Highly capable small satellites arecommonplace today, but this was-n’t always the case. It wasn’t untilthe late 1980s that modern small

satellites came on the scene. This newbreed of low-profile, low-cost space systemwas built by maximizing the use of existingcomponents and off-the-shelf technologyand minimizing developmental efforts. Atthe time, many thought that because oftheir functional and operational character-istics and their low acquisition costs, these

small systems would become more preva-lent than the larger systems built during theprevious 30 years.

But exactly which spacecraft fell into thenew category? A precise description ofsmall satellites, or “lightsats,” as they werealso called, was lacking in the space litera-ture of the day. The terms meant differentthings to different people. Some estab-lished a mass threshold (e.g., 500 kilo-grams) to indicate when a satellite wassmall; others used cost as a criterion; still

others used size. Even scarcer than gooddescriptions of small satellites, however,were guidelines for cost estimation of small-satellite projects. Clearly, a more useful def-inition of small space systems was needed.

By the 1990s, because of increased in-terest in small satellites for military, com-mercial, and academic research applica-tions, the Air Force Space and MissileSystems Center (SMC) and the NationalReconnaissance Office (NRO) asked TheAerospace Corporation for information about

Small-Satellite CostsDavid A. Bearden

Crosslink Winter 2000/2001 • 33

34 • Crosslink Winter 2000/2001

capabilities and costs of such systems. In re-sponse, Aerospace commissioned a study tocompare cost and performance characteris-tics of small satellites with those of larger,traditional systems. Of specific interest wasthe ability to examine tradeoffs betweencost and risk to allow assessment of how tra-ditional risk-management philosophiesmight be affected by the adoption of small-satellite designs.

Estimating costs for small systemsraised many questions. What parametersdrove the cost of small satellites? Were tra-ditional parameters known to drive the costof large systems still applicable? How didsmall systems compare with large ones?Did small-satellite acquisition philoso-phies, which prompted reductions in levelsof oversight, independent reviews, and pa-perwork, enable a reduction in cost-per-unitcapability? What advantages might small

satellites offer for rapid incorporation of newtechnologies? Could they help reduce thelong development cycle for military spaceprograms? Were small satellites really eco-nomical for operational applications, such asnavigation and communication?

These questions led to a series of studieson technical and economic issues involvedin designing, manufacturing, and operatingsmall satellites. The studies found that ex-isting spacecraft cost models, developedduring the previous 30 years to support theNational Aeronautics and Space Adminis-tration (NASA) and the Department of De-fense (DOD), were of limited utility becauseof fundamental differences in technicalcharacteristics and acquisition and develop-ment philosophies between small-satelliteand traditional-satellite programs.

This finding prompted NASA and DODto seek cost-analysis methods and models

specifically tailored to small-satellite pro-grams. To meet this need, Aerospace even-tually developed the Small Satellite CostModel, a small-satellite trade-study soft-ware tool that captures cost, performance,and risk information within a single frame-work. Before looking at the developmentof Aerospace’s trade-study tool, though, itwill be valuable to backtrack to the late1980s and review just exactly how small-spacecraft programs had been perceived.

Streamlined Development ActivitiesIn the 1980s, the DOD Advanced ResearchProjects Agency and the United States AirForce Space Test Program served as the pri-mary sources of funding for small satellites,which typically were used for technologyexperiments. The Space Test Program co-ordinated experimental payload flights forthe Army, Navy, Air Force, and other gov-ernment agencies. Reduced developmentcomplexity and required launch-vehiclesize enabled affordable, frequent access tospace for research applications. Relativelylow acquisition costs and short develop-ment schedules also allowed university lab-oratories to participate, providing individ-ual researchers access to space—a privilegepreviously reserved only for well-fundedgovernment organizations.

Small satellites were procured under aspecifically defined “low cost” philosophy.They were smaller in size and were builtwith maximum use of existing hardware. Asmaller business base (i.e., a reduced num-ber of participating contractors) was in-volved in the development process, and itwas perceived that small satellites repre-sented a niche market relative to the moreprevalent large systems. Mission timelinesfrom approval to launch were on the orderof 24 to 48 months, with an on-orbit life of6 to 18 months. Launch costs, either for anexisting dedicated small launcher or for asecondary payload on a large launcher, re-mained high, but developments such as thePegasus air-launched vehicle and newsmall launchers (such as Taurus andAthena) offered promise of lowering thesecosts. Additionally, small-satellite flightand ground systems typically used themost mature hardware and software avail-able to minimize technology-developmentand flight-certification costs.

Emerging advances in microelectronics,software, and lightweight components en-abled system-level downsizing. Spacecraftoften cost more than $200 thousand per

Spa

cecr

aft b

us c

ost

(FY

97 d

olla

rs in

mill

ions

)

1000

100

1

10 DOD large satellitesModern DOD small satellitesTraditional DOD small satellites

0 200 400 600 800 1000 1200 1400 1600Spacecraft bus dry mass (kilograms)

Dollars-per-kilogram comparison of DOD large satellites (500 dollars per kilogram), modern smallsatellites (100 dollars per kilogram), and traditional small satellites (150 dollars per kilogram). Datapoints for these three categories cluster differently, and regression analysis shows that each set ofpoints determines a different cost-estimating relationship.This information confirms the need for a newmodel using contemporary small satellites as its basis.

Tota

l sat

ellit

e co

st(F

Y98

dol

lars

in m

illio

ns)

1000

100

1

10

10000

Satellite launch mass (kilograms)

Traditional NASA missionsNASA faster-better-cheaper missions

0 1000 2000 3000 4000 5000 6000 7000

This graph compares the dollars-per-kilogram ratio for traditional NASA missions (900 dollars per kilo-gram) with the ratio as noted in NASA’s faster-better-cheaper missions (120 dollars per kilogram). It’sclear that the different sets of data points determine markedly different cost-estimating regimes.

Crosslink Winter 2000/2001 • 35

kilogram and could reach $1 million perkilogram with delivery-to-space costs in-cluded. With miniaturization, every kilo-gram saved in the spacecraft bus or instru-ments represented a possible saving of upto five kilograms in launch, onboardpropulsion, and attitude-control systemsmass. Reduced power demands from mi-croelectronics, instruments, and sensorscould produce similar payoffs. For inter-planetary missions, reduced mass had thecapability to produce indirect cost savingsthrough shorter transit times and missionduration. All this downsizing eliminatedthe need for large facilities and costlyequipment such as high bays, clean-roomareas, test facilities, and special handlingequipment and containers.

Engineering development units—proto-types built before the actual construction offlight hardware—were not built; instead aprotoflight approach was favored, where asingle unit served as both the engineeringmodel and the actual flight item. Qualityparts were used where possible, but strictadherence to rigid military specificationswas avoided. Redundancy—the use ofmultiple components for backup in theevent the primary component fails—wasalso avoided in favor of simpler designs.Designers relied on multifunctional sub-systems and software to allow operationalwork-arounds or alternate performancemodes that could provide functionality ifsomething went wrong during a mission.

