30 Jul/Aug 2003

18225 69th Pl WLynnwood, WA 98037kc7yxd@arrl.net

D-STAR: New Modesfor VHF/UHF

Amateur Radio, Part 1

By John Gibbs, KC7YXD

Our friends in JARL have created a newdigital communication standard. Let’s look attheir new system, and what’s in it for hams.

This article is the first in a three-part series describing and ana-lyzing a new communication

standard developed by the JapanAmateur Radio League (JARL). Thefirst part focuses on the advantagesof upgrading our VHF/UHF equip-ment to a new, more capable system.The second article in the series ad-dresses the technical design consider-ations of a digital voice and high-speeddata system in the VHF/UHF spec-trum. The third, and final, article dis-cusses the D-STAR standard and howit addresses the needs and technicalissues raised in the two previous parts.

JARL has developed a new openstandard for VHF/UHF digital radiocalled D-STAR. The system supportsboth digitally modulated voice trans-mission and data transmission, includ-ing Internet connections, at DSL rates.

At a time when the third-generation(3G) cell phone proposals for high-speeddata have been severely delayed,

D-STAR presents Amateur Radio op-erators with the opportunity to bringthe Internet Age to mobile and portableoperation.

I have been fortunate to be one ofthe first hams to see and use the proto-type transceivers of this new Amateurcommunication system. Therefore, Iwould like to take this opportunity toshare some of the knowledge and expe-rience I have gained. In this article, Iwill present the objectives of the newJARL D-STAR system and provide aglimpse of the capabilities of this newsystem and the engineering tradeoffsthat went into the system design. I hopeyou will find it interesting both in de-veloping an understanding of this sys-tem and as an insight into the designprocess of a digital radio system.

One of the major goals of D-STARis that it be an open system. Thisseries of articles contains enough sys-tem details for a skilled ham to de-velop a homebrew D-STAR digitalvoice transceiver.1

If you are interested in this system,you may soon have a chance to try ityourself. Because the FCC encourages

the Amateur Radio community to de-velop new digital modes, the US hasthe regulatory structure in place forhams to use an all-digital voice andhigh-speed data radio system withoutspecial licensing or permits. US hamswill therefore be the first in the worldwith the opportunity to use the newD-STAR system illustrated in Fig 1.

Regular readers of QEX know thathams have been experimenting withdigital voice.2 In the US, the FCC en-courages hams to continue such ex-perimentation to demonstrate ourstewardship of our spectrum. In addi-tion to individual efforts, AmateurRadio organizations have also beenpromoting digital radio. For years,TAPR has focused exclusively on ad-vancing the state of the art in Ama-teur Radio digital communications.The ARRL has increased its efforts inthis area by sponsoring the establish-ment of Technology Task Force Work-ing Groups on digital voice, high-speedmultimedia and software radio.

In addition to the efforts of indi-vidual hams and their organizations,one manufacturer has already intro-duced a digital voice option to handheldVHF/UHF radios. Unfortunately, these1Notes appear on page 34.

gibb.pmd 6/3/2003, 12:14 PM30

Jul/Aug 2003 31

radios are of limited usefulness becausethe necessary repeater infrastructurefor VHF/UHF digital voice operationswith these radios does not exist. Theseradios will not work through existinganalog repeaters and the necessarydigital repeaters have not been devel-oped. Early on in the D-STAR planning,the JARL recognized that developingVHF/UHF digital capabilities also re-quires developing new standards fordigital voice repeaters and the links be-tween repeaters.

D-STAR HistoryThe D-STAR standard not only ad-

dresses the needs of VHF/UHF voiceand data communications with mobileand handheld radios, but also providesthe standards for repeater-to-repeaterlinking and Internet access. It was clearthat developing and testing such a com-plex system would take many man-years of engineering and testing. Theefforts that would be required to achievethis in a timely manner would far ex-ceed any reasonable expectations ofvolunteer ham labor, no matter howdedicated. So, the JARL contracted withthe Amateur Radio manufacturerICOM to develop and evaluate D-STARprototype hardware. D-STAR has beenunder development since 1998 and thesystem operation has been proven in laband field tests. The result of all thisdevelopment effort is about to bearfruit. The JARL expects to finalize theD-STAR standard this summer.

A D-STAR mobile transceivercalled the ICOM ID-1 was used forfield trials in the Tokyo area (see thecover photo on this issue) and shownat three US Amateur Radio shows:Dayton Hamvention 2002 and 2003and the Digital Communication Con-ference (DCC) in Denver last fall.Since then, repeaters and microwavelinks have been developed and are cur-

rently available on a limited basis toapplication developers in the US. Allthese D-STAR compatible componentswill soon be shipping in quantity, andwe expect that other manufacturerswill be shipping D-STAR-compatibleradios in the future.

Existing VHF/UHF SystemProperties

To replace any existing system witha new standard, there must be compel-ling reasons for incurring the expenseof new equipment. So it is good to startthe discussion of D-STAR with a lookat the capabilities and limitations of ourexisting VHF/UHF Amateur communi-cation systems in Tables 1 and 2. To dothis, let us look at the capabilities of arepresentative voice repeater systemthat covers the Pacific Northwest andbeyond: the Evergreen Intertie.

The Evergreen Intertie connectsmore than 23 repeaters by full-duplexUHF radio links that are transparentto the user. From my location, thereare two main links in the system,a North-South link that connectsWestern Washington and Oregon andan East-West link that crosses theCascade Mountains and connects tocities in Eastern Washington.

Users can control switches usingDTMF tones to connect repeaters tothe link. The way this particular sys-tem is configured, a minimum of threeswitches must be set by a user or con-trol operator to connect two repeaters.In a more-extreme case, sevenswitches must be set to talk from Se-attle repeater K7NWS to Portland re-peater KJ7IY. Of course, each switchthat connects to the next link may bealready in use, so it can be difficultfor a user to establish such links if thesystem is heavily used.

On a repeater link, only one con-tact can be held on a link at a time.

That is, unlike the telephone system,there is no multiplexing on links. If alink is in use or out of service, there isno way to link the repeaters unless analternate path is available.

Another difference from the tele-phone trunk system lies in how a linkis established. In the telephone sys-tem, the system automatically picksthe link based on the call destinationand the trunk lines currently avail-able. In the Evergreen Intertie, theuser must determine the logical paththrough the repeater links and toknow the DTMF codes for each of theswitches.

Amateur Radio packet systems of-fer an analogous set of featuresthrough dedicated packet nodes (sim-plex repeaters.) Packet radio is usedto transfer data (for example, com-puter files) and for keyboard contacts.It might even be possible to have adigital voice contact on a 9600-bpssystem. However, the system is packet-oriented, which means that real-timecommunications are not guaranteed.Unless the system was very lightlyloaded, some of the voice packetswould be unacceptably delayed.

Both the amateur voice repeatersand packet nodes are FM systems.Within the limitations of the existinganalog FM technology, some very cre-ative communication solutions havebeen developed. For instance, sub-au-dible tone codes are used to protect re-peaters against accidental activationby interference. DTMF is used to con-trol some repeater functions, activatea phone patch or for selective callingof amateurs.

Enterprising hams continually addnew capabilities to the systems. A few

Fig 1—JARL’s proposed D-STAR system offers digital voice and data communication on1.2-GHz with repeaters linked on 10 GHz and Internet gateways.

Table 1—Existing VHF/UHF AmateurRadio System Features

• Voice is FM, half duplex• Data is FSK, simplex• CTCSS protection• DTMF control• Linked repeaters• 1200/9600 packet• APRS• Voice over Internet

Table 2—Existing VHF/UHF AmateurRadio System Limitations

• Spectrally inefficient• Low speed data• One QSO per link• Difficult to establish links• Cannot mix data and voice

gibb.pmd 6/3/2003, 12:14 PM31

32 Jul/Aug 2003

examples that come to mind includesatellite gateways, GPS-based locationsystems (APRS) and worldwide com-munications with VHF/UHF trans-ceivers when voice-over-Internet-pro-tocol (VOIP) is added to a repeater. Yetwe are rapidly approaching the limitof what we can do with the existinginfrastructure, as we can see by inves-tigating the limitations of existingVHF/UHF Amateur Radio systems(see Table 2).

