased complexity of military space systems netted together into larger systems of systems posed many new challenges for the space acquisition and cost-estimating communities. The so-called Young Report , a widely influential joint Air Force/DoD study published in 2003 on the growing challenges in the space acquisition process, pinpointed the system-of-systems concept involving multiple users and extensive user requirements as a major contributor to the causes of cost growth on military space systems. It noted that the proliferation of users and requirements led to increasingly complex systems of systems, which greatly increased the difficulty of managing cost, schedule, and risk. At the same time, the complexity of individual systems, subsystems and technologies was also increasing, as sensors and other payloads became technologically more sophisticated, and much more complex processing and software tasks migrated to space vehicles. 
Alternative solutions and architectures should combat the growth of complexity seen to date. It should have the characteristic that allows growth to be better managed as well. One method of accomplishing this is to evolve more highly disaggregated architectures, in the process introducing greater diversity and flexibility for the Government. Figure 4 shows an example of such a disaggregated architecture, one that is both distributed and tiered, with a clear hierarchy that is diverse and includes both core assets and augmentation assets.
There are two primary ways to disaggregate or diversify. The first is the decomposition of a large, aggregated system into smaller components. The second is to augment the core system by adding smaller components to create a more diverse and proliferated system of assets. In space terms, a system such as Iridium, a global commercial SATCOM service provider, represents the former, while a system such as SBIRS may be the first step towards the latter (with HEO payloads augmenting the large GEO satellites). Here we examine both, with an example for each: tiered augmentation of the WGS system and decomposition of the AEHF system.
The first case is illustrated in Figure 5, and shows the addition of smaller satellites and/or hosted payloads to supplement the capabilities of the core system. In this notional example, the new assets are tailored for specific military Ka-band missions, such as tactical (or intra-theater) AISR, long-track AISR, and Communications on the Move (COTM). Specific orbital locations are selected according to immediate COCOM needs and to ensure non-interference with the WGS constellation. The new assets are both hosted payloads and free flyers.
The second case, illustrated in Figure 6, shows disaggregation through the decomposition of the existing AEHF satellites into separate satellites or payloads to separately support both the strategic and tactical missions. In this case the resulting satellites are each significantly smaller and less complex than the original.
Benefits Of A Distributed, Disaggregated Architecture
Many methods can be employed to make a system more resilient. The Milstar system incorporates autonomous recovery in individual satellites to assure that strategic links are restored within a certain minimum time period following an attack. SATCOM service providers employ on-orbit hot sparing of their commercial satellites to recover from on-orbit failures. A third option is to maintain payloads or satellites in ground storage, available for quick call up for launch on demand to replace lost capability.
Using disaggregation to disperse mission capability among a larger population of satellites and thus limiting the impact of the loss of any single one through system-level redundancy provides a means of achieving resiliency. The level of similarity among satellites is a trade. If all possess the same basic capabilities, then system redundancy may be maximized, and there is less motivation by an adversary to target any specific satellite in the system. If the object is to customize the capability by geographic region, for example, then this benefit may outweigh the desire for strict commonality and associated cost savings.
A distributed architecture should be cost competitive with existing systems to be attractive. When considering the benefits, cost must be considered; although in the end this may not be the most compelling factor in selecting the optimal degree of disaggregation. In addition, there is value in increased cost certainty for budgeting purposes, as well as overall cost.
Three major cost components dominate the space segments total cost: the developmental (non-recurring) cost, the production (recurring) cost, and the launch cost. Each of these costs is highly correlated to the size and complexity of the satellite. These costs are also influenced by longer timelines for large, complex programs that have certain required fixed costs for large organizations such as program office, contracts, security, and other functions that do not directly affect the end product. As program timelines lengthen, these fixed costs increase proportionally. Thus it is always beneficial to execute in the shortest span possible.
Developmental costs: The non-recurring engineering (NR) costs in the developmental phase are usually much higher than the recurring cost of the first satellite (T1). Typical NR/T1 ratios for space programs are historically greater than 2:1. A number of recent studies have shown correlations between development costs and complexity for both software and hardware . Generally this cost relationship is not linear, but becomes somewhat exponential at a certain level of complexity. Smaller, disaggregated payloads and satellites have the potential to maintain a complexity level below the knee of the curve. To attack developmental costs, we need to attack complexity, and its collateral impacts upon system development.
