NASA Technical Memorandum 87215

Application of Intersatellite Linksto Domestic Satellite Systems

(NASA-TM-87215) APPLICATION OFINTEHSATELLITE LINKS TO DOMESTIC SATELLITESYSTEMS (NASA) 17 p HC A02/MF A01 CSCL 09*

N86-16249

DnclasG3/17 05207

Denise S. Ponchak and Rodney L. SpenceLewis Research CenterCleveland, Ohio

Prepared for thellth Communications Satellite Systems Conferencesponsored by the American Institute of Aeronautics and AstronauticsSan Diego, California, March 16-20, 1986

MASA

https://ntrs.nasa.gov/search.jsp?R=19860006779 2018-06-11T10:40:31+00:00Z

APPLICATION OF INTERSATELLITE LINKS TO DOMESTIC SATELLITE SYSTEMS

Denise S. Ppnchak and Rodney L. SpericeNational Aeronautics and Space Administration

Lewis Research CenterCleveland, Ohio

O!CMCOCM

Abstract

This paper presents the results of a studyon intersatell.it'e link (ISL) applications, fordomestic satellite communications. The objectivewas to. determine if any .technical, economic, orperformance benefits could be gained by intro-ducing' intersatellite Links into a domestic sat-ellite communication network. In this study,several key systems issues of domestic ISL's areaddressed., These include the effect of a skewedtraffic distribution on the selection of ISL sat-ellite orbit locations, tolerable satellite spac-ing, and crosslink traffic-handling requirements.An ISL technology assessment, is made by performinga parametric link analysis for several microwaveand optical implementations. The impact of thecrosslink on the end-to-end link performance isinvestigated for both regenerative and nonregen-erative ISL architectures. Lastly, a comparisonis made between single satellite systems operat-ing at C-', and Ku-bands and the corresponding ISLsystems in terms of ground segment cost, spacesegment cost! 'and net link performance. Resultsindicate.that ISL's can effectively.expand theCONUS ̂ d.rbital arc, with a 60 GHz ISL implementa-tion'being the most attractive.

.... . Introduction

,0ver the.past year, Lewis Research Centerhas addressed the applicability of IntersatelliteLinks (ISL'sj..to U.S. domestic satellite communi-cations. Although there have been many ISL stud-ies, .the'majority of them deal with ISL technologywithout, addressing the costs and benefits of ISLsystems .architectures. In this study we addresseda broad range ,of ISL issues arising from an ini-tial set of "considerations." These considera-tions are briefly examined below:

(1). The use of ISL's permits the placementof spacecraft outside the slot locations whichprovide full CONUS or 48-state coverage. Normallysuch "outside" locations would allow only partialCONUS coverage. However, by linking such space-craft to .another satellite in an appropriateposition,, full CONUS connectivity can be achieved.

.(2):Given that spacecraft location impactsISL..throughput requirements, this report quanti-tatively addresses.the following questions:

, ,(a) What are the required ISL capacitiesas a.function of orbit location?

(b) What is the potential impact ofskewed ISL traffic (west to east versus east to •west) on ISL design? <

. (c) What amount of orbital arc expansionis "realistic".? ,

(3) With the prospect of less than fullCONUS coverage, single beam spacecraft would havehigher gain antennas. This, in turn, would lead

to increased spacecraft receive G/T and thepotential to decrease HPA output power. Thereport quantitatively addresses the communica-tions payload weight and power requirements todetermine if any margin exists for the additionof 'an ISL.

(4) The addition of an ISL to any domesticsystems architecture adds delay and decreases end-to-end link performance. The report addressesthe following questions:

(a) Are there particular classes ofservice which cannot tolerate additional delay?

(b) What are the best ways to compensatefor decreased link performance? Is basebandprocessing (BBP) necessary?

(c) How do' the current ISL technologiesperform against satellite separation and through-put requirements?

(5) Given the above, what are the ISL servicecosts compared to conventional satellite systems?What are the ISL technology, cost and performancerequirements to allow ISL implementation atequivalent service costs?

To address the above considerations, it wasnecessary to establish some study ground rules.First, only CONUS traffic was considered. Sec-ondly, the ISL systems were treated as closedentities competing with conventional U.S. domesticsatellite systems. And finally, competition forslot assignments from international or otherregional or domestic satellite systems wasignored.

The main text of this report is organizedinto the following study areas:

A. General ISL Systems Considerations

B. ISL Technology Assessment

C. Impact of Crosslink on End-to-EndPerformance

D. Spacecraft System Cost Comparison

E. Summary of Conclusions

A. ISL Systems Considerations

In this section, the use of an ISL to alle-viate congestion of prime locations in the geo-stationary orbit is examined assuming a simple2-satellite ISL configuration, shown in Fig. 1.Service is provided to CONUS through the ISLnetwork composed of two single-beam spacecraftsupplying aggregate coverage. This allows place-ment of spacecraft outside the normal visible arc.

Satellite Visibility

The CONUS orbital arc is defined to be theregion of orbital slots from which a satellitecan provide coverage to all locations in CONUS atelevation angles exceeding a specified minimum. -The minimum elevation angle required for eachfrequency band was chosen to limit the amount ofrain attenuation incurred by a signal as it trav-els through the atmosphere. The angles were cho-sen to be 5°, 10°, and 30° for C, Ku, and Ka-bandrespectively. • • . - - . . .