Spectral EfficiencyThe first major limitation is spec-

tral efficiency. The amateur commun-ity’s VHF/UHF spectrum usage hasnot changed despite dramatic im-provements in communication tech-nology that have occurred in the lastfew years. The FCC views the radiospectrum as a finite resource thatmust be efficiently shared amongmany users. There are many new po-tential users appearing for the VHF/UHF spectrum, and they are oftenlooking at the spectrum that has beenallocated for Amateur Radio. A grow-ing part of ARRL resources are beingdevoted to spectrum-defense.

However, defending our usage ofthese valuable frequencies will be-come more difficult because the cur-rent amateur FM system is not spec-trally efficient. Today, the FCC onlygrants new licenses in Land Mobileservices to users that meet reducedspectrum-occupancy requirements.The FCC calls this “refarming.” TheFCC has extended this principle toother radio services. For example, theexisting GMRS spectrum wasrefarmed with FRS channels placedbetween the old GMRS channels.There is no reason why we should feelthat the Amateur Radio Service wouldbe exempt from the requirement forspectral efficiency.

As a matter of fact, the quest forspectral efficiency is increasing. Cur-rent Land Mobile services refarmingis from 25-kHz channel spacing to12.5 kHz. However, in the next fewyears, the FCC plans to repeat thisprocess and force all new Land Mo-bile Service licenses to use equipmentcompatible with 6.25-kHz channelspacing. It is not clear that FM radioscan be developed that will meet thestability and bandwidth requirementsof such a system and be sold at an af-fordable price, so radios using othertechnologies may need to be developed.But what is certain from modulationtheory is that as the deviation is re-duced, the signal quality advantage ofFM over AM systems (including SSB)quickly disappears.

Data RateA second limitation is the 9600-

baud rate limitations of existing com-mercially available radios. In thisdata-intensive Internet age, this speedis woefully inadequate. Any new sys-tem should have the capability of sup-porting data transfers at speeds rival-ing DSL.

Limited LinkingAs mentioned above in the Ever-

green Intertie example, a severelimitation of the current FM-analog re-peater system is the number of contactsthat can be handled by a link. An idealrepeater-to-repeater link would havemuch wider bandwidth than the exist-ing links. This bandwidth would thenbe dynamically allocated between voicecontacts and high-speed data users.

In addition, as we have seen, it isdifficult for the user to establish linksbetween repeaters. With today’s low-cost computing power, a more auto-mated method of calling a distant hamcould certainly be developed.

Data and VoiceThe final and very significant limi-

tation of the existing systems is thatrepeaters can only handle voice or data,not both simultaneously. As we shall seelater, there are many applications thatcould be opened to Amateur Radio op-erators if this feature could be incorpo-rated in a new VHF/UHF system.

Desired PropertiesHaving considered the features and

limitations of current analog FM sys-tems, let us next consider what prop-erties any new analog or digital sys-tems should have. Ideally, any newsystem should solve the limitations ofthe existing systems without losingany of the features. In addition, anynew system should have the proper-ties described below and in Table 3.

Compatible with RegulationsIf a new system required changes in

the FCC regulations, it might takeyears before it could be adopted. Fortu-nately, this is not necessary for digitalvoice and data transmissions in the US.

The FCC encourages the AmateurRadio community to develop digitalvoice and new digital data communica-tion systems. One example is the FCC’sattitude toward the new HF digitalmodes such as PSK31. There was someconcern in the amateur community thatthe encoding used in PSK31, calledVaricode, would be considered a formof encryption and hence barred by thePart 97 regulations. However, the FCChas clearly and repeatedly stated thatencoding is not encryption and that aslong as the encoding method is publicthere is no regulatory problem.

Where should digital voice transmis-sions occur in the current band plan?Again the FCC in its encouragement ofdigital radio has already decided thatdigital voice operations belong in thephone bands.3

What regulatory issues are therefor new linked systems? WilliamCross, W3TN, of the FCC Policy andRules Branch, made it clear at his pre-sentation at Dayton last year thatthere are no rules specifically writtenfor linked systems; the FCC regulatesstations, not systems.

Finally, emergency operation isclearly one of the requirements for anyAmateur Radio system to meet one ofthe Part 97 justifications for the Ama-teur Radio Service. Any new systemmust not only be available to supportcommunication needs during an area-wide emergency, but an amateur mustbe able to break in and use the sys-tem during an accident or other localemergency.

A Worldwide StandardWhy is it desirable to be compat-

ible with an international standard?The primary reason for all countriesto share the same system architectureis to make the radios affordable. If theradios and the repeater infrastructureare not within the financial reach ofthe majority of the ham community,no amount of extra features will makea new system successful.

Radio costs are dramatically depen-dent on sales volume both because re-search and development costs can bespread over a larger number of radiosand because manufacturing unit costsdecrease as volume increases. TodayNorth America accounts for about athird of the world’s Amateur Radiolicenses. Japan, with a far smaller popu-lation, has about another third. So if

Table 3—Desired Properties of NewVHF/UHF Amateur System

• Compatible with regulations• Worldwide standard• Enables new applications• Enhancement friendly• Scalable• Open standard• Repeater operation• Linking repeaters• Simplex• High speed data• ANI• Expandable• Affordable

gibb.pmd 6/3/2003, 12:15 PM32

Jul/Aug 2003 33

Japan and the US agree on a singlestandard, the manufacturing volumecould double, which would dramaticallyreduce radio costs for all amateurs. Ifthe rest of the world also joined in thestandard, further cost reductions wouldfollow. As an example of what can hap-pen when there is not a world standard,consider the 222-MHz band.

Because the band is not availableworldwide, manufacturers offer a lim-ited number of transceiver models. Andwe find that equivalent rigs (if evenavailable) tend to be more expensivethan the high-volume 2-meter rigs.

A second reason for a worldwidestandard is that tying repeaters to-gether via the Internet is becoming apopular feature of today’s repeatersystems. Any new system must sup-port this trend for both voice and high-speed data. This could be done byspecifying a protocol that two other-wise incompatible systems would useto exchange data; but due to the eco-nomic issues discussed above, this isa less-than-ideal solution.

New Applications andEnhancements

To take full advantage of the digi-tal revolution in Amateur Radio, mini-mum standards will need to be estab-lished. Unlike a telecom system, whichneeds rigid standardization, an Ama-teur Radio system must have justenough standardization to allow com-munication, without inhibiting inno-vation.4 This is a difficult balance andrequires a great deal of work duringthe system design to properly blendthese conflicting requirements.

Ideally, any new communicationsystem would be a perfect “wirelesscable.” Of course, one of the things ofinterest to QEX readers is that nocommunication system is perfect. Thestudy of the impairments and experi-menting with ways to improve com-munications over an impaired chan-nel are interesting areas of our hobby.Every system involves a great deal ofcompromise; that is a part of daily lifefor the communication-system designengineer. An ideal system for AmateurRadio would allow a great deal of ex-perimentation that could be layeredon top of a well-functioning, but notoverly constraining, radio system.

A Scalable SystemA cell-phone system will not work

until the complete infrastructure isdeployed in an area. Clearly this is notpractical for Amateur Radio. Any newAmateur Radio VHF/UHF systemmust be able to work with only onerepeater and even—within the limitsof line-of-sight propagation—without

repeaters at all. Multiple repeatersand the linking of repeaters can comelater as the user base develops or asfunds become available.

In addition, it is important to beable to communicate with other hamswho have not upgraded to the newsystem. This can be done in two ways:The radios themselves could have ana-log FM capability, or the repeater canbe capable of interfacing with the ex-isting analog radio repeaters.

The system should also be scalableto facilitate emergency operation. Natu-ral or man-made disasters can destroyboth the commercial and amateur com-munication infrastructure. A new com-munication system should be able towork immediately without repeatersand be flexible enough so that sparetransceivers can be connected to quicklyform an emergency, temporary repeater.