Production costs: The recurring engineering (RE) costs track more linearly with satellite size and complexity. As there are certain base costs associated with individual satellite builds, including program management and systems engineering, this cost element may favor the larger satellites, given similar technologies. On the other hand, the larger quantity of smaller satellites and payloads will yield some accelerated learning curve and greater block buy cost efficiencies.
Consider a satellite with a recurring cost of $1B. If the satellite could be disaggregated into two identical satellites costing $500M each, and each built in half the cycle time, we can compare the effect of the learning curve as shown in Figure 7. It is common to use an 85 to 90 percent cumulative average learning curve for aerospace programs . A 90 percent factor was used in this example. As we see, the learning benefits will be realized sooner with the smaller product.
In this case the smaller satellites total cost is $3.5B for 10 items, while the larger five satellites would cost $3.9B, or about 11 percent more, due solely to fact that the higher volume of the smaller product takes greater advantage of the learning curve than for the shorter production run. And the functionality of the disaggregated satellites would be deployed more quickly than with the aggregated case, as half increments could be fielded twice as fast.
Eventually, learning curve savings and production efficiencies are offset by obsolescence costs during long-term production. Unless a lifetime buy is made for high-reliability and high value parts, obsolescence costs will increase the recurring cost over time and add some non-recurring costs for redesign as well. At some point the flattening of the learning curve may result in obsolescence costs overcoming any future learning curve savings. It can be argued that at this point the continued extension of the product is costing the program without returning value other than the immediate cost avoidance of non-recurring engineering to modernize the product design. The arrival of this cross-over point, notionally illustrated in Figure 8, is one indicator that the product may be past its useful life and requires a major redesign.
Maintenance of skill sets and material inventory is also more difficult with longer cycle times. If obsolescence can be contained, the larger satellite is more likely to enjoy a slight cost advantage for recurring costs. However, by that time, another evolutionary block upgrade is likely to have occurred, effectively resetting the learning curve again.
Launch costs: Launch costs represent a significant percentage of the total space segment cost for Government programs. Disaggregation results in a greater number of launches. If launch cost was directly proportional to lift capability, launching twice as many satellites of half the size and complexity at half the cost would be roughly equal to launching half of the number of aggregated satellites. However, launch vehicle cost is not proportional; there are a limited number of options with fixed capacities available. U.S. space policy further limits the allowable launch vehicles for MILSATCOM satellites to U.S. supplied items only.
Figure 9 shows a comparison of several medium and heavy launch vehicles (MLVs) suitable for launch to GEO. For the U.S. EELV launch vehicles, the launch cost is amortized over a large number of launches and thus there is little benefit in using a smaller vehicle. The Delta II is still available, though, at reduced cost for medium payloads. If non-U.S. launch vehicles can be considered, the number of alternatives for medium launch vehicles (MLVs) expands (e.g., Proton M), and lower prices can be found. In the future, new U.S. entrants such as the SpaceX Falcon 9, with a price of $60M, could further encourage launches of smaller assets much more affordably. If smaller satellites could allow a dual-manifest launch, the cost for a single satellite could drop to $30M.
The introduction of hosted payloads can also significantly improve affordability. Hosted payloads are MILSATCOM payloads hosted as a secondary mission on either existing U.S. Government or commercial satellites. For smaller payloads, hosted payloads will in many cases prove to be a more affordable alternative, due to the cost sharing associated with fixed costs such as launch and accommodation costs and a hosting fee. These are balanced against the free flyer cost of the spacecraft bus and full cost of the launch vehicle. The reduced cost for hosted payloads will at least partially balance the greater number of assets in the disaggregated architecture.
Figure 10 illustrates the relationship between cost per capacity and payload size and capacity for both hosted payloads and free flyers. Actual cost values will be highly dependent upon the specific mission(s) and overall payload design and capacity. Free flyers can maximize affordability by more completely using available satellite and launch capacity at reduced cost per communications throughput (e.g., $M/Gbps) but sacrificing some mission flexibility.