In determining the CONUS orbital arc corres-ponding to each frequency band, two referencepoint locations were'selected to define the •eastern and western bounds of CONUS. They wereAugusta, Maine -(44.25° N/69.750 W) and-Seattle,Washington (47.60° N/123.400 W). Using.theselocations and the appropriate minimum elevationangle criteria, the CONUS-orbital'arcs for C, Ku,and Ka-band were' computed to be:

FrequencyBand

CKu;Ka

Min. Elevation' Angle

5°10°30°

CONUSOrbital Arc

53 to 141° W61 to 134° W97 to 102° W

Traffic Considerations

The distribution of satellite addressabletraffic plays a major role in determining primeorbit locations for ISL satellites. Major metro-politan areas have very large traffic volumescompared to rural areas. This results in a highlyskewed 'traffic distribution for CONUS, with heavytraffic concentration on the east..and west coastswhere large^population densities exist. As asatellite' is moved beyond the limits of the CONUSorbital arc,-the-amount of traffic it can addressdecreases as a result of its diminishing coveragearea. .Using"a 45-node in-house trafficrdistribu-tion model,' the estimate of addressable trafficbased on the satellite longitude was determinedand is shown on Figs. 2, 4, and 6 for C, Ku, andKa-band, respectively. Some of. this traffic will-be intra-regional and remain in the satellite'smain beam, the rest requires an ISL to reach-its-destination. Results of a downlink/crosslinktraffic routing analysis are shown in Figs. 3, 5,and 7 for C, Ku, and Ka-band respectively. .Alsoshown on these graphs is the 50 percent geographiccoverage limit. Within this range of slots, asatellite can provide coverage to at least halfof CONUS.

Examination of Figs. 2 through 7 reveals twoimportant system design considerations. First,for a balanced trafficxload between two spacecraftusing an ISL interconnect, area coverage differsbetween the east and west spacecraft. This, inturn, implies differing spacecraft antenna gainand G/T characteristics. Thus, an "ISL system"may be designed for specific orbit locations andrequire unique'earth terminal design dependingupon which spacecraft is to be accessed. Second,and conversely, forcing equal spacecraft coverageareas to gain advantages in uniform design of .both spacecraft and earth terminals results in anasymmetric ISL crosslink design. By inspectionof Figs. 3, 5, and 7, we conclude that the "west

to-, east" crosslink is-Hkely to require morecapacity than the return link.

. Additionally, crosslink capacity requirements.can be-.approximated using Figs. 3, 5, and 7. Aconventional C-band CONUS coverage spacecraft willuse dual polarization to achieve a gross capacity*of twenty-four 60 Mbps transponders. For easterlyorbit slot assignments, the ISL capacity (Fig. 3)grows slowly until the 50 percent level isreached. Assuming a fully loaded satellite, theapproximate ISL capacity at 14° W is 720 Mbps(twe.lve 60 Mbps transponders). An equivalent-spacecraft-displacement to the west (39°) resultsin an orbit slot, at 180° (east or west) and a, net,crosslink capacity of 85 percent of full capacityor approximately 1.2 Gbps (twenty 60 Mbps trans-ponders). We would-suspect that designs requir-ing the vast majority of traffic to use thecrosslink would likely be rejected by system .planners as inefficient. Thus, it appears thatcrosslinks for C and Ku-band systems would notlikely exceed approximately 1.0 Gbps. By simple"bandwidth ratios, crosslinks at Ka-band are notlikely to exceed 3 to 4 Gbps.

In addition to capacity, satellite separa- .tion is an important ISL design parameter. Wefind from F.ig. 3 .that the maximum orbital separa-tion between two ISL satellites maintaining fullCONUS coverage is 140° for C-band. Figures 5 and7 show .that Ku-band has a maximum separation o'f125° while Ka-band has a 58° separation limit. ,Satellite separations, exceeding these -values will,not permit full CONUS coverage between the two. .spacecraft.

Based on the above discussions if we restrictISL capacity to no more than 70 percent of totalspacecraft capacity, since it is unlikely .thatthis value would be exceeded in a real system,then the potential increase in orbital arc for C,Ku and Ka-band, respectively, are 71 percent(slots from 8 to -162° W), 97 percent (slots from14 to 158° W), and 1433 percent,-(slots/rom 39 to130° W). These figures were calculated assumingthat the orbital-spacing for C and Ku-band sat7,ellites would remain at 2° while the Ka-band sat-ellites would be located at 1° intervals. Weconclude that ISL's enable dramatic .increases inorbital arc utilization for the higher^frequency,bands.

ISL Delay ;Impact

The traffic distribution model used up tothis point represents total long-haul communica-tions traffic that is addressable by satellite.The total traffic can be divided into three ser-vice categories. The first and largest serviceis for voice communications, accounting forapproximately 77 percent of the total traffic.The second is data services which accounts for 8percent of the total. The remaining 15 percentof the total traffic is comprised of-videoservices..

In a system that employs intersatellitelinks, the transmitted signal will.be subjectedto an additional time delay proportional to thesatellite spacing. The CCITT recommends as anetwork performance objective that the mean one-way transmission delay not exceed 400 msec forvoice circuits. This is provided that echo

suppressors and cancellers are used. Of this400 msec, approximately 290 msec is used by theup and downlinks and the terrestrial extensions.This leaves 110 msec delay for the ISL which willallow a maximum satellite separation of 50°,which for C and Ku-band prohibits the placementof both ISL satellites outside of the CONUSorbital arc.

Data systems normally employ a form of auto-matic repeat request error correction (ARQ),either stop and wait ARQ or continuous ARQ, theformer being most commonly used. After a blockof data is sent, the transmitter must pause andwait for an acknowledgment signal before any fur-ther transmissions occur. If the acknowledgmentis positive (no errors), the next block of datais sent; if it is negative (errors), the sameblock of data is retransmitted. When the pathlengths, and subsequently path delays, increase,the link throughput is decreased thus loweringtransmission efficiency. Continuous ARQ is moreimmune to transmission delays. With this proto-col, the transmitter continually sends data blocksand receives back a steady stream of acknowledg-ment signals. The system requires sufficientstorage to hold the data until an acknowledgmentis received in case the data must be retransmit-ted. The amount of storage required depends onthe bit-rate and the length of the round-tripdelay. The incremental amount of buffering neededwith an ISL satellite system over conventionalsatellite systems would be proportional to theincrease in the transmission delay.