As a result of these unorthodox ap-proaches that sought ways to save time and

money, small-spacecraft programs came tobe perceived as fast-paced, streamlined de-velopment activities. Dedicated projectleaders with small teams were given fulltechnical and budgetary responsibility withgoals tailored around what could be doneinexpensively on a short schedule. Fixed-price contracts became the norm, and re-quirement changes (and associated budget-ary adjustments) were held to a minimum.

The Next DecadeWith the advent of the 1990s came a move-ment toward realizing routine access to

space. The development of a broad array ofexpendable launch vehicles provided in-creased access to orbit for many differentkinds of payloads. Satellite programs at-tempted to incorporate advanced technol-ogy and demonstrate that fast developmentcycles, low acquisition costs, and smallflexible project teams could produce highlyuseful smaller spacecraft. This differentparadigm opened up new classes of spaceapplications, notably in Earth science,commercial mobile-communications, andremote-sensing arenas.

Cost vs. Pointing Accuracy

y = 1.67 + 12.98 x –0.5

16 data pointsMin = 0.25Standard

Spa

cecr

aft c

ost (

FY

94 d

olla

rs in

mill

ions

)

2.000

8.00

0

32.00

24.00

16.00

Cost vs. Satellite Dry Mass

EOL Power (W)

y = 0.7 + 0.023 x 1.26

20 data pointsMin = 20, Max = 210Standard error = 3.3 $M

40.00

6.00

0

24.00

18.00

12.00

80.0

Spa

cecr

aft c

ost (

FY

94 d

olla

rs in

mill

ions

)

Cost vs. End-of-life Power

EOL power (W)

y = 0.507 + 1.55 x 0.452

14 data pointsMin = 5, Max = 440Standard error = 6.2 $M5.00

0

20.00

15.00

10.00

100.00 200.0 300.0 400.0

Spa

cecr

aft c

ost (

FY

94 d

olla

rs in

mill

ions

)

System-level cost-estimating relationships that were developed for early versions of the Small SatelliteCost Model. The first cost-estimating relationships related total spacecraft bus cost to individual pa-rameters such as mass, power, or pointing accuracy. These were the early predecessors of today’smore sophisticated cost model that represents costs at the subsystem level utilizing a variety of costdrivers.

MACSAT Clementine

Cos

t mod

el e

stim

ate/

actu

al c

ost (

perc

ent)

Microsat LOSAT-X ALEXIS STEP 0 STEP 1 HETE NEARRADCAL SAMPEX

600

800

1000

400

200

0

SatelliteA cost-percentage comparison that makes use of an older model and the up-dated dollars-per-kilogram relationships shown in previous graphs to estimatemodern small-satellite costs. Each bar’s height represents the percentage dif-ference between a satellite’s estimated cost and its actual cost. Thus forClementine, with a percentage of 109%, the older model’s estimate was twice

the actual cost, and for RADCAL, with a percentage of 801%, the oldermodel’s estimated cost was nine times the actual cost. Because the esti-mates far outweighed the real cost in many cases, the chart illustrates theinadequacy of a traditional cost model for modern small satellites.

36 • Crosslink Winter 2000/2001

Small-spacecraft designers, in their questto reduce costs through use of off-the-shelftechnology, in many cases pioneered theuse of microcircuitry and other miniaturizeddevices in space. Whereas small satelliteshad been unstabilized, battery-powered,single-frequency, store-and-forward space-craft with limited applicability to opera-tional space endeavors, the level of func-tionality achievable in small spacecraft tooka dramatic leap forward in the early 1990s,mainly because of the availability of in-creased space-compatible computationalpower and memory. These advances led tothe current rich suite of spacecraft bus capa-bilities and the large array of missions usingsmall spacecraft.

The trend toward cost reduction andstreamlined management continued to gainmomentum with increased interest in smallspacecraft from NASA and DOD. A shiftin philosophy, where a greater tolerance forrisk was assumed, was evident in programslike the NASA-sponsored Small andMedium Explorer Programs, the Ballistic

The Need for a New ModelIt was against this backdrop that Aerospacebegan collecting a large body of informa-tion concerning technologies and program-management techniques that affectedsmall-satellite cost-effectiveness. The pro-grammatic aspects of traditional satelliteprograms (e.g., long schedules, largeamounts of documentation, rigorous proce-dures, and risk abatement) were known todramatically affect cost. In particular, twodistinct but interrelated factors drove thecost of the system: what was built and howit was procured. In many cases, how thesystem was procured appeared to be as im-portant as what was procured because costand schedule were dependent on the acqui-sition environment.

A study that compared spacecraft massversus cost for traditional small spacecraftof the 1960s and 1970s, traditional largespacecraft of the 1970s and 1980s, andmodern (post-1990) small spacecraft re-vealed two important messages. First, themodern small spacecraft differed dramati-cally from traditional large spacecraft aswell as their similarly sized cousins of thepast. It was postulated that the latter differ-ence, as evidenced by cost reduction, wasthe result of a combination of new businessapproaches and advanced technology. Sec-ond, cost and spacecraft sizing modelsbased on systems or technologies for tradi-tional spacecraft were inappropriate for as-sessing modern small satellites.

This was an understandable departurefrom traditional-spacecraft cost trends.New developments in technology are oftenbased on empirical models that character-ize historical trends, with the assumptionthat future missions will to some degree re-flect these trends. However, in cases wheremajor technology advancements are real-ized or where fundamental paradigmsshift, assumptions based on traditional ap-proaches may not apply. It became clearthat estimating small-system costs was onesuch case.

Early small-satellite studies showed thatcost-reduction measures applied to small-satellite programs resulted in system costssubstantially lower than those estimatedby traditional (primarily mass-based)parametric cost-estimating relationships(equations that predict cost as a functionof one or more drivers). The studies ana-lyzed the applicability of available costmodels such as the Air Force UnmannedSpacecraft Cost Model and the Aerospace

5. Develop cost model• Perform parametric weighting• Perform statistical analysis

7. Compare with known costs

8. Apply model• Use in trade study• Use in cost analysis• Consider sensitivities

6. Validate cost model

2. Examine existing cost models• Determine utility• Determine adaptability

1. Define problem

4. Do parametric analysis• Perform regression• Identify correlation• Consider multiple parameters

3. Collect cost and technical data• Identify potential cost drivers• Normalize cost data

9. Deliver model to user community

Mass

$

The cost modeling process. This is an ongoing iterative process that involves collecting data and per-forming regression analysis to arrive at cost-estimating relationships. The data are validated againstactual program costs. The model is delivered to the users for trade analyses.

Missile Defense Organization-sponsoredClementine, DOD-sponsored Space TestExperiment Platforms, and the SmallSatellite Technology Initiative’s Lewis andClark, among others. The end of the ColdWar (in 1991) and the drive toward reduceddevelopment and launch costs created a po-litical and budgetary imperative wheresmall satellites were viewed as one of thefew vehicles available for science and tech-nology missions.