Other costs: Other cost advantages may be realized by disaggregating missions. In the case of nuclear hardened missions such as AEHF, disaggregation provides an additional cost reduction. The combination of strategic and tactical missions levies a survivability tax on shared hardware (including the bus) that must meet the more stringent nuclear requirements. Disaggregating frees the tactical payload of those requirements.
Though the focus here is the space segment, attendant ground segment and mission planning costs may also rise with a more disaggregated architecture. However ground assets are generally more scalable, and can often be shared among many spacecraft. In addition, todays highly networked terrestrial systems provide extensive interconnectivity to allow coordination of system management activities.
It is clear that many factors influence the system life cycle cost, some positively and some negatively. The challenge is to develop a solution that takes advantage of the more affordable aspects of a distributed architecture.
Disaggregation also can respond to the need to be more agile by more rapidly fielding new capabilities in response to changing threats and environments. Programs with long cycle times effectively remove the option of rapidly making changes to counter the moves of an adversary if a new vulnerability appears.
One response to this concern has been to add flexibility to MILSATCOM payloads such that they can be reconfigured from the ground. Specific examples include the inclusion of more on-board digital processing and software, and configurable components such as phased array antennas. While these all provide flexibility, they do not necessarily add capabilities, and all represent cost and schedule drivers to the program, extending cycle times even further. Though software content in space continues to increase, a significant amount of functionality is determined by application specific integrated circuits (ASICs) and firmware, neither of which can be modified after launch. In the distributed world, smaller responsive payloads could be launched to augment the core missions with very specific capabilities in a much shorter period of time, addressing new vulnerabilities in a time-critical manner, resulting in improved security. To borrow a term from the software world, this would effectively be a security patch at a system level.
Technology Refresh + Modernization
Disaggregation also provides more technology insertion opportunities with reduced risk for any single opportunity. With only a relatively small number of satellites, the opportunity to enhance the products are limited. In fact, as the number of assets shrinks, there is more conservatism, leading to upgrades only occurring as major block changes, which can take a decade to implement. With smaller satellites, there is a reduced risk of inserting new technologies for each, including demonstrations. The larger, core programs can be less disturbed, integrating new capabilities at a planned interval, after the viability of the technology has been proven. This allows technology development and demonstration to be effectively decoupled from larger, ongoing operational programs.
Examples of technology evolution applicable to space payloads are the increased speed and density of microprocessors, increases in power efficiency for a wide variety of semiconductors, and the higher capacity of solid-state memory. Cycle times for incremental improvements of these parameters are often between 12 and 24 months. Figure 11 summarizes the rate of growth in integrated circuit (IC) complexity and processor power over the period 1965 through 2005. These gains are realized in the designs of spacecraft digital processors, digital storage units, encryptors, sensors, and other functional blocks, among others.
The disaggregated architecture can also provide functionality and capabilities that are difficult to implement with more aggregated ones. In particular, modern networks, including the terrestrial and cellular systems, and the Internet, depend upon distributed architectures. These architectures allow for high levels of redundancy, yielding extremely high availability. A trait of these networks is a large number of nodes. Network utility increases in such networks proportional to the number of nodes . If the future is to include space-based network extensions of the Global Information Grid (GIG) beyond simple transponded services, then an increased number of nodes will be highly desirable.
Mission planning segments will need to be expanded to accommodate more payloads and satellites in the distributed world, encompassing core assets, hosted payloads, and smaller free flyers. Coordinated responses will still require a centralized mission planning element, but with potentially many more interfaces. Capabilities may be distributed so that command and control systems will have to be smarter to be aware of the inherent capabilities of each asset relative to the entire system. New or expanded planning tools may also be required. In certain cases, single-mission payloads may be operated independently, though, without substantial coordination with other existing systems.
Additional systems engineering will likely be required to develop the more detailed CONOPS for disaggregated MILSATCOM systems. This includes the work required to allocate mission capabilities among distributed assets, such as core and augmentation, to produce a requirements set for each, and support the tiered architecture concept. While not absolutely required, this process is more likely to ensure higher value to the Government by reducing duplication of capabilities among the various system elements.
Spectrum management is already challenging in many military and commercial frequency bands worldwide. Modern technologies and techniques have continued to allow satellites to be placed ever closer to one another in the GEO belt, but closer frequency coordination continues to be required to limit interference. Disaggregation will continue to complicate spectrum use, with more payloads in operation than ever before. One benefit of disaggregated systems is that they will inherently provide the opportunity for greater frequency reuse due to spatial isolation in the GEO belt allowing greater capacities in smaller theaters. It is also easier to plan and operate single-frequency systems at a single orbital location than multi-band systems.