The additional propagation delay of theintersatellite link can severely limit the amountof voice and data traffic that the system cancapture. Voice traffic requirements restrain thesatellite separation to 50°. The length of thedelay increases the amount of storage needed fordata transmission, or lowers the efficiency.Propagation delay has no significant effect ontelevision transmission and other one-wayservices.

B. ISL Technology Assessment

In order to assess the suitability of milli-meter wave (MMW) and optical systems for use inintersatellite links, a parametric analysisinvolving the basic system variables (antenna oroptic size, transmit power, data rate, etc.), wasperformed under certain system assumptions. Thena comparison of the various ISL types was made ona weight, prime power, and performance basis.

Selection of RF and Laser Systems

For MMW systems, crosslink frequencies inthe 23, 32, and 60 GHz bands were studied forwhich the allocated bandwidths are 1, 1, and5 GHz, respectively. Higher microwave frequen-cies were not considered because of the lack ofcomponent technology in these bands. Among theavailable laser sources are the FD Nd:YAG, HeNe,GaAs, and C02 lasers. Characteristics of theseprincipal laser sources are shown in Table 1.The FD Nd:YAG diode pumped laser and the GaAssemiconductor laser were chosen to be best suitedfor ISL application primarily because theseapproaches are amenable to simple and efficientdirect-detection techniques. Thus, the receiveris essentially a power detector or "photon bucket"

and the phase of the received signal is unimpor-tant. This is in contrast to the C02 gas lasersystem which requires heterodyne detection andhence highly stable transmitters, an additionallaser at the receiver for the local oscillator,as well as precise tracking of the received opti-cal wavefront and polarization. Further, theC02 laser requires cryogenically cooled detec-tors to suppress internal noise and its presentlifetime is not sufficient for space qualifica-tion. Other reasons why the YAG diode-pumpedlaser and GaAs laser are the preferred opticalISL system approaches are long lifetime and reli-ability, production of modest output power, andavailability of wide-band traveling wave electro-optic modulators and high-gain low-noise photo-detectors. It should also be mentioned that theother common laser source, the HeNe laser, whilebeing the most mature of laser sources with avery high reliability, is severely limited inoutput power and efficiency; therefore the HeNelaser should probably be considered only for veryshort-range or intra-cluster ISL applications.

System Parameters and Requirements

A summary of some of the basic ISL systemparameters and requirements used in the analysisis given below. Values of transmit power repre-sent current or near-term state of the art.

(1) Frequency: 23, 32, 60 GHz and FD Nd:YAG(0.53 tin), GaAs (0.83 jn)

(2) Satellite spacings: 30° to 140°(maximum separation for line-of-sight viewis 163°)

(3) BER: TO"7

(4) Data rate: 1 Gbps crosslink

(5) Modulation: uncoded QPSK - MMW systems;continuous optical carrier intensitymodulation, direct detection - lasers

(6) Antenna/optics diameters: 3, 4, and5 ft parabolic reflectors - MMW; 6, 12,and 24 in. optics apertures - laser (samesize antennas assumed on both ends ofcrosslink)

(7) Maximum transmit power: 75 W - MMW,100 mW - GaAs, 500 mW - FD Nd:YAG

(8) Noise source: 1500 K receiver noisetemp - MMW; 4xlO'10 W background solarradiation noise - laser (sun modeled as ablackbody radiator at T = 6000°K)

ISL Systems

In order to make a comparison of the dif-ferent ISL technologies in terms of performancecapability, prime power required, and weight bur-den on the satellite, the following approach wasused: (1) curves relating the required crosslinkEIRP for a nominal crosslink bit-error-rate (BER)of 10'' were generated as .a function of satel-lite longitudinal separation for parametricvalues of data rate and antenna/optics diameterfor each of the ISL types, (2) average RF andlaser output power values necessary to producethe above EIRP's were determined, (3) weight and

power models (MMW model was developed in-house,while the laser model was developed by COMSAT)using the above parameters as inputs were appliedto the MMW and optical systems to obtain esti-mates of the ISL subsystem weight and prime power.From these data, curves relating ISL weight andprime power to transmitting EIRP were generated,and (4) a comparison was made among the variousISL types,'cognizant of the state of art (SOA)technological constraints applying to each, todetermine which approach offered the least weightand power'burden on the spacecraft for the givendata rate crosslink operating over a range ofsatellite angular spacings.

The results of the analysis procedure areshown in Table 2, listing each of the ISL systemsalong with their corresponding variation inrequired EIRP, transmit RF or laser power, primepower, and subsystem weight as the satelliteseparation is increased from 30° to 140°. Inexamining the table, the following comments apply:

(1) Among lasers, systems using smallerdiameter optics weigh less than ones using largerdiameters since most of the ISL mass is concen-trated in the acquisition and tracking subsystem(40 to 70 percent of total weight) where sub-microradian tracking and pointing accuracy isrequired. As an example of the extremely narrowoptical beams and formidable pointing problemsinvolved, an ISL system using a FD Nd:YAG laserwith 6 in. optics will transmit a beamwidth ofabout 4.2 yrad. Therefore, to constrain trans-mit and receive pointing losses to less than3 dB, a pointing accuracy of 2.1 prad is neces-sary, roughly the angle subtended by a dime froma distance of 5 miles. The majority of primepower is also absorbed in the acquisition/trackingand optical modulation subsystems rather than asinput to the laser itself. Thus, at a given datarate and optics diameter, ISL subsystem weightand prime power are relatively insensitive tolaser output power and angular separation.