An Open StandardAs this is an Amateur Radio system,

the system should be available frommore than one manufacturer. It is de-sirable to have competition between theradio manufacturers to keep prices lowand encourage innovation within theframework of a new standard.

Yet it is at least as important to thoseof us with QEX leanings that the sys-tem technology is such that a ham witha sufficient technical background canmake any part of the system, includingradios, repeaters and repeater links.Because so much of leading-edge com-munication technology is the intellec-tual property (IP) of communicationcompanies, the requirement that hamsbe able to develop and publicize equip-ment without violating patent rightsbecomes a system-design challenge.

Further RequirementsHigh-Speed Data

Fixed site-to-site data links atgreater than 9600 bits/s are rare, butnot unknown in Amateur Radio. Anynew VHF/UHF system should supporthigh-speed data, not only for these fixedlinks but also for mobile and portableoperation. This means the system musttolerate channel impairments likemultipath and Doppler shift.

To be able to interface into the vastarray of low-cost hardware and soft-ware available today, the new VHF/UHF system should appear as a “wire-less Ethernet cable” to a PC. Youshould be able to use any software thatcan interface with the IEEE 802.3(10Base-T) Ethernet, connect a cablefrom the PC to the transceiver and usethe computer just as if it was a wiredconnection. For example, if the otherhalf of the RF link is connected to an

ISP, then an Internet browser willwork seamlessly and the Amateurworld will have high-speed wirelessInternet connections.

Repeater OperationBecause the FCC requires that the

control operator is able to shut downthe repeater and repeaters are oftenin remote locations, remote control ofthe repeater must be designed into anynew system. With today’s systems,shutting down the system is the onlyoption available if a user abuses therepeater. A new system should havethe ability to block offenders from re-peater access while still allowing oth-ers to access the system.

Control over landlines is certainlyrequired, but radio control operation isnecessary for those sites without phone-line access. Of course, it is highly desir-able that the control operator can usethe data capabilities of the system tomonitor the status of the system andcontrol many other features.

Linking RepeatersAny new Amateur Radio system

must have a wide-bandwidth links ca-pable of supporting multiplexed con-tacts. Multiple contacts are necessarybecause the system should supportmultiple repeaters at a single site aswell as different pairs of sites using thelink at the same time. The link mustalso support both voice and data so thatwe do not have to invest in two links.The best way to meet this need is forthe link itself to be digital and the voicedigitized for transmission over the link.Because more than one pair of repeat-ers is using the link at a time (multi-plexed), the link must be full duplex.

Each site in the system repeats thehigh-speed link signal and extractsand adds the contacts that are appro-priate for its site. Normally only onerepeater in the system would broad-cast the contact. The other repeatersignore contacts that are not directedto them. Otherwise one contact wouldtie up the entire repeater system.

Most often, these repeater linkswould be microwave links. However,because of the distances involved, itmay be more attractive to use Internetlinking with some repeater systems.Any new VHF/UHF Amateur systemshould support both types of links.

SimplexAny new system, no matter what

benefits are available from repeateroperation, must be able to work sim-plex without a repeater. And unliketypical fixed digital radio networks, itis critical that anyone tuning the bandscan immediately listen in on a contact.

gibb.pmd 6/3/2003, 12:15 PM33

34 Jul/Aug 2003

This requires two properties that arenot available on many digital radio sys-tems. First, the digital voice systemmust work without handshakes. Thatway it is possible to have one talker andmany listeners. Second, it must not benecessary to wait for the start of a newtransmission to acquire the carrier,frame and bit synchronization neces-sary to demodulate the contact.

Automatic Number Identification (ANI)Any new Amateur system should

support a higher level of automationin establishing a contact. An equiva-lent feature to what the commercialLand Mobile market calls “AutomaticNumber Identification” (ANI) needs tobe developed. With this feature en-abled, your radio opens squelch onlywhen your call sign is received. Somehams do this today with a DTMF coderather than their call sign, but DTMFcodes are not unique and DTMF sig-naling is very slow. If desired, theradio can beep when you are called orin mobile applications, the horn cansound.

This call-sign squelching principleshould also be extended to repeaters.Repeaters today use CTCSS tones tokeep from being opened accidentallyby interfering signals. On any newsystem, you should use the repeatercall sign to unambiguously and easilyopen the repeater. This would be fol-lowed by the call signs of the party youare calling and the repeater they useso that the system can route your call.

AffordabiltyThe system must be designed to be

tolerant of the performance limita-tions of reasonably priced components.Particularly with high-speed data atUHF, the frequency and time accuracyrequirements of many modern digitalradio systems are so great as to beprohibitively expensive for amateurusage.

Another reason for designing thesystem to be reasonably tolerant ofcomponent and system variations isto allow enterprising Amateurs theopportunity to homebrew their ownD-STAR hardware.

Also to save user cost, the systemshould not be designed for full duplexoperation. Full-duplex operation re-quires expensive isolation between thetransmitter and the receiver. Half-duplex and simplex operations allowthe sharing of many expensive com-ponents between the receiver andtransmitter. Finally, radio amateursalmost always operate in these low-cost modes so there is no problem withconversion to new modes.

Advantages of Digital ModulationIn the next part of this series, we

delve into the engineering designconsiderations that were made in de-veloping D-STAR and the technical de-tails of its implementation. Yet, let usconclude by investigating the advan-tages of a new system based on digi-tal instead of analog modulation.

The first advantage of digital modu-lation is the ability to reduce occupiedspectrum. To meet the regulatory pres-sures discussed above and to reducethe congestion on our bands, any newsystem must be spectrally efficient.One solution would be to stay with ananalog FM system and reduce the de-viation, as the FCC has required of theLand Mobile Service. However, doingso reduces the audio quality that is themajor benefit of FM.

It is a better solution to changethe modulation completely andtransmit voice using digital modu-lation. However, without careful sys-tem design, switching to digitalvoice could actually increase thebandwidth required for voice com-munication because of the high bitrates required by uncom-pressedvoice. For instance, pulse-codemodulation (PCM), as used by theUS telephone standard, requires adigital stream of 64,000 bits/s. Evenwith very elaborate modulationschemes, that high bit rate would re-quire a much wider bandwidth thancurrent FM voice radios.

The enabling technology for digitalvoice is digital signal processing(DSP). It has long been realized thatthe information in a voice signal ishighly redundant and that it shouldbe possible to establish good transmis-sions without sending the redundantinformation. Modern high-speed, low-cost signal processors and very cleveralgorithms can dramatically reducethe bit rate required to accurately re-produce a human voice in real time.We shall see that it is possible to getsimilar voice quality at only 2400 bits/s and therefore occupy far less spec-trum than today’s FM systems.

A second advantage of digitalmodulation is improved quality. Withwide-band systems like HDTV andhigh-speed wireless Internet service,the most important advantage of digi-tizing transmissions is the ability touse DSP to correct for transmissionerrors. This results in improved per-formance over the vast majority of theoperating area. In analog radio com-munication systems we have littlechoice but to live with the errorscaused by propagation, noise and in-terference (both natural and man-

made). We can sometimes increase thereceived signal-to-noise ratio by in-creasing transmitter power and/orusing gain antennas. However the fad-ing caused by multipath propagationis not improved by increasing thetransmitted signal power. Particularlyin mobile wide-band systems, multi-path can be a serious problem. Youhave probably heard multipath im-pairment if you listen to FM broad-casts in your car. It is perhaps mostnoticeable if you are at a stop light andhear distortion but move a few feetand the distortion disappears.

But perhaps the greatest advantagesof digital voice transmission are theadded features that are possible whena digital data payload is added to a voicecontact. The availability of simulta-neous low-speed data transmissionwith voice transmissions opens up awhole world of new possibilities forAmateur Radio. Imagine sending stillpictures, maps, small data files andGPS position while rag chewing. Whatwould it be like to have “instant mes-saging” on your radio? What a great wayto politely break into a contact!

Notes1Remember, however, the D-STAR specifi-

cations discussed in this article have notyet been finalized.

2See for example “Practical HF DigitalVoice,” by Charles Brain, G4GUO, andAndy Talbot, G4JNT, QEX May/June2000, pp 3-8.