In response to budget pressures and in thewake of several highly publicized lost orimpaired billion-dollar missions, NASA’sadministrator Dan Goldin in 1992 em-braced small spacecraft and promoted thenotion of a “faster-better-cheaper” approachfor many of NASA’s missions. The pro-grams implemented as a result of this tacticdictated faster and cheaper by specifyinglaunch year and imposing a firm fundingcap. These constraints laid the groundworkfor what would become a decade of ongo-ing controversy about the definition andsuccess of faster-better-cheaper.

Crosslink Winter 2000/2001 • 37

100

180014601120440 780

54

42

Beginning-of-Lifepower (watts)

16000

140001200010000

8000600040002000

0

TT

C/C

DH

cos

t(F

Y97

dol

lars

in th

ousa

nds)

Telemetry, Tracking, Communications/Command and Data Handlingsubsystem CER (S-band)

Structures mass (kilograms)

16000

140001200010000

8000600040002000

0S

truc

ture

s co

st(F

Y97

dol

lars

in th

ousa

nds)

Structures subsystem CER (composite)

90725436180

180

60

120

0

Solar-array area(square meters)

Three-dimensional cost-estimating relationships(CER). Later versions of the Small Satellite CostModel used multiparameter cost-estimating relation-ships derived at the subsystem level. Emphasis wasplaced on a combination of mass- and performance-based cost drivers.

Satellite Cost Model to predict costs ofsmall satellites.

These cost models—based on historicalcosts and technical parameters of tradi-tional large satellites developed primarilyfor military space programs—were foundinappropriate for cost analyses of small-satellite programs. It became readily appar-ent in comparing actual costs against costsestimated by these models that a newmodel, dedicated to this new class of mis-sion, was needed. Credible parametric costestimates for small-satellite systems re-quired new cost-estimating relationshipsderived from a cost and technical databaseof modern small satellites.

The Making of a ModelDeveloping a small-satellite cost modelthat related technical parameters and phys-ical characteristics to cost soon became theprimary objective of small-satellite studies.To accomplish this, a broad study of smallsatellites was performed, with emphasis onthe following tasks:• definition of small satellite and identifi-

cation of small-satellite programs

• collection of small-satellite cost andtechnical data from the Air Force,NASA, and university, government lab-oratory, and industry sources

• examination of cost-reduction tech-niques used by small-satellite contrac-tors and sponsors

• performance of parametric analysis todetermine which factors should be usedin the derivation of cost-estimating rela-tionships by using best-fit regressionson data where cost correlation is evident

• development and validation of a costmodel with parametrics and statistics;evaluation of the cost model by per-formance of cost and cost-sensitivityanalyses on small-satellite systems un-der development

• creation of a corporate knowledge baseof ongoing small-satellite activities andcapabilities, technology-insertion oppor-tunities, and project histories for lessonslearned, systems studies, etc.

• maintenance of a corporate presence inthe small-satellite community to advisecustomers about relevant developments

• development of a cadre of people withexpertise and tools for continued studiesof the applicability of small satellites tomilitary, civil, and commercial missions

The cost-modeling process entailed ag-gressive data acquisition through collabo-ration with organizations responsible fordeveloping small satellites. One unantici-pated challenge was actually gainingaccess to cost data. Small-satellite contrac-tors, in their quest to reduce costs, wouldoften not be contractually bound to deliverdetailed cost data, so in many cases costswere not available. Despite this difficulty,Aerospace collected data over a period oftwo to three years for about 20 small-satellite programs at the system level (i.e.,total spacecraft or spacecraft bus costsonly). From this initial database, analystsderived parametric costing relationships as

a function of performance measures andphysical characteristics. The model esti-mated protoflight development costs andcost sensitivities to individual parametersat the system level.

The model was of great value in in-stances where evaluations needed to be per-formed on varying proposals with differingdegrees of detail or when limited informa-tion was available, as is often the case in anearly concept-development phase.

The Second-Generation Cost ModelWhile initial system-level small-satellitestudies were sponsored by DOD and inter-nal Aerospace research and development,in 1995, the need to respond to increas-ingly frequent questions about NASA-sponsored small-satellite architectures anda need for refined small-satellite systemanalysis at the subsystem level promptedNASA to seek better cost-analysis methodsand models specifically tailored to small-satellite programs. Consequently, NASAasked Aerospace to gather informationregarding capabilities and costs of smallsatellites and to develop a set of subsys-tem-level small-satellite cost-estimatingrelationships.

To allow assessment of a completespacecraft bus cost, Aerospace collectedmore data in order to be able to derive cost-estimating relationships for each of thespacecraft bus subsystems:

38 • Crosslink Winter 2000/2001

Program Sponsor SpacecraftLaunch

Launch Launch MissionContractorMass

Date Vehicle(kilograms)

NASA Small Planetary SatellitesClementine BMDO/NASA NRL 494 Jan 94 Titan II Lunar mappingNEAR NASA JHU/APL 805 Feb 96 Delta II Asteroid mappingMGS NASA Lockheed Martin 651 Nov 96 Delta II Mars mappingMars Pathfinder NASA JPL 890 Dec 96 Delta II Mars lander and roverACE NASA JHU/APL 785 Aug 97 Delta II Low energy particleLunar Prospector NASA Lockheed Martin 296 Jan 98 Athena II Lunar scienceDS1 NASA JPL/Spectrum Astro 486 Oct 98 Delta II Technology demoMCO NASA JPL/Lockheed Martin 629 Dec 98 Delta II Mars remote sensingMPL NASA JPL/Lockheed Martin 576 Jan 99 Delta II Mars scienceStardust NASA JPL/Lockheed Martin 385 Feb 99 Delta II Comet sample return

NASA Earth-Orbiting Small SatellitesSAMPEX NASA NASA GSFC 161 Jul 92 Scout Science experimentsMICROLAB NASA/Orbital Orbital 75 Apr 95 Pegasus Lightning experimentMETEOR NASA CTA (Orbital) 364 Oct 95 Conestoga Microgravity experimentsTOMS-EP NASA TRW 295 Jul 96 Pegasus XL Ozone mappingFAST NASA NASA GSFC 191 Aug 96 Pegasus XL Auroral measurementsHETE NASA MIT/AeroAstro 128 Nov 96 Pegasus XL High energy experimentsSAC-B CONAE/NASA CONAE 191 Nov 96 Pegasus XL Science experimentsSeastar NASA Orbital 372 Aug 97 Pegasus XL Ocean colorSNOE NASA GSFC LASP 132 Feb 98 Pegasus XL Space physicsTRACE NASA NASA GSFC 250 Apr 98 Pegasus XL Solar coronalClark NASA CTA (Orbital) 266 cancelled Athena I Science experimentsLewis NASA TRW 385 Jul 98 Athena I Hyperspectral imagingSWAS NASA NASA GSFC 288 Dec 98 Pegasus XL AstronomyWIRE NASA, JPL NASA GSFC 255 Mar 99 Pegasus XL Astronomical telescopeTERRRIERS NASA, BU AeroAstro, LLC 288 May 99 Pegasus XL Space physicsFUSE NASA, APL Orbital 1360 Jun 99 Delta II Space scienceQuikSCAT NASA, NOAA JPL/Ball Aerospace 870 Jun 99 Titan II Ocean wind measureACRIMSat NASA, JPL Orbital 115 Dec 99 Taurus Sun-Earth atmosphereIMAGE NASA GSFC Lockheed Martin 494 Mar 00 Delta II Neutral atom/UV measure