(2) Unlike the optical systems, MMW ISL'sexhibit a high dependency of prime power andweight on EIRP and longitudinal separation. Thisis because TWTA output power is the major factordriving the prime power requirement and resultanttransponder and power subsystem weights. There-fore, in contrast to optical systems, it is moreadvantageous to use larger diameter antennas andreduce the rf output power requirement to aminimum.

(3) The 60 GHz ISL implementation using 5 ftreflectors offers significant weight and poweradvantages over the other methods. From Table 2,this package weighing about 35 Kg (77 Ib) andconsuming 70 W of prime power can sustain a 1 Gbpsdata rate link at a BER of 10"7 or better outto 140° separation. This implementation is 15 to25 Kg .lighter than the 6 in. laser systems. Itshould be noted, however, that COMSAT character-ized their optical ISL weight and power model asbeing very conservative.

A closer look at these implementations isshown in Fig. 8. Here the data rate has'beenextended out to 4 Gbps and ISL package weight isplotted as a function of satellite separation fora crosslink BER of 10"'. We see that for thesame data rate, 60 GHz systems are always lighter

than laser systems. Laser ISL's, however, exhibitapparent advantages over RF systems in certainoperating regimes. For data rates greater than 1Gbps, the laser systems can more easily achievehigh Eb/No.within SOA power levels. Note alsothat lasers have an important advantage over RFsystems at the higher data rates in that theirweight is for the most part distance insensitive.Therefore for a given data rate and performancerequirement, an ISL laser package will weighapproximately the same regardless of crosslinkseparation.

In contrast, RF ISL subsystem weight ishighly dependent on satellite spacing at highdata rates (especially 4 Gbps and higher) so thata major driver of the ISL package design would bethe projected ISL separation distance. As datarates decrease, however, 60 GHz ISL weight becomesmore distance insensitive and the technologyoffers reduced weight and power penalty on thesatellite, near-term availability of components,and simplicity of design. Also, in comparison toother MMW systems, 60 GHz offers the advantage ofreduced antenna size and TWT power, and largeravailable bandwidth as well as avoiding thepotential RFI problem that the 23 and 32 GHzbands may encounter.

From the models, laser and RF technologyneeds can be identified. For RF systems, higherpower, higher efficiency transmitters are neces-sary for large separation, high data rate appli-cations. For laser systems, reduced massacquisition and tracking systems and reducedpower consumption for the laser modulator systemare necessary for similar applications.

C. Impact of Crosslink on End-to-EndLink Performance

With the insertion of a crosslink, into a2-satellite ISL configuration, overall link per-formance is degraded, with the weakest of thethree segments (uplink, crosslink, or downlink)determining end-to-end BER, assuming conventionaltranslating repeaters. Reasonable BER objectivesare dependent on the type of service beingoffered; generally BER's in the range of 10"^to lO"^ are considered acceptable for digitizedvoice data while BER's in the range of 10"° to10"' are considered satisfactory for digitalvideo and data signals.

In our study of domestic ISL's, the impactof crosslink insertion was measured by comparingthe BER performance of a typical C-band andKu-band single satellite system, providingsingle-beam CONUS coverage, to that of their ISLcounterparts in which each satellite of the2-satellite ISL system provides single beamcoverage to half the CONUS area. In making thiscomparison it was noted that: (1) crosslinkdegradation is partially offset by an improveduplink G/T resulting from the use of a largerantenna to provide the reduced amount of satel-lite coverage, and (2) the larger antenna allowsthe reduction of downlink HPA power while main-taining the same net EIRP. This combinationtranslates into a net weight savings which canthen be reallocated to the ISL subsystem.

It should be emphasized that the crosslinkdegradation results presented in this section are

entirely specific to the-link budget characteris-tics that were chosen for these systems. Ulti-mately, the effect on overall performance ofintroducing the ISL will depend on the relativevalues of uplink, crosslink, and downlink Eb/No,the coding technique used, as well as whether IFor baseband processing is employed on the satel-lite. One can imagine an ISL scenario in whichthe crosslink Eb/No is significantly greater thanthe uplink or downlink Eb/No so that, performance-wise, it is essentially "transparent" to thesystem except for the added propagation delay.However, such large crosslink Eb/No requirementslead to large output power levels. To illustratethis, Fig. 9 is a plot of the received crosslinkEb/No a'rid the corresponding BER as a function ofsatellite spacing for a 60 GHz, 1 Gbps link using5-ft transmit and receive reflectors and a laserlink using 6 in. optics. The separate curvescorrespond to different values of RF and laseroutput power.

From the-figure, we see that more than 130 Wof RF power at 60 GHz is necessary to maintain acrosslink Eb/No of at least 20 dB out to 140°separation. This is currently beyond the capa-bility of conventional helix-type TWT's. For thelaser, the required power level is about 400 mW,which 'is within the SOA. By going to 12 in.optics, this power is reduced to an easilyachieved value of 25 mW. Thus, in designingtransparent ISL's for wide angular spacings,optical links appear more suitable than MMW giventhe current state of technology.

It should also be mentioned that if thecrosslink was effectively made independent fromthe uplink and downlink through the use of IF orBBP at both its transmitting and receiving ends,then "transparency" can be much more easilyachieved since the overall BER is now the sum ofthe individual link segment BER's. For example,from Fig. 9, 10"10 BER can be obtained on thecrosslink, at 140° spacing, using approximately50 W and 100 mW for the 60 GHz and laser systems,respectively.