3See, for instance, the editors preface andPaul Rinaldo’s, W4RI, comments in asidebar in “Practical HF Digital Voice,”pp 3 and 4.

4See “Technical Standards in Amateur Ra-dio,” Doug Smith, KF6DX, QEX Mar/Apr2003, p 2.An Extra class license holder, John

usually is found on the HF bands, pri-marily operating PSK31. At age 3,John exhibited early talents in elec-tronics by “helping” his dad fix a TV.He plugged the speaker into a wallsocket! Despite this traumatic start, hespent his youth building Heathkit andEico equipment, repairing vacuum-tube radios and TVs and designingand building numerous homebrewprojects including a Morse decoderhigh-school project built with resistor-transistor logic in the mid 1960s.

With BSE and MSEE degrees incontrol and communication theory, hehas worked for Hewlett-Packard in thefields of spectrum and network analy-sis and frequency synthesis. He is cur-rently the research department engi-neering manager at ICOM America,where his primary interests are digi-tal communications and DSP. Johnhas eight patents and is currently ap-plying for four more.

gibb.pmd 6/3/2003, 12:15 PM34

22 Sept/Oct 2003

18225 69th Pl WLynnwood, WA 98037kc7yxd@arrl.net

D-STAR, Part 2:Design Considerations

By John Gibbs, KC7YXD

Come learn what JARL put into theirproposed new VHF communication standard.

In the first segment of this series,we considered the attributes wewould like to see in any new VHF/

UHF Amateur Radio system. In thissecond segment, we discuss the tech-nical issues involved with selecting theparameters of an Amateur Radio digi-tal system to meet those attributes.

Perhaps the easiest way to organizethe design considerations for a digitalradio system is to use the OSI Model.The OSI Model is officially known asThe Basic Reference Model for OpenSystems Interconnection. We start thedescription with the bottom level of theOSI model, the physical layer. Then wewill work our way up to the top levels,which are open to amateur experimen-tation and application development.

Physical LayerTransceiver Frequency

We hams have a large spectrum al-location at 1.2 GHz (60 MHz wide in

the US) that is little used today. In fact,we should use this spectrum or risk los-ing it to commercial interests. Also, ifwe want to develop a system to sendhigh-speed data, we will need a wide-bandwidth signal and there is littleavailable spectrum at 70 cm. If it is de-sired to use the D-STAR protocol atlower frequencies, the high-speed datamode could be dropped. In fact, a pro-totype portable 2-m HT using only thedigital voice mode has been developedand was shown this spring at IWCE.

Like the previous FM system, thesystem design logically calls for half-duplex operation for digital voice andsimplex for high-speed data.

Repeater Link FrequencySince it is desired to have multiple

contacts and high-speed data packetson the repeater link, a wide bandwidthis required. Therefore the repeaterbackbone must be at microwave fre-quencies and the amateur band at10 GHz is a logical choice. Today’sequipment generates usable 10 GHzpower, is affordable and the 10 MHzof bandwidth is practical. The bi-

directional, asynchronous nature ofmultiple contacts demands a full-duplex repeater link.

Given the high-speed data require-ments of the system and the desire touse digital voice to reduce transceiverspectrum requirements, the repeaterlink should be digital. With the asyn-chronous nature of the system, packetmode is a natural choice. However,because a packet system does notguarantee real-time communication,voice should be given priority overdata to minimize the possibility ofvoice disruptions.

ModulationAn ideal modulation system should

generate a signal that has a narrowspectrum, with low side lobes so thatit does not interfere with other nearbyusers. On our crowded ham bands, thisis becoming more of an issue daily.

There are several commonly usedmodulations for digital data. The digi-tal modulation scheme chosen will sig-nificantly affect the performance of acommunication system. Generally wewant to maximize the data rate within

gibb.pmd 8/4/2003, 9:49 AM22

Sept/Oct 2003 23

the constraints of acceptable level oflatency, available bandwidth, accept-able error rate, product costs and op-erating environment (that is mobile,portable, fixed-link). In particular,mobile and portable operation causesvariable multipath fading and fastphase shifts that can wreak havocwith digital radios.

Less spectrally efficient modulationsgenerally have better operation charac-teristics in poorer SNR conditions. Also,they are more forgiving of frequency-offset errors between the transmitterand receiver and frequency and phaseresponse error on the channel—animportant consideration if costs are tobe kept down in a UHF system.

4 FSK— FSK, MSK and GMSK arevery attractive because they are con-stant amplitude modulation. Thismeans that the power amplifier canbe class C, which offers low cost andexcellent power efficiency.

FSK has of course been used inamateur systems for years, datingback at least to the introduction ofRTTY. Newer variations on FSK usemore frequencies than just mark andspace. For instance, the new weak sig-nal mode, JT44 uses very slow FSK(about 5 Hz data rate) with 44 differ-ent frequencies each corresponding toa character.1 But at the higher datarates needed for a VHF/UHF digitalvoice system, four FSK frequenciesoffers an attractive option forimproved FSK performance.

GMSK— Among FSK, MSK andGMSK, GMSK offers the best spectralefficiency with only a slight degrada-tion in the BER compared to FSK andMSK. These advantages have madeGMSK one of the most popular digi-tal modulations worldwide. Othermore complex modulations like QPSKrequire a more expensive linear poweramplifier that also typically requiresmore current, which is critical in por-table operation.

GMSK low-pass-filters the datastream with a filter that approximatesa Gaussian time and frequencyresponse. A Gaussian filter is usedbecause of its desirable properties inboth the time and frequency domains.This filtering reduces the high-fre-quency content of the modulation andtherefore narrows the frequency spec-trum of the modulated signal whilewidening the data response minimally.However, as you continue to narrowthe filter, the spectrum continues tonarrow and the time response of thefilter lengthens. This causes the peakamplitude to decrease and the adja-cent data tails of the time response to

interfere with the decoding of the de-sired symbol, a phenomenon calledinter-symbol interference (ISI).

A typical compromise between ISIand bandwidth used by many systemsis for the bandwidth/data rate ratio tobe equal to 0.5. This yields almost nodegradation due to ISI compared toMSK and yet dramatically reduces thespectral occupancy of GMSK com-pared to MSK.

QPSK— In theory, quadraturephase-shift modulation could have aconstant amplitude format. However,the rapid switching of the input datacauses a QPSK signal to have largesidebands that destroy its spectral effi-ciency. Therefore in practice, raised co-sine filters are used on input data toreduce these sidebands. To preserve thewave shape induced by these filters re-quires the use of a more expensive andless power-efficient linear amplifier. Ifa class-C amplifier were used withQPSK, the sidebands that wereremoved by the cosine filter would beregenerated.

In the presence of additive whiteGaussian noise (AWGN), QPSK re-quires about 3 dB less signal-to-noisethan does FSK. However, in real chan-nels, with multipath and poorlysynchronized receivers, the 3-dBadvantage quickly disappears.

Data Link LayerTime Division Multiple Access (TDMA)

TDMA is one of the two commonlyused multiplexing standards for cellu-lar phones. The cell tower site acts asthe master clock and assigns a time slotto each of several cell phones that areassigned the same frequency. For properoperation, it is critical that each phonetransmit and receive exactly in its as-signed time slot. This is not attractivefor amateur simplex operations becauseoperations are as two or more equals,and there is no master to determine theclock and assign time slots.

Any Amateur Radio system has towork without a centralized frequencyreference and master clock. In addi-tion, the radios must be able to acquiresignals that are somewhat off fre-quency and acquire timing without theneed for a separately transmittedclock signal. These requirements maymake an amateur system less spec-trally efficient than a centrally-con-trolled system like the cell phone, butthey are more in keeping with thespirit of Amateur Radio, particularlythe capability to operate when theinfrastructure is destroyed.

Code Division Multiple Access(CDMA)

CDMA (also known as spread spec-

trum) is also used for multiplexing cel-lular phones. In CDMA, several cellphones share the same frequency andtransmit simultaneously. Each phoneon a frequency is modulated with a codesequence that spreads the spectrum ina unique way. If the receiver is synchro-nized and has the same code sequence,then the signal is restored. Otherwise,the signals from other phones becomepart of the background noise.