Other U.S.-Built Small SatellitesGLOMR I DARPA DSI (Orbital) 71 Nov 85 Shuttle Message relayPEGSAT DARPA DSI (Orbital) 68 Apr 90 Pegasus Message relayPOGS/SSR STP, ONR DSI (Orbital) 68 Apr 90 Atlas I Geomagnetic surveySCE ONR DSI (Orbital) 56 Apr 90 Atlas I CommunicationsTEX STP, ONR DSI (Orbital) 67 Apr 90 Atlas I CommunicationsMACSAT DARPA DSI (Orbital) 61 May 90 Scout CommunicationsREX STP DSI (Orbital) 77 Jun 91 Scout RadiationLOSAT-X SDIO Ball Aerospace 76 Jul 91 Delta II Sensor experimentsMICROSAT DARPA DSI (Orbital) 26 Nov 91 Pegasus CommunicationsMSTI-1 BMDO JPL/Spectrum Astro 144 Nov 92 Scout Sensor experimentsALEXIS DOE AeroAstro 113 Apr 93 Pegasus X-ray mappingRADCAL STP, NRL DSI (Orbital) 91 Jun 93 Scout Radar calibration testsDARPASAT DARPA/AF Ball Aerospace 161 Mar 94 Taurus ClassifiedSTEP 0 STP TRW 489 Mar 94 Taurus Autonomy experimentsMSTI-2 SDIO Spectrum Astro 170 May 94 Scout Sensor experimentsSTEP 2 STP TRW 180 May 94 Pegasus Signal detect/modulationSTEP 1 STP TRW 352 Jun 94 Pegasus XL Atmospheric physicsAPEX STP Orbital 209 Aug 94 Pegasus Power experimentsSTEP 3 STP TRW 295 Jul 95 Pegasus XL Science/communicationsREX II STP DSI (Orbital) 110 Mar 96 Pegasus XL RadiationMSTI-3 SDIO Spectrum Astro 212 May 96 Pegasus Hyperspectral imagingFORTE DOE LANL/SNL 210 Aug 97 Pegasus XL ScienceSTEP 4 STP TRW 386 Oct 97 Pegasus XL Science/communicationsGFO U.S. Navy Ball Aerospace 357 Feb 98 Taurus Radar altimetryMIGHTYSAT STP CTA (Orbital) 69 Jul 98 STS/GAS ScienceSTEX NRO Lockheed Martin 691 Sep 98 Taurus Tether experimentTSX-5 BMDO/STP Orbital 249 Jun 00 Pegasus XL Remote sensing

Small-Satellite Database

Crosslink Winter 2000/2001 • 39

of the subsystems, using a subset of themore than 70 technical parameters col-lected on each of the small satellites. Theeffort to develop a cost-estimating relation-ship for a small-satellite subsystem tookfull advantage of advanced developmentsin regression techniques. Choosing theproper drivers involved combining aknowledge of statistics, sound engineeringjudgment, and common sense. Graphicssoftware tools assisted in the developmentof these cost-estimating relationships,enabling the analyst to view the shape of afunction against its data points and to iden-tify the function (whether linear, logarith-mic, exponential, or some other form).

The end product was a set of subsystem-level bus-related cost-estimating relation-ships based entirely on actual cost, physi-cal, and performance parameters of 15modern small satellites. This was a majoradvancement over available tools for esti-mating small-satellite costs. Analysts alsodeveloped factors to use in estimating bothrecurring and nonrecurring costs of bussubsystems, to enable studies of multiplebuilds—such as the ones that are neededfor constellations of small satellites. Thecost-estimating relationships enabled theinclusion of cost as a variable in system de-sign tools. They were also incorporatedinto a stand-alone, menu-driven computer-ized model that could be distributed to gov-ernment organizations and private compa-nies that contributed data.

Cost Model Leaves Earth OrbitIn 1996, NASA was moving to smallerplatforms for planetary exploration. Thismovement afforded an important applica-tion for the Small Satellite Cost Model.Following well-publicized problems withthe Galileo and Mars Observer spacecraft,there had emerged in the early 1990s agrowing apprehension in the NASA plane-tary science community that opportunitiesfor planetary science data return weredwindling. After Galileo was launched in1989, the next planetary mission scheduledwas Cassini, which would launch in Octo-ber 1997 and begin returning data in 2003,a full six years after Galileo had stoppedsending data. Since a steady stream of newdata is important to maintaining a vigorousprogram of planetary and scientific investi-gation, the situation was naturally a causefor concern. Out of this concern emerged anew NASA small-spacecraft programcalled Discovery.

• attitude determination and control

• propulsion

• electrical power supply

• telemetry, tracking, and command

• command and data handling

• structure, adapter

• thermal control

Emphasis was placed on obtaining dataon spacecraft bus subsystem characteris-tics. In addition to technical data, costs in

the areas of spacecraft integration, assem-bly and test, program management andsystems engineering, and launch and or-bital operations were requested.

To gather information on the state of theindustry as a whole, as well as specificdata, analysts surveyed and interviewedcontractors who build small satellites orprovide small-satellite facilities (e.g., com-ponents, launchers). A cost and technicalsurvey sheet was distributed to virtuallyevery organization and contractor in thesmall-satellite industry. It was important toobtain information about mass, power, per-formance, and other technical characteris-tics because the development of crediblesubsystem-level cost analyses of small-satellite missions depends on the analyst’s

ability to relate cost to those characteristics.Programs either already completed or await-ing launch in the next year were targeted.

Because Aerospace operates a federallyfunded research and development center, itwas in a unique position to receive propri-etary data from private companies and en-ter it into a special-purpose database tosupport government space-system acquisi-tion goals and provide value added to theindustry as a whole. Proprietary informa-

tion delivered to the corporation wastreated in a restricted manner, used only forthe purpose intended, and not released toorganizations, agencies, or individuals notassociated with the study team. The infor-mation was used exclusively for analysispurposes directly related to cost-model de-velopment. Only derived information de-picted in a generalized manner was re-leased, and the database itself has remainedproprietary. In some cases, formal nondis-closure agreements between the companiesand Aerospace were necessary to facilitatedelivery of proprietary data.