Representative link budgets for the C-bandand Ku-band single-beam CONUS coverage systemsare shown in Tables 3 and 4. Major system param-eters are typical of those from the Ku-band SBSsystem and C-band RCA Satcom system. Slightdifferences exist, however, due to the elliptic-i-ty of the spacecraft antenna. A comparison ofthe end-to-end link performance between thesingle satellite baseline and ISL configurationsis shown in Table 5 for both C-band and Ku-band.Indicated on the table is the net Eb/No valuesand the BER's they produce for coding/noncodingmodes of operation. Also shown for the ISL casesis the BER performance of the system with IFdemodulation/ remodulation on the satellites atthe point where the crosslink interfaces with thedownlink. In this type of processing, thereceived crosslink signal is frequency translatedto IF and the entire IF signal is remodulatedonto the downlink carrier. By this method, theaccumulated noise on the uplink and crosslink isisolated from the downlink, which is normally theweakest link in the chain. Overall BER is thenthe sum of the BER's on each of the two isolatedsegments (uplink/crosslink + downlink). Theobvious penalty in this approach is the necessityfor additional downlink bandwidth when compared

to uplink bandwidth. For this reason, such a .system is unlikely to find application at C orKu-band, but may offer some appeal for certainapplications at Ka-band where a large bandwidthallocation exists.

By examination of Table 5, the followingobservations can be made:

(1) If the single satellite system isdesigned for a BER performance acceptable forvoice, then it appears that the insertion of thecrosslink has no appreciable impact in moving theBER performance out of the acceptable range.Related to this point is the fact that propaga-tion delay factors for voice will generally con-strain ISL spacings to less than about 50° inwhich case present or near-term technology per-mits a much higher quality crosslink (Fig. 9}than the 11.3 dB Eb/No (10-' BER) link assumedhere.

(2) If the single satellite system isdesigned for a BER performance acceptable fordata and video traffic (10~^ to 10"' rangerequires R = 7/8 code for C-band, R = 3/4 codefor Ku-band), then the resulting BER's aftercrosslink insertion are still satisfactory.

(3) For the satellite configurations studiedhere, the use of demodulating/ remodulatingtransponders for the ISL is unnecessary unlessone desires an improvement in performance overthe single satellite system.

As mentioned before, the selection ofuplink-crosslink-downlink design parameters,coding scheme, modulation format, etc., are madeto conform to a particular system application andits requirements. The primary purpose for ana-lyzing these particular configurations was todemonstrate quantitatively what the impact of anISL can be, how it relates to the present SOA,and where compromises can occur.

D. Spacecraft Weight and Cost Estimates

The next stage in studying the feasibilityof ISL systems was to determine if any payloadweight margins would result from the trade-off ofreduced downlink HPA power for increased antennagain, which could accommodate an ISL packagewithout increasing the overall satellite weight.The satellite payload was defined to consist ofthe transponder subsystem, the antenna subsystem,and the portion of the power subsystem dedicatedto the transponders. The ISL package was treatedas a separate entity from the communicationspayload though it is recognized that the two willbe interconnected.

Using in-house payload weight algorithms,the communications payload weight of the ISLsatellite was calculated and compared to that ofa baseline CONUS satellite having equal downlinkperformance. The design of the payloads wereidentical except for the transmitting power andthe antenna size (obtained from Tables 3 and 4).The increased antenna size of the ISL satelliteproduces a weight penalty, whereas the reducedtransmit power lowers the weight and operatingpower requirements. A payload comparison forboth C and Ku-band is shown in Table 6. As canbe seen from these results, ISL weight margins

were achieved for both frequency bands. Referringto Fig. 8, we see that a 75 1b ISL weight marginfor a C-band ISL satellite corresponds to theestimated weight of a 60 GHz package consistingof a 5-ft antenna and a 25-W TWTA. Implementingthis package has the capability of offering a1 Gbps capacity crosslink at TO"7 BER forseparations as large as 140°. For medium-rangeseparations compatible with voice transmission(40° to 60°), this package would provide veryhigh quality crosslinks (Eb/No in the 17-20 dBrange) with BER's better than 10"10. This mar-gin would have to be slightly greater in order toaccommodate a laser package (weighing approxi-mately 85 Ib) although the laser ISL weight modelis considered conservative. Also, from Fig. 8 weobserve that a 279 Ib ISL weight margin for aKu-band ISL satellite will easily allow implemen-tation of either a laser or microwave ISL packageproviding crosslink capacities in excess of4 Gbps over all angular separations.

The final stage was to determine if any eco-nomic benefits could be gained in using ISL's fordomestic satellite communications. The approachthat was taken was to compare two ISL satellitescenarios to a baseline system. The first ISLscenario was one in which full advantage of theincreased antenna gain was taken into accounton-board the satellite to provide a weight/costmargin for the ISL package. For this scenario,it was assumed the earth stations would have thesame design as the baseline scenario. The second -ISL scenario used the increased antenna gain toprovide a cost-reduction of the earth stationsthrough reduced earth station antenna gain andtransmitting power. This cost savings could thenbe used to provide for the larger satelliteantenna-and ISL package. The baseline scenarioconsisted of two CONUS satellites so that eachscenario provided the same capacity.

The spacecraft cost model (launch costincluded) used for this study was an in-housemodel based on the satellite beginning-of-life(BOL) weight and end-of-life (EOL) power whichwere determined by a statistical approach basedon the payload weight and power requirements.

-The earth terminals for the various scenarioswere compared by the cost of their RF equipment(antenna, HPA, and LNA). Once again, in-housemodels were utilized to determine the costs basedon the following RF system parameters: antennadiameter', HPA RF power, and system noise tempera-ture. Given the required earth station EIRP andG/T (obtained from Tables 3 and 4), these threeparameters were determined so as to minimize theRF equipment cost. The earth station character-istics are given in Table 7. All cost values aregiven in 1984 dollars.