An important limitation on the sys-tem is that undesired signals are notcompletely rejected. Depending on thelength of the codes used and the atten-dant difficulty in synchronizing, per-haps 20-30 dB of so-called processinggain can be attained. Therefore, a strongnearby CDMA signal can overpower amore distant signal. This classic prob-lem with spread-spectrum communica-tions is called “the near-far problem.”

In a cell-phone system, this problemis addressed by power control. Since allthe nearby phones are communicatingwith the same nearby cell site, the cellsite remotely controls the power levelof each phone to minimize the possibil-ity of interference. However, in AmateurRadio, particularly with multiple sim-plex contacts, this is not a solution.

Frequency Domain Multiplexing(FDM)

TDMA, CDMA and other modernmultiplexing schemes require coordi-nation between the units that is in-compatible with the basic goals of theAmateur Radio Service. One of themajor justifications for our service inthe US is emergency service. TDMAand CDMA require an infrastructureto provide the coordination. This in-frastructure would quite possibly bedestroyed in an emergency. So, the bestsolution for Amateur Radio is what wehave traditionally used, FDM.

Network LayerIn the network layer, the binary

data stream is divided into discretepackets of finite length. In addition,error checking is performed by cyclicredundancy check (CRC) at this level.If an error is detected, it is correctedby the retransmission of packets.

Transport LayerIn the transport layer, we multiplex

and split all the data streams we needto send and receive. In an AmateurRadio system, we would typically needto include repeater control data;source, destination and routing infor-mation (that is, call signs of both op-erators and repeaters used); and whatis called the payload, which is the voiceor data to be sent.1Notes appear on page 28.

gibb.pmd 8/4/2003, 9:50 AM23

24 Sept/Oct 2003

Table 1—D-STAR Transmission Characteristics

Mode Transmission Speed BandwidthBackbone 10 Mbps or less 10.5 MHzData 128 kbps or less 130 kHzDigital Voice ITU 8 kbps 9 kHz AMBE 2.4 kbps 5 kHz

Presentation LayerCodec

As mentioned in the first segmentof this series, simple PCM encoding ofvoice results in a 64,000 bits/s datastream. Codecs have been developedto compress voice with good qualitydown to 2400 bits/s and lower. Thesecodecs develop their extreme datacompression by modeling short seg-ments of the human voice and onlytransmitting the reduced informationneeded to describe the voice model.

One of the major difficulties indesigning a digital voice radio is in test-ing the voice quality. High-compressioncodecs are designed to work with a hu-man voice; traditional tests like fre-quency response and harmonic distor-tion with sinusoidal tones do not gen-erate meaningful results. Consequently,a subjective method of testing calledmean opinion score (MOS) has beendeveloped. MOS is estimated by a testwith a group of normal listeners whoare asked their opinion on a five-pointscale (1 = bad, 5 = excellent) and the

Fig 1—Header additions with TCP/IP protocol.

results are averaged together.2A very important factor in conduct-

ing MOS tests is the acoustical envi-ronment. Since the codec is designedto highly compress the information ina human voice, it is easy to imaginethat the presence of other signals andnoise can severely affect the perfor-mance of the system. An excellent testfor Amateur Radio is performance inan automotive environment includingengine noise and wind noise from anopen window.

A final issue in codec selection isthe MOS performance in the presenceof the channel impairments we com-monly find in VHF/UHF commun-

ication paths. As Digital VoiceSystems point out on their Web site,“Vocoders…designed for extremelylow bit-error rates, such as those en-countered in land-line communica-tions, often experience seriousdegradation when applied to the muchhigher bit-error rates found in wire-less communications. Consequently,it is important to consider robustnessto channel degradations during thevocoder-algorithm design process.”

ScramblingBit synchronization and accurate

level slicing in the receiver require fre-quent transitions in the data (no long

gibb.pmd 8/4/2003, 9:50 AM24

Sept/Oct 2003 25

strings of ones or zeros). To ensure thatcondition, most digital radio systemsuse a device called a scrambler to ran-domize the input data stream.

“Scrambler” is an unfortunate termfor people who are familiar with ana-log radios; it is common to interpret thisas encryption. The FCC of courseforbids encryption for amateurs. Scram-bling is not an attempt to hide the mes-sage content, however; it is a fixed andpublished method known by all poten-tial receivers for converting the inputdata stream into a data stream withshort strings of ones or zeros.

Scrambling is typically done witha shift register and exclusive OR gates.CCITT recommendation V.26 recom-mends this procedure.

Application LayerThis is the layer where hams can

begin to customize the system and addtheir own applications. In addition,this is the level where the system de-sign allows for user control entry anddata entry, both data from the IEEE10BaseT Ethernet and analog audiofrom the microphone.

D-STAR Proposed StandardsAs stated in the first part of this

article, D-STAR is not a finalized stan-dard at the time of this writing. How-ever, the field trials are finished andstandard publication begins here.Table 1 shows the system as it standsat this writing.

To describe the proposed D-STARstandard, we will start at the input

Table 2—Call-Sign Combinations and the System Function

(Uses Evergreen Intertie Call sign examples3)Called Departure Destination OwnStation Repeater Repeater Station FunctionCQCQCQ KB7WUK K7NWS KC7YXD CQ PortlandN7ABC KB7WUK K7NWS KC7YXD Call N7ABC in PortlandN7ABC K7NWS K7NWS KC7YXD Call N7ABC on local repeaterN7ABC DIRECT DIRECT KC7YXD Simplex

Fig 2—Proposed bit pattern in high-speed digital mode.

side of the transceiver and work ourway out to the antenna, working ourway down the OSI model. We willthen see how the standard definesthe repeater operation and the linksbetween repeaters. First we will con-sider the high-speed data mode andthen the digital voice.

High-Speed DataThe standard interface for high-

speed data into and from the D-Starsystem is IEEE802.3 (10BaseT Ether-net.) In Fig 1 you can see how theD-STAR transmitter adds a radioheader extension to the Ethernet mes-sage just as the Ethernet protocol addeda header to the Internet Protocol, TCP/IP. Since this radio header is strippedoff in the receiver, the radio link appearsto be a “wireless Ethernet cable.” There-fore, it is possible using existingsoftware (such as browsers) to commu-nicate the same images, text and voiceas is handled by Ethernet, includinglinks to the Internet, without modi-fication.

Data MultiplexingFig 2 shows the details of a com-

munication packet from the radio partin Fig 1. Each packet consists of aradio header and the Ethernet packetdescribed above followed by an error-checking frame. The radio header isworthwhile to study in some depth asit shows many of the D-STAR systemcapabilities.

Each frame of the radio header isidentical in both the high-speed data

mode and in the digital voice mode.If the standard is approved as pro-posed, it will contain the followinginformation:

The first two fields are common tomost digital radios, the bit sync andthe frame sync. These preambles aredesigned to allow the receive modemto establish timing and level lock asquickly as possible.

The flag field describes the contentof the data field.Bit 7 Data or voice communication

flag.Bit 6 Repeater or simplex flag.Bit 5 Communication-interruption

flag.Bit 4 Control signal, data or voice

signal flag.Bit 3 Emergency/normal signal

flag.Bits 2-0 T r a n s m i s s i o n - c o n t r o l

bits (see below).The ID field can hold four call signs:1. The local repeater you are access-

ing (optional).2. The linked distant repeater the

called party is using (optional).3. The station you are calling (can be

CQ).4. Your own call sign.

The PFCS field is a check word forthe header. Some of these bits requirea little more explanation if we are tounderstand the operation of the sys-tem. Notice that when bit 3 of the flagfield is set, you are asking for an emer-gency break-in. (On many FM repeat-ers you would say “break” today.) For

gibb.pmd 8/4/2003, 9:51 AM25

26 Sept/Oct 2003

instance if you want to report an acci-dent, pushing the emergency key onthe radio will set bit 3 and all D-STARradios within range will open squelchand their volume to be set high.

The flag field bits 0, 1 and 2 areused for transmission control. Theyimplement functions like ACK, ARQand repeater control.