After properly categorizing cost data,adjusting it for inflation, and breaking itout on a subsystem basis, analysts devel-oped cost-estimating relationships for each

Inputs Estimates Print sheet

Database

SSCM98 DP

Cost Input Data

Code* Cost Driver (Units) Value Valid RangeLow High Outside range

Re # Major Contractors 1- Mission Orbit 1

He Apogee (km) 600 400 4,200Tp,Ip Design life (months) 48 6 60

R,L,Pp Satellite Wet Mass (kg) 805. 75 489 64.6%Ap,Ip,Se,Pp Satellite Dry Mass (kg) 468.1 26 410 14.2%

L Stabilization Type 2A-spin Pointing Accuracy (degrees) 0.6 1 5 -40.0%

A-3 axis Pointing Knowledge (degrees) 0.029 0.016 1.5Ap-3 axis Attitude Control Subsystem Mass (kg) 33.9 2.4 59

Pe Number of thrusters 1 0 8Pp Fuel Type 1Pp Propulsion Subsystem Dry Mass (kg) 118. 10.5 118.2Pp Hydrazine propulsion type 2 ERROR: Cold gas cannot be dual!

Ep,Tp Beginning-of-Life Power (W) 1,880 57 1,880Ie End-of-Life Power (W) 105 8 740E Solar Array Cell type 2

Hp Power Subsystem Mass (kg) 103.2 7 124He Solar Array mounting type 1Hp Battery Capacity (A-hrs) 9. 9 40E Battery Type 1S Solar Array Area (m2) 8.9 0.24 11.6T Communications Band 1

Te,Le Downlink Data Rate (kbps) 256. 2.4 3,000Ee Transmitter Power (W) 5. 1 18Hp TT&C/C&DH Subsystem Mass (kg) 43.6 13 115Sp Structures material 2Sp Structures Mass (kg) 102.2 20 183He Thermal Subsystem Mass (kg) 3. 0.1 10.5

Legend required input do not change! do not change!

SSCM98 DP

3-axis Control

Cold Gas

Dual

UHF/VHF

Gallium Arsenide (GaA

NiCd

Estimates Cost-Risk Print sheetSave InputsRetrieve Inputs Database

Single contractor

Composite

Deployable

Earth Orbiting

SMALL SATELLITE COST MODEL (SSCM)

Developed for NASA Headquarters

by D.A. Bearden, N.Y. Lao, T.J. Mosher, and J.J. Muhle

' 1998, The Aerospace Corporation. All Rights Reserved.

This model may not be redistributed without the expressed written consent of The Aerospace Corporation

Start Help Exit

Cost-Probability Distribution Mean Std Error Percentiles of Cost38864 6131

Percentile Cost (FY97$K)

10% 31,401

15% 32,631

20% 33,643

25% 34,536

30% 35,359

35% 36,138

40% 36,894

45% 37,640

50% 38,389

55% 39,153

60% 39,945

65% 40,780

70% 41,679

75% 42,672

80% 43,805

85% 45,163

90% 46,933

95% 49,684

0 10,000 20,000 30,000 40,000 50,000 60,000 70,000

Estimated Cost (FY97$K)

Pro

ba

bil

ity

D

en

sit

y

These selected screen shots from the Small Satellite Cost Model demonstrate the parametric cost-estimating model.The model is easy to work with and provides useful outputs such as cost probabilitydistributions.

participate in a National Research Councilworkshop, from which a report titled “Re-ducing the Costs of Space Science ResearchMissions” was generated. Aerospace wasalso invited to join the editorial board of anew international peer-reviewed technicaljournal (Reducing Space Mission Cost, pub-lished by Kluwer Academic Publishers) andto become a member of the Low Cost Plan-etary Missions Subcommittee of the Inter-national Academy of Astronautics Commit-tee on Small Satellite Missions.

Launched in February 1999, Stardust is journeyingto the comet Wild-2. It will arrive in 2004 and, duringa slow flyby, will collect samples of dust and gas in alow-density material called aerogel. The samples willbe returned to Earth for analysis in 2006. Stardustwill also photograph the comet and do chemicalanalysis of particles and gases. The JPL/LockheedMartin-built spacecraft is one of NASA’s DiscoveryProgram missions. It is the first NASA mission dedi-cated to exploring a comet and the first U.S. missionlaunched to robotically obtain samples in deep space.

NEAR was launched in February 1996. Its objectivewas to orbit the asteroid Eros for one year startingin January 1999, collecting scientific data. Devel-oped in 29 months at JHU/APL, NEAR was part ofthe NASA Discovery Program. Its payload wascomposed of a multispectral imager, a laserrangefinder, an X-ray/gamma-ray spectrometer,and a magnetometer. A software-error-inducedburn abort that occurred in December 1998resulted in delaying the rendezvous and subse-quent data acquisition until February 2000.

The Lunar Prospector, a NASA-sponsored lu-nar polar orbiting probe developed by LockheedMartin, was launched aboard Athena II in Janu-ary 1998. Its primary mission was to map themoon’s chemical, gravitational, and magneticproperties. Data from instruments, including agamma-ray spectrometer, a neutron spectrome-ter, an alpha particle spectrometer, a magne-tometer, and an electron reflectometer, wereused to construct a map of the surface composi-tion of the moon.

Mars Pathfinder, the second launch in the Dis-covery Program developed by JPL, consists of acruise stage, entry vehicle, and lander. The mis-sion of Mars Pathfinder was to test technologiesin preparation for future Mars missions, as wellas to collect data on the Martian atmosphere,meteorology, surface geology, and rock and soilcomposition. On July 4, 1997, Mars Pathfindersuccessfully landed on Mars and subsequentlyrolled out the Sojourner rover to analyze nativerock composition.

Missions Evaluated by the Small Satellite Cost Model

science missions designed to provide fre-quent investigative opportunities to theresearch community. Aerospace served onthe Technical, Management, and Cost re-view panel. Fifty-two Small Explorer mis-sion concepts were evaluated, from whichtwo final missions were chosen.

NASA commended Aerospace for itswork on Discovery and Small Explorer mis-sions. Because of the work it had done onthese programs, Aerospace was invited to

The Discovery program’s primary goalwas to conduct frequent, highly focused,cost-effective missions to answer criticalquestions in solar-system science. For-mally started under NASA’s fiscal-year1994 budget, the Discovery program fea-tured small planetary exploration space-craft—with focused science goals—thatcould be built in 36 months or less andwould cost less than $150 million (fiscalyear 1992), not including the cost of thelaunch vehicle.

To apply its cost model to this new do-main, Aerospace performed, in collabora-tion with Johns Hopkins University’s Ap-plied Physics Laboratory (JHU/APL), acost-risk assessment of the Near Earth As-teroid Rendezvous (NEAR) mission. Thismission, one of NASA’s first two Discoverymissions, was designed to leave Earth orbiton a trajectory to the near-Earth asteroidEros. The study identified a number of lim-itations in applying the Small Satellite CostModel to interplanetary missions. Out ofthis information came a concerted effort togather data on small interplanetary missionsto enhance the model. Analysts collecteddata on missions such as Mars Pathfinder,Lunar Prospector, Clementine, and Star-dust, developing cost-estimating relation-ships appropriate to a Discovery-class mis-sion. Less than a year later the model wasagain applied successfully to the Near EarthAsteroid Rendezvous spacecraft, demon-strating cost estimates within a few percentof the actual costs.