The results given in Table 8 indicate thatthe first ISL scenario in which the increasedantenna gain is advantageously used on-board thespacecraft provides a direct cost margin thatcould be applied toward the two required ISLpackages. The concept of the second ISL scenariodoes not provide such a direct margin. For thiscase, many earth terminals would be needed tosupport the larger satellite antennas and ISLpackages. Even though the exact cost of an ISLpackage is unknown, it can be seen that thisscenario would not be comparable to the baseline

scenario unless an unrealistically high number ofearth terminals were required.

E. Conclusion

ISL's have the capability of relievingorbital arc congestion by allowing satellites tobe placed outside the normal visible arc withreduced coverage area requirements per space-craft. For moderately high data rates (up toapproximately 2 Gbps), a 60 GHz MMW implementa-tion seems to currently be the most attractive ofthe various ISL approaches. Compared to lasers,60 GHz has a more mature technology base, issimpler in design, has reduced weight and powerpenalty on the satellite, and avoids the acquisi-tion and tracking problems associated with micro-radian beamwidth optical systems. However, formulti-gigabit rate, long-range applicationsexceeding the current capability of MMW SOA,laser systems seem to be more appropriate. Incomparison with the 23 GHz and 32 GHz implementa-tions, 60 GHz has the advantage of reduced TWTApower, reasonable antenna size, larger availablebandwidth, and immunity to RFI. A 60 GHz ISLusing a modest 25 W of RF power and 5 ft reflec-tors is capable of supporting a 1 Gbps crosslinkat 10~7 BER for a 140° satellite separation.This maximum separation corresponds to each sat-ellite covering half the CONUS area at 5° eleva-tion angle minimum for C-band. At separationsless than 100°, BER performance exceeds 10"10so that for most ISL applications (especiallyvoice service), crosslink degradation on end-to-end performance is negligible.

For the C and Ku-band single beam casesexamined, it has been demonstrated that improve-ments in uplink and downlink performance can betranslated into weight and power margins to com-pensate for ISL weight and power penalties. Inthis way, a two-satellite system connected by anISL provides a nearly equivalent or perhaps some-what superior capability over a simple unconnectedtwo-satellite single-beam system (consider the ;increased connectivity from a single earth sta-tion). However, such advantages are not nearlyas evident for multibeam satellite systems. Forthese satellites, there is no inherent advantagein less than CONUS coverage. Size, weight, andpower savings are the direct result of decreasedcoverage and the resultant decrease in traffic.In order to maintain the same net traffic as thesingle CONUS multibeam satellite, two smallermultibeam satellites are required, each equippedwith an ISL package. Thus, a net increase inservice costs is expected.

In improving uplink and downlink performanceover single satellite systems, ISL's have theirgreatest advantage at Ka-band where the abilityto place satellites over locations experiencinghigh rainfall (normally outside the Ka-band CONUSarc) allows a 2 to 4 dB reduction in rain fade.As far as economic advantages are concerned, arealistic comparison involves making assumptionson the ISL and single satellite systems beingstudied. In the satellite configurations con-sidered in this report, ISL system costs were notderived as such. Rather, a cost comparison wasmade between two CONUS satellites and two satel-lites operating with an ISL (equal capacities) todetermine the cost margin that can be allocated

to the ISL system. For the C and Ku-band systems,these margins were $23 and $82 M, respectively.

References

1. Gagliardi, R.M., Satellite Communications,Lifetime Learning Publications, 1984.

2. Sivo, O.N., "Advanced Communication SatelliteSystems," Microwave Journal; Vol. 27, No. 1,Jan., 1984, pp. 77-95->(ll pp.).

3. Stevenson, S., Poley, W., Lekan, J., andSalzman, J.A., "Demand for Satellite-ProvidedDomestic Communications Services to the Year2000," NASA TM-86894, 1984.

4. "The Effects of Transmission Delay in theFixed Satellite Service," Report 383-4,Recommendations and Report of the CCIR 1982,Vol. IV, Part I - Fixed Satellite Service,International Telecommunication Union, Geneva,Switzerland, 1982, pp. 29-36.

5. Welti, G.R. and Lee, Y.S., "PlanningAssistance for the 30/20 GHz Program, Task 6Final Report, Study of Intersatellite Links,"COMSAT TASK 123-6031, COMSAT Laboratories,Clarksburg, MD, Nov. 1981. (Also NASA Report1-11-C-1-T6)

ORIGINAL PAGE ISOF POOR QUALITY

ORIGINAL PA££ 58

TABLE 1. - CHARACTERISTICS OF LASER SYSTEMS

Laser

FD Nd:YAGHeNeGaAsC02

Wavelength,ym

0.530.630.8310.6

Average poweroutput, mW

TOO to 5002 to 5

' 20 to 1001000 to 2000

Transmitterefficiency

0.5 to 2 percent1 percent

5 to 10 percent10 to 15 percent

Lifetime,hr "'•'_

50 00075 00050 000

5 000 to 10 000

TABLE 2. - PRIME POWER REQUIREMENT AND WEIGHT RANGES OF ISL SYSTEMS FOR

30° TO 140° SPACINGS (1 Gbps LINK)

System

Laser (FD Nd:YAG)

Laser (GaAs)

23 GHz

32 GHz

60 GHz

Diameter

6 inch12 inch24 inch

6 inch12 inch24 inch

3 ft4 ft5 ft

3 ft4 ft5 ft

3 ft4 ft5 ft

RequiredEIRP (dBW),10-' BER

94 to 10588 to 9982 to 93

94 to 10588 to 9982 to 93

66 to 7763 to 7461 to 72

66 to 7763 to 7461 to 72

66 to 7763 to 7461 to 72

Transmit power

4 to 55 mW0.3 to 3.5 mW.02 to .22 mW

11 to 136 mW.7 to 8.5 mW.04 to .5 mW

117 to 1476 W33 to 416 W13 to 168 W

61 to 763 W17 to 215 W7 to 87 W

17 to 217 W5 to 61 W2 to 25 W

Prime power,W

80 to 9585 to 86

90

80 to 9085 to 86

90 .