One of these repeater control func-tions is repeater lockout. Repeaterlockout is used mainly to block illegalstations. A D-STAR repeater can holda black list of call signs that have con-sistently violated repeater and/or FCCrules. If a blacklisted station calls therepeater, the repeater does not repeatthe message but instead calls theoffending station back with the lockoutbit set. The offending station’s radio willthen display a message indicating thatit is blocked from the repeater. So now,it is not necessary to shut down therepeater for everyone when one indi-vidual is misusing the repeater.

Another important field for under-standing the capabilities of the systemis the ID field. Understanding the IDfield is important because it shows thegreat flexibility available in the sys-tem calling capabilities. The first thingto notice is that the D-STAR protocolautomatically IDs at every transmis-sion. This easily meets the FCC IDrequirements for ID at start, end andevery 10 minutes of transmission.

Next, to understand how the four IDfields work, Table 2 illustrates the con-tents of each field if KC7YXD were totransmit on a fictional D-STAR Ever-green Inter-tie system. It is not neces-sary to always fill in all the ID fields. Ifyou respond to a CQ or a call directedat you, your D-STAR transceiver willautomatically fill in the fields for you.

Digital VoiceCodec— Those of you who had a

chance to see the D-STAR presentationat last year’s Dayton Hamvention or atthe DCC in Denver last fall may recallthat the Digital Voice mode occupied8 kHz of bandwidth using the ITUG.723.1 Codec standard. At that time,

Table 3—AMBE Vocoder-BasedSystems

InmarsatThurayaIridiumAPCO Project 25 (IMBE)G4GUO & G4JNT HF Digital VoiceSystem

Fig 3—Bit pattern in digital-voice mode.

the two codecs were undergoing fieldtrials; today the JARL has selectedAMBE as the standard. The two stan-dards under consideration were the ITUstandard G723.1 and a Digital VoiceSystems proprietary codec that uses theAMBE algorithm.

ITU G.723.1 uses an ACELP (alge-braic code-excited linear prediction)algorithm that generates a 5.3-kbpsdata stream. With an algorithm delayof 37 ms, the total wireless-communi-cation-throughput delay is a little over100 ms, quite reasonable for half-duplex communications.

AMBE stands for advanced multi-band excitation. AMBE can use differ-ent levels of compression to trade offvoice quality and bit rate. Tests showthat at the 2.4-kbps data rate, the voicequality was at least as good as thehigher-data-rate ITU G.723.1 over realradio links. The algorithmic delay isonly slightly longer than G723.1(44 ms), so the factor-of-two improve-ment in data rate (and spectral effi-ciency) comes with no noticeable latencyincrease.

The data-rate reduction from theAMBE codec is particularly significantbecause of worldwide pressure fromregulatory agencies to reduce the oc-cupied bandwidth of voice communi-cations sufficiently to allow 6.25-kHzsignal spacing. When using a modula-tion scheme sufficiently robust to givereliable communication in mobile andportable applications, only the AMBEdata rate meets this signal-spacingrequirement.

The decision between codecs is com-plicated by the fact that G.723.1 is an

open public standard codec whereasAMBE is the patented intellectualproperty of Digital Voice Systems.Unlike many companies, however, thepresent owner of this technology sup-ports the Amateur Radio communityand is willing to sell these parts insmall quantities.

The JARL is not alone in decidingon AMBE for its high voice quality andvery low bit rate. Table 3 shows severaldigital systems that have standardizedon this codec technology. For instance,the Telecommunications Industry Asso-ciation (TIA) selected DSVI’s codec tech-nology over CELP and other codecs forthe APCO Project 25 North Americanland-mobile radio-communication sys-tem. This is particularly significant be-cause at least two Amateur Radiogroups are evaluating Project 25 radiosas an alternative digital radio standardfor amateur usage.

Fig 3 illustrates the bit pattern usedin the digital voice mode when theAMBE codec is used. As mentionedbefore, the radio header is identical tothe high-speed digital mode radioheader and so will not be discussedhere.

The most interesting part of Fig 3is that the digital-voice data framesare interleaved with data frames.These frames are currently reservedby the D-STAR standard with no dedi-cated usage by the system overhead.This means that the system is capableof supporting a 2400-bits/s datastream from a user application whilethe user is talking on the system!Notice that the D-STAR system itselfprovides no error detection for thisdata, so it would be up to the user’sapplication to provide error detectionand error correction. This and otheroverhead would decrease the end-to-end data rate slightly; but if radios arebuilt to exploit this capability, hamscould potentially add many interest-ing features to the D-STAR system.

What is not shown in Fig 3 is thatthe frame and sync fields are repeatedoften so that the errors between thetransmitter and receiver clocks can be

gibb.pmd 8/4/2003, 9:51 AM26

Sept/Oct 2003 27

corrected without requiring a masterclock signal. It also means that anotheramateur can tune into the middle of acontact and listen to the conversationwithout waiting for the sync frames ofthe radio header at the next over.

ModulationSeveral modulation methods were

investigated during the development ofthe D-STAR standard. Modulationstested included GMSK, FSK, 4-FSK,MSK and QPSK. GMSK has beenselected for the backbone line betweenrepeaters. The standard for the portableand mobile transceivers may includemore than one modulation format.

Gaussian minimum-shift keying(GMSK) and quadrature phase-shiftkeying (QPSK) are the two finalists. Athird, 4-FSK, has been recently pro-posed as an alternate standard and isnow under investigation. The reason forthe delay is that selecting the bestmodulation for D-STAR real worldapplications is not a trivial exercise. Inreal mobile communications systems,the link between a moving node and abase station will be subject tomultipath, which results in Rayleighfading. This will have a significant ef-fect on the resultant BER performance,possibly increasing the required C/N fora specific BER by as much as 10 dB.

QPSK is commonly used in fixed-link communication systems. Underideal conditions, QPSK would give

Fig 4—Occupied bandwidth of digital radio.

better performance than GMSK or4-FSK and its higher spectral effi-ciency is obviously attractive. How-ever, QPSK’s higher spectral efficiencyalso leads to higher susceptibility totransmission impairments such asmultipath and phase hits. Yet, the big-gest disadvantage of QPSK is the needfor extremely linear power amplifiersto avoid spectral growth—what weAmateurs call splatter.

The front-runner at the time thisarticle is written is GMSK. In its favor,GMSK is a well-proven technology, andprobably the most commonly used digi-tal modulation in the world for portableapplications (see Table 4). GMSK hastwo basic advantages. First, it is morerobust than QPSK to common transmis-sion impairments. Second, GMSK, as aform of FSK, has constant amplitudeand can therefore use very efficientclass-C power amplifiers. Third, GMSKis not as sensitive to frequency errorsbetween the transmitter and receiver.Because no master frequency referenceis available in the D-STAR system, tun-ing errors on a 1.2-GHz signal can be

substantial, particularly with the ex-tremes of temperature found in portableoperation. The alternatives are to suf-fer the expense of a precision frequencyreference in all the radios or adopt amodulation method like GMSK that ismore tolerant of frequency errors.

However, GMSK is not so spectrallyefficient as QPSK. For instance, at128 kbps, GMSK with a BT product of1/2 occupies a bandwidth of 135 kHz.For the same data rate, QPSK requiresonly 83 kHz.

The best solution is probably to usea codec with the AMBE algorithm de-scribed earlier that reduces the datarate as far as possible and then useGMSK for more robust communica-tions, but the tests still continue.

Fig 4 is a somewhat busy graphthat dramatically shows the differencein occupied bandwidth. The existingFM system bandwidth can be deter-mined by Carson’s Rule to be about16 kHz. While not all combinations ofmodulation and codec algorithms areshown, you can clearly see that youcan fit many more digital voice con-tacts into the same spectrum.

RepeaterThe digital voice mode is half-

duplex with a 20-MHz offset betweentransmit and receive frequencies.High-speed data is simplex.