Once Aerospace demonstrated the abil-ity to assess small interplanetary missioncosts, NASA’s Langley Research CenterOffice of Space Science asked the corpora-tion to participate in the Discovery missionevaluation process. Aerospace evaluated 34Discovery proposals submitted by govern-ment, industry, and university teams. Theseproposals included a wide variety of pay-loads (including rovers, probes, and pene-trators)—more than 120 in all. The goalswere to provide independent cost estimatesfor each proposal, identify cost-risk areas,determine cost-risk level (low, medium, orhigh) for each proposal, and evaluate pro-posals in an efficient and equitable manner.Five finalists were selected.

In 1997, as a follow-on to the successfulDiscovery mission evaluation, the NASAOffice of Space Science asked Aerospaceto assist in the selection of Small Explorermissions. This was a series of small, low-cost interplanetary and Earth-orbiting

Illus

trat

ions

cou

rtes

y of

NA

SA

40 • Crosslink Winter 2000/2001

Crosslink Winter 2000/2001 • 41

The ACE (Advanced Composition Explorer)spacecraft carried six high-resolution sensors,mainly spectrometers, and three monitoring instru-ments. It collected samples of low-energy solarand high-energy galactic particles and measuredconditions of solar wind flow and particle events.An Explorer mission sponsored by the NASAOffice of Space Science and built by JHU/APL,ACE orbits the L1 libration point, a location900,000 miles from Earth where the gravitationaleffects of Earth and the sun are balanced, toprovide near-real-time solar wind information.

SWAS (Submillimeter Wave Astronomy Satellite) wasthe third NASA Small Explorer mission. It waslaunched aboard a Pegasus XL rocket in December1998. The overall goal of the mission was to under-stand star formation by using a passively cooledCassegrain telescope to determine the composition ofinterstellar clouds and establish the means by whichthese clouds cool as they collapse to form stars andplanets. SWAS observed water, molecular oxygen,isotopic carbon monoxide, and atomic carbon.

Launched in October 1998, DS1 (Deep Space 1)was a NASA New Millennium Program mission. Itsobjective was to validate several technologies inspace including solar electric propulsion and au-tonomous navigation. Instruments on board includeda solar concentrator array and a miniature integratedcamera and imaging spectrometer. The spacecraft,built by JPL and Spectrum Astro, was designed tomonitor solar wind and measure the interaction withtargets during flybys of an asteroid and a comet.

The primary mission objectives of the Clementine lunarorbiter, launched in January 1994 aboard Titan IIG, wereto investigate long-term effects of the space environmenton sensors and spacecraft components and to take mul-tispectral images of the moon and the near-Earth aster-oid Geographos. The Naval Research Laboratory-builtspacecraft incorporated advanced technologies, includ-ing Lawrence Livermore National Laboratory lightweightsensors. After Clementine completed lunar mapping, itsonboard computer malfunctioned on departure fromlunar orbit and depletion of onboard fuel resulted.

Examining the Faster-Better-Cheaper ExperimentSuccessful NASA programs such as theMars Pathfinder and the Near Earth Aster-oid Rendezvous mission effectively de-bunked the myth that interplanetary mis-sions could only be accomplished withbillion-dollar budgets. They set a new stan-dard against which all later missions werenot only forced to measure up but go be-yond. Designers were asked to meet unre-lenting mission objectives within rigid cost

and schedule constraints in an environmentcharacterized by rapid technological im-provements, immense budgetary pressure,downsizing government, and distributedacquisition authority.

As a result of these constraints, NASAhad greatly increased its utilization ofsmall spacecraft to conduct low-cost scien-tific investigations and technology demon-stration missions. The original tenets of thesmall-satellite paradigm, including lowcost, maximum use of existing components

and off-the-shelf technology, and reducedprogram-management oversight and devel-opmental effort, had been applied to in-creasingly more ambitious endeavors withincreasingly demanding requirements. Thismove had clearly benefited the scientificcommunity by greatly diversifying the num-ber and frequency of science opportunities.

A number of failed small scientificspacecraft, however, such as Small Satel-lite Technology Initiative’s Lewis andClark, and the Wide-field Infrared Experi-ment, fueled an ongoing debate on whetherNASA’s experiment with faster-better-cheaper missions was working. The loss ofthe Mars Climate Orbiter and the Mars Po-lar Lander within a few months of eachother sent waves of anxiety throughoutgovernment and industry that the recipe forsuccessful faster-better-cheaper missionshad been lost. Impaired missions or “nearmisses,” such as the Mars Global Surveyor,contributed to the debate as well, and manywondered whether programs currently onthe books or late in development were tooambitious for the time and money they hadbeen allotted.

At the heart of the matter was allocationof cost and schedule. Priorities hadchanged. During the last few years the tra-ditional approach to spacecraft design,driven by performance characteristics andhigh reliability to meet mission objectives,had completely given way to developmentsdominated by cost- and schedule-relatedconcerns. While it was readily apparentthat the faster-better-cheaper strategy re-sulted in lower costs per mission andshorter absolute development times, thesebenefits may have been achieved at the ex-pense of reduced probability of success.Some questions lingered. When was a mis-sion too fast and too cheap with the resultthat it was prone to failure? Given a fixedamount of time and money, what level ofperformance and technology requirementswould cause a mission to stop short of fail-ure due to unforeseen events?

Risks often do not manifest ahead oftime or in obvious ways. However, whenexamined after the fact, mission failure orimpairment is often found to be the resultof mismanagement or miscommunicationin fatal combination with a series of low-probability events. These missteps, whichoften occur when a program is operatingnear the budget ceiling or under tremen-dous schedule pressure, result in failurescaused by lack of sufficient resources to

42 • Crosslink Winter 2000/2001

To determine whether the faster-better-cheaper experiment was successful, ana-lysts plotted a comparison of complexityrelative to development time and cost, not-ing failures. Some interesting trendsemerged. Correlation between complexity,cost, and schedule was evident. A thresh-old, or “no-fly zone,” was apparent whereproject resources (time, funds) were possi-bly insufficient relative to the complexityof the undertaking. While it is unknownwhether allocation of additional resourceswould have increased the probability ofsuccess of a given mission, this much isclear: When a mission fails or becomes im-paired, it appears that it is too complex rel-ative to its allocated resources.