300 to 370090 to 100040 to 400

150 to 190042 to 54023 to 220

50 to 560IB to 16010 to 70

Weight,kg

47 to 5068180

47 to 5068

.. 180

76 to 66035 to 22028 to 100

48 to 35027 to 12024 to 55

25 to 12020 to 5022 to 35

TABLE 3. - C-BAND SINGLE BEAM CONUS COVERAGE REPRESENTATIVE LINK BUDGET

Link parameter

PT (Transmit power)Ltf (Feed loss)Gn (On-axis gain)EIRP (On-axis)Ltp (Transmit pointing loss)LA (Atmospheric loss)Lfs (Freespace loss)Lrp (Receive pointing loss)Edge of coverage lossGR (Receive on-axis gain)TS (System noise temp)G/TS (Receive figure of merit)K (Boltzmann's Constant)Rb (Bit rate, Mbps)Implementation lossRain loss (99.99 percent)Received Eb/No

Uplink, dB(fup = 6 GHz)

27.0 (500 W)-1.0

53.7 (32.8 ft)79.7

-0.24 (0.05° Error)-0.5 (5° Elevation)

-200.3 (5° Elevation)-.005 (.07° Error)

-3.034.2 (5.4 by 2.3 ft)

+30.0 (1000°K)4.3

-228.6+77.8 (60 Mbps)

-1.5-3.0

26.3 dB

Downlink, dB(fdown = 4 GHz)

7.0 (5 W)-1.0

30.8 (5.4 by 2.3 ft)35.8

-0.002 (0.07° Error)-0.44 (5° Elevation)-196.8 (5° Elevation)

-.11 (.05° Error)-3.0

50.2 (32.8 ft)+24.8 (300°K)

25.5-228.6

+77.8 (60 Hbps)-1.5-1.0

10.3 dB

TABLE 4. - Ku-BAND SINGLE BEAM CONUS COVERAGE REPRESENTATIVE LINK BUDGET

Link parameter

Pj (Transmit power)Ltf (Feed loss)6n (On-As1x gain)EIRP (On-axis)Ltp (Transmit pointing loss)LA (Atmospheric loss)Lfs (Freespace loss)Lrp (Receive pointing loss)Edge of coverage lossGp (Receive on-axis gain)TS (System noise temp)G/TS (Receive figure of merit)K (Boltzmann's Constant)RD (Bit rate, Mbps)Implementation lossRain loss (99.99 percent)Received Eb/No

Uplink, dB(fup = 14 GHz)

23.0 (200 W)-1.0

55.1 (16.4 ft)77.1

-0.33 (0.05° Error)-0.5 (10° Elevation)

-207.5 (10° Elevation)-0.0-3.0

32.0 (1.8 by 0.7 ft)+30.0 (1000K)

2.0-228.6

+79.5 (90 Mbps)-1 .5-4.0

11.3 dB

Downlink, dB(fdown = 12 GH*>

13.0 (20 W)-1.0

30.6 (1.8 by 0.7 ft)41.7

-0.003 (0.07° Error)-0.35 (10° Elevation)-206.2 (10° Elevation)

-0.24-3.0

53.7 (16.4 ft)+24.8 (300K)

29.0-228.6

+79.5 (90 Mbps)-1.5-1.5

8.0 dB

TABLE 5. - COMPARISON OF C-BAND/Ku-BAND SINGLE SATELLITE BASELINE AND ISL

END-TO-END LINK PERFORMANCE

Net Eb/No (dB)BER (OPSK uncoded)BER (R = 7/8 code)BER (R = 3/4 code)

C-Bandsingle sat.baseline

10.23x1 O-6<io-io<io->o

ISL system(C-Band

up/down links)

a7.73x1 O-4

lO-'10-10

Ku-Bandsingle sat.baseline

6.310-32x10-52x10-'

ISL system(Ku-Band

up/down links)

D5.910-2

BxlO-5

10'6

aAssumes uplink Eb/No improvement of 4.2 dB and 11.3 dB crosslink Eb/No.^Assumes uplink Eb/No improvement of 4.7 dB and 11.3 dB crosslink Eb/No.

TABLE 6. - COMMUNICATIONS PAYLOAD COMPARISON

Antenna size, ftHPA Power, WPayload power, WPayload weight, IbISL weight margin, Ib

C-Band

Baseline

5.4 by 2.35.0

376325

ISL

6.6 by 5.02.0

15025075

Ku-Band

Baseline

1 .8 by 0.720.0

1079557

ISL .

2.2 by 1.67.2

396278279

TABLE 7. - EARTH STATION COMPARISON

EIRP, dBWG/T, dB/KHPA power, WAntenna diameter, ftNoise temp, K

C-Band

Baseline/ISLScenario 1

79.7

25.51175.0

21.3125.3

ISLScenario '2

75.5

21.3525.0

19.7

281.2

Ku-Band

Baseline/ISLScenario 1

77.129.0

140.019.7

430.0

ISLScenario 2

72.4

24.356.018.0

1050.0

TABLE 8. - SYSTEM COSTS

First unit, $MSecond unit, $MNonrecurring, $HSpace segment cost, JME/S RF cost, H

C-Band

BaselineScenario

38347614867

ISLScenario 1

322964

^12567

ISLScenario 2

'393578

a!5247.5

Ku-Band

BaselineScenario

5751

114 '22253.5

ISLScenario 1

363272

auo53.5

ISLScenario 2

5852

_U6a226

37

aDoes not include ISL.

JGfNAL PAGE ISOF POOR QUALITY

Figure 1. Two-satellite ISL configuration.