As shown in Fig 5, much of the re-peater site function is to provide a

Table 4—GMSK is used in SystemsWorldwide

GSM cell phoneDECTCellular Digital Packet Data (CDPD)Mobiltex

gibb.pmd 8/4/2003, 9:51 AM27

28 Sept/Oct 2003

gateway to other repeaters, both atother sites and to repeaters on differ-ent modes. The repeater also providesan interface to the repeater backbonelink to the Internet if desired. The sys-tem is designed to support remote con-trol of the repeater over radio and/orlandline links.

Repeaters could be linked via theInternet instead of the backbone, butbecause of bandwidth limitations,much of the high-speed multiple con-tact capability would be lost.

Repeater Call Sign ProtectionTo protect repeaters from co-chan-

nel interference, CTCSS tones areused to prevent interfering signalsfrom triggering the repeater. In the D-STAR system, the digital header con-tains the call sign of the repeater tobe accessed. If the repeater does notsee its call sign in the message, therepeater is not opened.

The Backbone Repeater LinkOne of the major advantages of the

D-STAR system is the full-duplex

10-MHz-bandwidth backbone link be-tween repeaters. This wide bandwidthallows multiple voice and data con-tacts to occur simultaneously on thelink. An analysis of the frequency ofuse of data and voice communicationsdemonstrated that a 10-Mbit/s full-duplex link would support the needsof up to 12 linked repeaters.

The high-speed data and digital-voice data streams from multiple re-peaters are multiplexed into a singledata stream according to the asynchro-nous transfer mode (ATM) standard.This 10-Mbit/s data stream is GMSKmodulated onto a 10-GHz carrier, re-sulting in an approximately 10-MHzwide signal.

The ATM cell is made up of a short53-byte packet that consists of a 5-byteheader and a 48-byte payload. The ATMcell is sent to the required destinationaccording to the preset list that is setby the ATM switch set at each repeatersite. Because the priority level can bedesignated in the header, voice signalsarrive in real time. This avoids the de-lays that happen with VoIP on the

existing Internet Protocol.Backbone field tests have been car-

ried out with a 36-dB-gain parabolicantenna and a 1-W transmitter. Heavyrains in southern Japan of more than12 inches per hour limit the practicaldistance that the repeaters can beseparated. It was found that taking intoconsideration these extreme weatherconditions, the maximum range foruninterrupted communications is about12 miles. Obviously, the fog and rain-fall at the location and the acceptableprobability of communication interrup-tion dramatically affect this number.

Notes1“JT44: New Digital Mode for Weak Signals,”

(World Above 50 MHz) QST, June 2002,pp 81-82.

2D. Smith, KF6DX, “Digital Voice: The NextNew Mode?” QST, January 2002. For adiscussion of MOS, see the sidebar, “HowDo I Sound?” on pp 29.

3K7NWS is a Seattle, Washington, repeater andKB7WUK is a Portland, Oregon, repeater onthe Evergreen Intertie. These examples as-sume an identical system to the EvergreenIntertie but based on D-STAR.

Fig 5—Integrated site with analog and digital radio repeaters, high-speed backbone and Internet connection.

gibb.pmd 8/4/2003, 9:52 AM28

gibbs.pmd 10/1/2003, 12:30 PM42

D-STAR, Part 3:•Implementation•

We’ve seen the “whys” and “hows” of D-STAR. Let’s look at the hardware and possible uses for the system.

This article, the final part of the series, investigates the block diagram and performance of the

prototype equipment to better under­stand the design issues of a D-STAR digital radio.

The hardware used in testing the D-STAR standard is shown in Fig 1 and the performance of the mobile unit is summarized in Table 1. Some of this hardware is available today and we expect several manufacturers will of­fer hardware soon.

Recall that the D-STAR standard has only recently finalized the selec­tion of the modulation and codec. Pro­totype testing demonstrated that GMSK modulation and the AMBE 2020 codec gives the best combination of spectral efficiency and robust com­munications.

The IF and RF parts of the block diagram (see Fig 2) of the ID-1 shows a straightforward dual-conversion

18225 69th Pl W Lynnwood, WA 98037 kc7yxd@arrl.net

By John Gibbs, KC7YXD

superheterodyne design that should look familiar to those experienced with analog rigs. However, several issues in a digital-radio IF are not clear from the block diagram.

IF Design Issues The first issue with digital-radio

IFs is that the group delay of the IF structure is critical. While analog ra­dio designers can ignore phase linear­

ity, group-delay variations need to be less than about 10% of the data pe­riod to avoid excessive BER due to intersymbol interference.

The second issue with digital radio IFs is that IF bandwidth must be wider than that of an equivalent ana­log design. It must be wider so that significant energy does not fall near the band edges of the filter because there the group delay is not constant.

Table 1 ID-1 Specifications Summary

Operating frequency 1.2 GHz Amateur Radio Band Operating Modes FM (analog voice) (FDMA) 0.5GMSK (digital voice / data) Data Rate 4.8 kbps (voice) / 128 kbps (data) CODEC AMBE Data Interface IEEE802.3 (10Base-T) RF Power 10 W/1 W Receive Sensitivity FM –16 dBu (typical) 4.8 kbps GMSK Voice –10 dBu

128 kbps GMSK Data + 2 dBu Switching time 10 mS (digital mode) GMSK Modulation Quadrature Modulator / FPGA (baseband)

42 Nov/Dec 2003

gibbs.pmd 10/1/2003, 12:43 PM43

It often rises significantly and displays what are called “ears” (from their shape). This is particularly true in the receiver IF where transmitter and re­ceiver relative-frequency tuning errors may cause the signal to be off center in the IF. Unfortunately, this increases the noise and interfering signals that pass through to the detector.

The quality of these IFs is mea­sured by the sensitivity numbers in the specifications and in the “eye” dia­grams in Fig 3. The well-open eye means that the receiver can easily dis­tinguish between the plus and minus signal sent and therefore decode with very little BER. Fig 4 shows how the BER improves as S/N increases in the digital voice mode.

The final issue with digital radio IFs is the quality of the local oscilla­tors. First, as implied above, the fre­quency reference must be accurate and temperature-stable if communi­cation is to be established at UHF with a reasonably wide receiver IF. Second, the close-in phase noise of the local oscillators must be kept low, particu­larly if QPSK and other high-data-rate modulations are used. Excess oscilla­tor noise can increase the BER just as effectively as actual channel impair­ments. One of the advantages of using GMSK is its relatively low sen­sitivity to these receiver problems, as shown in Fig 5.

Baseband Design Issues The baseband hardware and modu­

lators have far more obvious differ­ences in this digital radio block diagram in Fig 2. For instance, on the transmitter side, the audio input is immediately converted to digital form, even if the radio is in the analog FM mode. This digital information is then signal-processed digitally and modu­lated onto the first IF. The modulation is accomplished by an I/Q modulator made with an FPGA. When teamed with DSP, an I/Q modulator is a very versatile component that can handle any form of modulation needed in the ID-1. It is even possible to produce narrowband-FM with the digitized voice. (The analog FM feature is de­sired for compatibility with existing analog radios.)

D-STAR Applications D-STAR is very much a “blank

slate,” waiting for amateurs to write upon it. We can exploit its capabilities for a variety of old and new uses. Here are a few of the many suggestions we have heard from the Amateur Radio community as possible applications of D-STAR.

Mobile and Portable Internet Access The application that springs almost

instantly to everyone’s mind is high­speed wireless Internet access. Part of the reason is that the Internet has be­come such an important communica­tion and information tool in hams’ lives today. Another reason is all the hype built up around third-generation (3G) cell phones and the DOCOMO system in Japan. Yet, with today’s meltdown in telecommunication commerce, it could be years before a 3G phone sys­tem is deployed in the US. So, with the deployment of D-STAR, hams could once again have a leading technology that the rest of the population would envy and that might encourage more people to get their tickets.

In support of this vision of D-STAR as an Amateur Radio community growth agent, it is interesting to watch the reaction of inactive no-code hams. For a variety of reasons, they got their tickets, but never really got interested in the hobby. Often when they see a D-STAR demonstration, you can see their eyes light up and almost hear the gears turning in their head! Several

have said that a system like D-STAR would get them active again.