The observation of a correlation be-tween cost and development time and com-plexity, based on actual program experi-ence (i.e., actual costs incurred anddevelopment time required as opposed tonumbers used during the planning phase),is encouraging because this model can beapplied to future systems. The index may

Large payload mass (~200–500 kilograms)Many (5–10) payload instrumentsThree-axis stabilized using reaction wheelsDeployed sun-tracking solar panels (multijunction cells or concentrator)Long design life (~3–6 years)Partially or fully redundantComposite structuresFine pointing accuracy (~0.01–0.1 degrees)Monopropellant or bipropellant system with thrusters (4–12)High-frequency communicationsDeployed high-gain parabolic antennasHigh data rate downlink (thousands of kilobits per second)High power requirements (~500–2000 watts)Deployed and/or articulated mechanismsSolid-state data recorders (up to 5 gigabytes)Onboard processing (up to 30 million instructions per second)Active thermal control using heat pipes, radiators, etc.

High-Complexity SpacecraftComplexity index 0.67–1

Small payload mass (~5–10 kilograms)One payload instrumentSpin or gravity-gradient stabilizedBody-fixed solar cells (silicon or gallium arsenide)Short design life (~6–12 months)Single-string designAluminum structuresCoarse pointing accuracy (~1–5 degrees)No propulsion or cold-gas systemLow-frequency communicationsSimple helix or patch low-gain antennasLow data rate downlink (~1–10 kilobits per second)Low power requirements (~50–100 watts)No deployed or articulated mechanismsLittle or no data storageNo onboard processing (“bent-pipe”)Passive thermal control using coatings, insulation, etc.

Low-Complexity SpacecraftComplexity index 0–0.33

catastrophic, where the mission was lostcompletely; or programmatic- or launch-related, where the mission was never real-ized because of cancellation or failure dur-ing launch.

A complexity index was derived fromperformance, mass, power, and technologychoices, as a top-level representation of thesystem for purposes of comparison. Com-plexity drivers (a total of 29) included sub-system technical parameters (such as mass,power, performance, pointing accuracy,downlink data rate, technology choices)and a few general programmatic factorssuch as heritage (reuse of a part that flew ona previous mission) and redundancy policy.The process used to estimate spacecraftcomplexity included the following steps.• Identify parameters that drive or signif-

icantly influence spacecraft design.

• Quantify the parameters so that they canbe measured.

• Combine the parameters into an averagecomplexity index (expressed as a valuebetween zero and one).

thoroughly test, simulate, or review workand processes.

Having maintained an extensive histori-cal database of programmatic informationon NASA faster-better-cheaper missions tosupport the Small Satellite Cost Model de-velopment, Aerospace was well positionedto examine the situation. With a decade ofexperience and more than 40 scientific andtechnology demonstration spacecraftflown, sufficient information existed foruse in conducting an objective examina-tion. To understand the relationship be-tween risk, cost, and schedule, Aerospaceanalyzed data for missions launched be-tween 1990 and 2000, using technicalspecifications, costs, development time,and operational status.

The study examined the faster-better-cheaper strategy in terms of complexitymeasured against development time andcost for successful and failed missions. Thefailures were categorized as partial, wherethe mission was impaired in some way;

Crosslink Winter 2000/2001 • 43

reveal a general risk of failure, but it won’tnecessarily specify which subsystem mightfail or how it will fail. Nevertheless, it doesidentify when a new mission under consid-eration is in a regime occupied by failed orsuccessful missions of the recent past. Thisprocess should allow for more informedoverall decisions to be made for new sys-tems being conceived.

ConclusionIn summary, early small-satellite studiesshowed that older cost-estimation modelsbased on historical costs and technical pa-rameters of large satellite systems couldnot be successfully applied to small sys-tems. It was necessary to develop a modelthat would be tailored specifically to thisnew category of spacecraft. To this day,there remains no formally agreed-upondefinition of “small spacecraft,” althoughsuch spacecraft are typically considered tobe Discovery-class in size or less (i.e., forinterplanetary applications, they fit on aDelta II launch vehicle; for Earth-orbitingapplications, they weigh less than 500 kilo-grams), and most are budgeted in the $50-to $250-million range.

Aerospace has been studying smallsatellites since 1991, and the main productof its ongoing work is the Small SatelliteCost Model. Based on actual physical andperformance parameters of small Earth-orbiting and interplanetary spacecraftflown during the last decade, this softwaretool was developed to estimate the cost of asmall spacecraft. It has addressed many ofthe questions that were originally raisedabout cost estimation for small systems.The model is used in assessment of small-satellite conceptual designs, technology

needs, and capabilities, and it is continuallyupdated to model state-of-the-art systems.

The Small Satellite Cost Model has de-veloped through several generations, withadditions to the database and improve-ments to the cost-estimating relationshipsserving as the primary drivers from versionto version. Currently, the small-satellitedatabase has evolved to include more than40 programs. While initial small-satellitestudies were funded by DOD and Aero-space internal funds in the early 1990s, thedevelopment of Small Satellite Cost Modelversion 1.0 was funded by NASA in 1995.The database and model were updated in1996 (version 2.0), and interplanetary ca-pabilities were added in 1998 (version 3.0or Small Satellite Cost Model 98). Thefourth release is now in development.

A version of the model is supplied toindustry members who provide data thatcan improve it. There is also a publiclyavailable version (see www.aero.org/soft-ware/sscm). The model is used extensivelyby DOD, JPL, and virtually all of theNASA field centers. It has become thestandard for parametric evaluation of smallmissions worldwide and is used by the Eu-ropean Space Agency and the French Cen-tre National d’Etudes Spatiales, amongother organizations. A number of foreign-built small spacecraft are included in thedatabase.

The most recent application of thesmall-satellite study is the assessment ofNASA’s approach, under constrainedbudgets and rigid schedules, to conductfaster-better-cheaper scientific investiga-tions. Recent instances of failed or im-paired spacecraft have brought into ques-tion the faster-better-cheaper strategy,

especially for interplanetary applications.While missions have been developed underthis strategy with lower costs and shorterdevelopment times, these benefits have, insome cases, been achieved at the expenseof increasing performance risk. Addressingthe question of when a mission becomes“too fast and too cheap” with the result thatit is prone to failure, studies have foundthat when a mission’s complexity is toogreat relative to its allocated resources, itfails or becomes impaired.

Aerospace’s Small Satellite Cost Modelhas been highly successful, as evidenced byNASA’s recognition for the model’s appli-cation in the Discovery and Small Explorerprograms. A critical component of manysmall-satellite evaluation activities and amajor contribution to the space industry asa whole, the model is a stellar example ofthe value-added role Aerospace can play.There is every reason to expect that moresuccess stories will be forthcoming.

Further ReadingR. L. Abramson and D. A. Bearden, “CostAnalysis Methodology for High-PerformanceSmall Satellites,” SPIE International Sympo-sium on Aerospace and Remote Sensing (Or-lando, FL, April 1993).

R. L. Abramson, D. A. Bearden, and D.Glackin, “Small Satellites: Cost Methodologiesand Remote Sensing Issues” (Paper 2583-62),European Symposium on Satellite Remote Sens-ing (Paris, France, September 25–28, 1995).