C-BANDCONUS ORBITAL-

ARC

I I I180 160 140 120 100 80 60

SATELLITE LONGITUDE, °W

I I40 20 0

Figure 2. - Amount of satellite addressable traffic accessiblefrom C-band orbit locations.

aio

Q-</l

O

UJQ_

100

80

60

40

20

0

MAINBEAM

_X TRAFFIC

rMAIN

C-BANDCONUS ORBITAL

ARC

I I I200 180 160 40 20 0140 120 100 80 60

SATELLITE LONGITUDE, °W

50% GEOGRAPHICCOVERAGE LIMIT

Figure 3. - ISL, mainbeam traffic routing from C-band orbitlocations.

- 100

§S. 80

i 60

a 40CO

20aa a

Ku-BAND*-CONUS ORBITAL-

ARC

I I I L I VI200 180 . 160 140 120 100 80 60 40 20 0

; SATELLITE LONGITUDE, °W

Figure 4. - Amount oi satellite addressable traffic accessiblefrom Ku-band orbit.locations. •

100

80

2£ 60

t— oLUO

UJQ.

40

20

0

Ku-BAND-CONUS ORBITAL-*

.ARC

I I200 180 160 40 20140 120 100 80 60

SATELLITE LONGITUDE, °W

50% GEOGRAPHICCOVERAGE LIMIT '

Figure 5. - ISL, mainbeam traffic routing from Ku-band orbitlocations.

aCO<s.l/lon

100

90

80

70

60

50

40

30

:20i

10

-Ka-BANDCONUSORBITAL.ARC

60 140 40J20120 100 80 60

SATELLITE LONGITUDE, °W

Figure 6. - Amount of satellite addressable traffic accessiblefrom Ka-band orbit locations.

Ka-BANDCONUSORBITALARC

SATELLITE LONGITUDE,

50% GEOGRAPHICCOVERATE LIMIT H

Figure 7. - ISL, mainbeam traffic routing from Ka-band orbitlocations.

A 500 MbpsO 1 GbpsD 2 GbpsO 4 Gbps

FD Nd: YAG (6" OPTICS)60 GHz (51 ANTENNAS)

CX60)

160

140

120

- 100i—~i :80LLJ

I 60

40

20

0

aoo)

20

10

0 1

NUMBERS IN PARANTHESES ARELASER (mW) AND TWT (W) OUTPUT .POWERS AT THE SEPARATIONS SHOWN

I I I I I I I30 40 50 60 70 80 90 100 110 120 130 140

SATELLITE SPACING, °

Figure 8. ISL weight versus 9 (10~7 BER).

2

COtoocco

10 _|

6 -

4 -

.aUJ

l/lOa;o

302826

2422

201816

1412

1086

4

20

60 GHz (51 REFLECTORS)FDNd: YAG (6" OPTICS)

OUTPUTPOWER400 mW130 W200 mW75 W

100 mW25 W50 mW

10W

5W

I I I I I I I I I I30 40 50 60 70 80 90 100 110 120 130 140

SATELLITE SPACING, 9. deg

Figure 9. - ISL BER and Eb/No versus 9, 1 Gbps crosslink.

1. Report No.

NASA TM-87215

2. Government Accession No. 3. Recipient's Catalog No.

4. Title and Subtitle 5. Report Date

Application of Intersatellite Links to DomesticSatellite Systems 6. Performing Organization Code

650-60-267. Authors)

Denise S. Ponchak and Rodney L. Spence

8. Performing Organization Report No.

E-2829

10. Work Unit No.

9. Performing Organization Name and Address

National Aeronautics and Space AdministrationLewis Research CenterCleveland, Ohio 44135

11. Contract or Grant No.

12. Sponsoring Agency Name and Address

National Aeronautics and Space AdministrationWashington, D.C. 20546

13. Type of Report and Period Covered

Technical Memorandum

14. Sponsoring Agency Code

15. Supplementary Notes

Prepared for the llth Communications Satellite Systems Conference sponsored bythe American Institute of Aeronautics and Astronautics, San Diego, California,March 16-20, 1986.

16. Abstract

This paper presents the results of a study on intersatellite link (ISL) applica-tions for domestic satellite communications. The objective was to determine ifany technical, economic, or performance benefits could be gained by introducingintersatellite links into a domestic satellite communication network. In thisstudy, several key systems issues of domestic ISL's are addressed. These includethe effect of a skewed traffic distribution on the selection of ISL satelliteorbit locations, tolerable satellite spacing, and crosslink traffic-handlingrequirements. An ISL technology assessment is made by performing a parametriclink analysis for several microwave and optical implementations. The impact ofthe crosslink on the end-to-end link performance is investigated for both regen-erative and nonregenerative ISL architectures. Lastly, a comparison is madebetween single satellite systems operating at C-, and Ku-bands and the corre-sponding ISL systems in terms of ground segment cost, space segment cost, andnet link performance. Results indicate that ISL's can effectively expand theCONUS orbital arc, with a 60 GHz ISL implementation being the most attractive.

17. Key Words (Suggested by Authors))

Intersatellite links (ISL)Millimeter-wave (MMW)Laser

19. Security Classif. (of this report)

Unclassified

18. Distribution Statement

UnclassifiedSTAR Category

20. Security Classif. (of this page)

Unclassified

- unlimited17

21. No. of pages 22. Price*

*For sale by the National Technical Information Service, Springfield, Virginia 22161

National Aeronautics andSpace Administration

Lewis Research CenterCleveland. Ohio 44135

Official BusinessPenalty for Private Use $300

SECOND CLASS MAIL

ADDRESS CORRECTION REQUESTED

Postage and Fees PaidNational Aeronautics andSpace AdministrationNASA-451

NASA