Full or partial disaggregation and diversification of existing MILSATCOM systems can provide significant benefits including increased cost and schedule certainty, reduced execution risk, improved security, increased resiliency and responsiveness, and the ability to more quickly integrate and demonstrate new technologies to avoid obsolescence. Challenges include providing effective mission planning and system management of a greater number of resources and coordinating their operation globally. While some key missions will always require a certain number of more exquisite satellites, a more diversified, disaggregated mixed architecture that prominently features less complex satellites and hosted payloads appears to have significant benefits to the U.S. Government.
Bernard Fox, Kevin Brancato, Brien Alkire, Report TR-418-AF: Guidelines and Metrics for Assessing Space System Cost Estimates, RAND Corp., 2008; Bertram M. Hadfield, Satellite Systems Cost Estimating,1973
Various public available sources: Lockheed-Martin, TRW/Northrop-Grumman, Hughes Aircraft/Boeing Co.
Government Accountability Office, report GAO-09-648T, Space Acquisitions Government and Industry Partners Face Substantial Challenges in Developing New DOD Space Systems, 2009
Government Accountability Office report, GAO-09-648T, Space Acquisitions Government and Industry Partners Face Substantial Challenges in Developing New DOD Space Systems, 2009
Improving the Cost Estimation of Space Systems: Past Lessons and Future Recommendations, RAND Corporation, 2008
Report of the Defense Science Board/Air Force Scientific Advisory Board Joint Task Force on Acquisition of National Security Space Programs, May 2003
David A. Bearden, Small-Satellite Costs, Aerospace Corporation Crosslink magazine, Winter 2001
NASA Cost Estimating Internet Website: http://cost.jsc.nasa.gov/learn.htm + http://cost.jsc.nasa.gov/learn.html
Edgar Zapata, A Review of Costs of US Evolved Expendable Launch Vehicles (EELV), NASA Internet website, February 5, 2008; http://science.ksc.nasa.gov/shuttle/nexgen/EELV_main.htm + http://science.ksc.nasa.gov/shuttle/nexgen/EELV_main.htm
Federal Aviation Administration, FAA Semi-Annual Launch Report, First Half of 2011; http://www.faa.gov/about/office_org/headquarters_offices/ast/reports_studies/
Federal Aviation Administration, FAA Semi-Annual Launch Report, First Half of 2011; http://www.faa.gov/about/office_org/headquarters_offices/ast/reports_studies/
SpaceX Internet Website: http://www.spacex.com/falcon9.phphttp://www.spacex.com/falcon9.php
Gordon E. Moore, Intel Corp., No Exponential is Forever: But Forever Can Be Delayed!, ISSCC 2003 Session 1 Plenary
George Gilder, Metcalfs Law and Legacy, Forbes ASAP, September 13, 1993
About the author
Mr. Burch is the Director of Advanced MILSATCOM for the Boeing Space & Intelligence Systems (S&IS) Missions and Systems Group in El Segundo, CA, U.S.A. His focus is developing advanced MILSATCOM solutions for U.S. Government and international customers.
Mr. Burch joined Hughes Space & Communications as a Member of the Technical Staff in 1982 as an RF design engineer. Later assignments included subsystem engineering, project engineering, and ultimately project management. During this time he provided technical management for a number of programs in including Milstar II (MDR) and the Interim Polar 1 payload. Boeing acquired Hughes, and from 1999 through 2002 Mr. Burch was the program manager for the Interim Polar EHF program, responsible for the delivery of two EHF communications payloads to the U.S. Navy. Subsequent assignments have included MILSATCOM systems development, as well as acting as both line manager and IPT lead for the development of laser communications (Lasercom) systems for space for the U.S. Air Forces Transformational Satellite (TSAT RRSD) program (2002 2009).
Mr. Burch received his Bachelor of Science degree in Electrical Engineering from California State University, Fresno in 1982 and a Masters Degree in Electrical Engineering with an emphasis in communications science from the California Institute of Technology (Caltech) in 1984.