H. Apgar, D. A. Bearden, and R. Wong, “CostModeling” chapter in Space Mission Analysisand Design (SMAD), 3rd ed. (Microcosm Press,Torrance, CA, 1999).

D. A. Bearden, “A Complexity-Based Risk As-sessment of Low-Cost Planetary Missions:When Is a Mission Too Fast and Too Cheap?”

Dev

elop

men

t tim

e (m

onth

s)

0.40.2 0.30 0.1 0.60.5 0.7 0.8 0.9 1Complexity index

144

0

120

96

72

48

24

SuccessfulFailedImpaired

200

150

100

50

0

Tota

l spa

cecr

aft c

ost

(FY

00 d

olla

rs in

mill

ions

)

0.40.2 0.30 0.1 0.60.5 0.7 0.8 0.9 1Complexity index

SuccessfulFailedImpaired

Cost and schedule plotted against a complexity index derived from per-formance, mass, power, and technology choices. The regression curvesmay be used to determine the level of complexity possible for a set budgetor development time. Although the complexity index does not identify the

manner or subsystem in which a failure is likely to occur, it does identify aregime by which an index calculated for a mission under consideration maybe compared with missions of the recent past.

44 • Crosslink Winter 2000/2001

T. J. Mosher, N. Y. Lao, E. T. Davalos, and D. A.Bearden, “A Comparison of NEAR ActualSpacecraft Costs with Three Parametric CostModels,” IAA Low Cost Planetary MissionsConference (Pasadena, CA, April 1998).

National Research Council report, Technologyfor Small Spacecraft (1994).

L. Sarsfield, “Federal Investments in SmallSpacecraft,” Report for Office of Science andTechnology Policy, RAND Critical Technolo-gies Institute, DRU-1494-OSTP (September1996).

L. Sarsfield, The Cosmos on a Shoestring:Small Spacecraft for Space and Earth Science(RAND Critical Technologies Institute, ISBN0-8330-2528-7, MR-864-OSTP, 1998).

D. A. Bearden, N.Y. Lao, T. B. Coughlin, A. G.Santo, J. T. Hemmings, and W. L. Ebert, “Com-parison of NEAR Costs with a Small-Space-craft Cost Model,” AIAA/Utah State UniversityConference on Small Satellites (Logan, UT,September, 1996).

E. L. Burgess, N. Y. Lao, and D. A. Bearden,“Small Satellite Cost Estimating Relation-ships,” AIAA/Utah State University Conferenceon Small Satellites (Logan, UT, September18–21, 1995).

M. Kicza and R. Vorder Bruegge, “NASA’s Dis-covery Program,” Acta Astronautica: Proceed-ings of the IAA International Conference onLow-Cost Planetary Missions (April 12–15,1994).

Fourth IAA International Conference on Low-Cost Planetary Missions, JHU/APL (Laurel,MD, May 2–5, 2000).

D. A. Bearden, R. Boudrough, and J. Wertz,“Cost Modeling” chapter in Reducing the Costof Space Systems (Microcosm Press, Torrance,CA, 1998).

D. A. Bearden and R. L. Abramson, “A SmallSatellite Cost Study,” 6th Annual AIAA/UtahState University Conference on Small Satellites(Logan, UT, September 21–24, 1992).

D. A. Bearden and R.L. Abramson, “SmallSatellite Cost Study—Risk and Quality Assess-ment,” CNES/ESA 2nd International Sympo-sium on Small Satellites Systems and Services(Biarritz, France, June 29,1994).

The Soviet Union’s October 1957launch of Sputnik, the first satellite,stunned the world. It kicked off the“space race” between the Soviet Unionand the United States, and in doing so,changed the course of history. Spacewould become an important setting inwhich nations could demonstrate polit-ical and scientific prowess. The UnitedStates responded to Sputnik in Janu-ary 1958, launching Explorer I, a sim-ple, inexpensive spacecraft built to an-swer basic questions about Earth andnear space.

Explorer and its immediate descen-dants were small satellites, but onlybecause of launch-vehicle limitations.Size and complexity of later spacecraftgrew to match launch capability. Notsurprisingly, the early years of thespace race saw U.S. projects expandon many levels. The Cold War spurredthe buildup of a massive space-baseddefense and communications infra-structure in the United States. Thegovernment and its contractors, es-sentially unchecked by budgetary re-strictions, developed large, sophisti-cated, and expensive platforms tomeet increasingly demanding missionrequirements. NASA followed the leadof DOD, building complex scientificand interplanetary spacecraft to maxi-mize research capabilities.

U.S. expertise in space science wasescalating. Launch-vehicle capabilitycontinued to grow from the 1960sthrough the early 1980s, with largesatellite platforms carrying more

powerful payloads (and, often, multiplepayloads). Engineers and scientistsworked to perfect the technologiesnecessary for mission success andlengthier operations. Major researchspacecraft took nearly a decade to de-velop, and they grew to cost more than$1 billion.

As years passed, however, severalfactors pointed to a need to scale back.With the end of the Cold War, govern-ment spending in science and technol-ogy received increased public scrutiny.Funding for large, complex flagshipmissions would no longer be available.Budget constraints forced programmanagers to look seriously at smallerplatforms in an attempt to get payloadsonto less-costly launch vehicles.

At the same time, the public voiceda growing concern over the potentialfor reduced research findings in thewake of several failures of large, high-profile, expensive NASA missions; forexample, a crippling manufacturingdefect was discovered on the HubbleSpace Telescope. NASA came underfire for its perceived inability to deliverquality scientific research.

The scientific community expressedfrustration about the lack of flight op-portunities because only a few flagshipmissions, with decade-long develop-ment times, were being undertaken. Af-ter the limited-capability launch ofGalileo in 1989 and the loss of MarsObserver in 1993, the next planetarymission to be launched was the Cassini

mission to Saturn in 1997, whichwouldn’t start transmitting data to Earthuntil 2003, six years after Galileostopped sending data from Jupiter.

All these issues—budgetarychanges brought about by the end ofthe Cold War, mission failures, pre-dicted gaps in scientific data return—meant that future space-science re-search and planetary explorationwould require a different approach.

In the mid-1980s, with new develop-ments in microelectronics and soft-ware, engineers could package morecapability into smaller satellites. Fund-ing from the DOD Advanced ResearchProjects Agency, the Air Force SpaceTest Program, and university laborato-ries allowed engineers to build low-profile, low-cost satellites with maxi-mum use of existing components andoff-the-shelf technology and minimalnonrecurring developmental effort. Re-search organizations, private busi-nesses, and academic institutions—allweary of waiting years for their instru-ments to be piggybacked on largesatellites—began to develop smallsatellites that could be launched assecondary payloads on the shuttle orlarge expendable boosters.

Small space systems were emerg-ing that were affordable and easy touse, and thus attractive to a larger,more diverse customer base. A newtrend had taken shape, and a wholenew era in the history of space sci-ence was beginning.

The History Behind Small Satellites