Because this is Amateur Radio, there will be some restrictions on this vision of high-speed wireless Internet. The FCC does not allow encryption, so there is no guarantee of privacy. Anyone can look over your shoulder and read your e-mail.

Some hams bring up the issue of advertising and pornography. Control operators will be responsible for the content passing through their repeat­ers exactly as they are today. However, this does not seem to be a very diffi­cult issue. Inexpensive software exists today that can filter out this offend­ing material. Control operators can easily incorporate so-called “kiddy fil­ter” software into the repeater’s Internet interface. If the existing soft­ware does not quite fit our application, then resourceful hams will develop better software!

Then there is the issue of third­party traffic. Again, the control opera­tor is responsible for ensuring that no illegal third-party traffic passes through his or her station.

Fig 1—Currently available hardware (counterclockwise, from upper right): RC-24 Control Head, ID-1 1.2 GHz transceiver, ID-RP1D 1.2 GHz data repeater, ID-RP1VS 1.2 GHz voice repeater, ID-RP1L 10 GHz backbone repeater and AH-1045/1080 parabolic antenna.

Nov/Dec 2003 43

gibbs.pmd 10/1/2003, 12:43 PM44

44

Fig 2—1.2 GHz transceiver block diagram.

Fig 3—“Eye” patterns for digital voice (A) and high-speed data (B).

2003Nov/Dec

gibbs.pmd 10/1/2003, 12:44 PM45

Nov/Dec

Fig 4—Bit-error rate versus RF level.

Fig 5—Frequency error versus sensitivity.

Combination of 802.11 and D-STARIt would be surprising if manufac-

turers did not quickly develop ahandheld D-STAR-compatible radio,but the high-speed data mode will nec-essarily have reduced range comparedto a mobile rig with a good antennaand more power. What could you do ifyou wanted to connect a notebookcomputer to the Internet, but you arebeyond the limited range of ahandheld?

When hams have a range problemwith handhelds on today’s analog FMsystem, they sometimes cross-bandrepeat using their car’s mobile radio.A similar solution could be imple-mented for high-speed data usingD-STAR and a wireless LAN accesspoint. A D-STAR mobile in your carcould be cross-band (and cross-mode)connected to an access point installedin your car. Only an Ethernet cable isneeded for this connection (no PC). Ifyou already had a wireless LAN cardin your notebook computer, you wouldbe ready to go. Your notebook com-puter now has high-speed Internetaccess with the range of the high-power mobile.

Other High-Speed Data ApplicationsThe Internet is so pervasive today

that we sometimes forget that thereare many other uses for high-speeddata transmission. Here are two high-speed data applications that havearisen in D-STAR discussions.

Local Amateur Intranet: Ratherthan connecting to the Internet, aclub-sponsored repeater could offer awireless, wide-area Intranet. Whatmight they put on the site? It certainlyis a good place to make available therepeater system’s operation guide andrules. To encourage D-STAR experi-mentation, it would be useful to have

posting of hams’ experiences with thesystem as well as freeware andshareware that they have found use-ful in D-STAR operation.

Visitors to the area could downloadinformation they need, even at 3 AM.Are you looking for a good Mexicanrestaurant, or do you need a quiet mo-tel away from the highway? TheIntranet could have suggestions fromother hams on file, and you coulddownload maps, driving directions andeven pictures.

The possibilities multiply enor-mously if your notebook computer hasGPS. Now D-STAR can guide you ex-actly to your destination with accuratemaps and directions that better reflectthe local driving conditions than thoseprovided by major services on theInternet. Local hams could help youavoid traffic problems caused by tem-porary road closures and accidents.

Emergency Communications: An-other D-STAR Intranet application is

emergency communications. Even ifthe local D-STAR repeater wereknocked out, temporary repeaterscould quickly be assembled using twotransceivers back-to-back. Trainingneeds are minimized by using stan-dard Internet browsers. When an op-erator comes onto the system, he caneasily access stored files and bringhimself up-to-date on the situationwithout distracting others.

Possible Add-Ons andEnhancements

We wrap up this discussion of thenew D-STAR system with a treatmentof the possible directions in which ap-plications might evolve. D-STAR is notmeant to be a turnkey communicationsystem like the cell-phone system. In-stead, it is an infrastructure that hamscan use to meet current and futurecommunication needs. Most impor-tantly, it is a flexible, highly capablesystem that allows amateurs, them-

452003

gibbs.pmd 10/1/2003, 2:55 PM46

selves, to expand their service. Rather than depending on manufacturers to provide new features and applications, we expect the amateur community will develop add-ons to the system that will address the major goals of Amateur Radio including emergency communi­cations, experimentation and just plain fun! Hams who have seen early demonstrations of the D-STAR system have generated the following ideas.

Power to the People! We hams have our own opinions of

how products should be designed and which features should be added. One of the great things about the D-STAR system is that for a large part, it is possible for us to try out our ideas and further the state of the art. Error cor­rection is one area that is ripe for con­tribution by hams.

As data rates increase or as we push the range, decoding errors begin to be significant in any digital radio system. This is less of a problem for a properly designed digital voice system, because it is not significantly dis­turbed by BER levels that would ren­der digital file transfers impossible. Yet any high-speed digital mode can use any help it can get.

Because of the importance of trans­ferring data quickly and accurately, there has been a great deal of theoreti­cal work done on coding and error cor­rection. Tom McDermott, N5EG, gives a good introduction to the many coding techniques used in digital radio includ­ing Reed-Solomon, Golay and convolu­tional codes.1 Newer codes called “turbo” codes have been developed that ap­proach the theoretical limit on how fast information can be transmitted over a noisy band-limited channel.

However, these codes are only opti­mal if the interference is what we call AWGN (additive white Gaussian noise). This is true because the mathematics of AWGN is well understood. The bad news is that most of the impairments we find in real radio communications do not match this nice mathematical model. The good news is this is an op­portunity for Amateur Radio to again advance the state of the art.

An interesting example of the pos­sibilities of error correction is the ubiq­uitous CD player. A few years ago, I saw a demonstration of the power of the error correction used in CDs. The professor had drilled large holes in a CD and despite these obvious faults in the data stream, the music played perfectly without a click, pop or drop out! Perhaps some enterprising ham can discover the way to make just as

1Notes appear on page 47.

dramatic improvement in radio.

Interleaving Spreads Bursts of Errors

Wireless communication channels with fades of the signal power are prone to errors occurring in bursts. Burst errors can cause problems by breaking error-control codes when the number of errors exceeds the maxi­mum number of correctable errors for the specific code used.For short bursts, intraframe interleaving improves per­formance by spreading the burst of errors over several different code words. For example, if four code words, each containing 23 bits that can cor­rect up to 3-bit errors, are used in a frame consecutively, then a burst er­ror 4 bits long will break a single code. However, if the four code words were interleaved (that is, bit 1 codeword 1, bit 1 codeword 2, bit 1 codeword 3, bit 1 codeword 4, bit 2 codeword 1, and so forth), each code word would contain only one error, which could easily be corrected. Since intraframe interleav­ing only modifies the bit ordering within the current frame, no addi­tional delay is generally needed for implementation. If additional delay can be tolerated, interframe (more than just the current frame) interleav­ing can be used to further increase the performance with longer burst errors.

Mixed Voice and Data As we saw in the section on the

D-STAR standard, the proposed digital voice protocol has the ability to trans­mit low-speed user data simultaneously with voice.The first-generation D-STAR transceivers minimally support this feature. However, as new radios are in­troduced, it is expected that hams will develop applications that exploit this capability. Notice that in the D-STAR system, this is referred to as low-speed data. Yet the data rate is actually about 2400 baud, faster than the old 1200 baud of amateur systems (and yes, slower than the 9600 baud used in higher-speed systems).

What could we do with this feature? How about the equivalent of the Internet’s “instant messaging”? With instant messaging, messages could be added by the sender or even from a third party (where legal) and added by the repeater. Imagine that you are in the middle of a contact when • A DX alert displays on your mobile

for a country you need, or • A printer attached to your trans­

ceiver prints out route instructions to your club’s Field Day site, or

• Your spouse sends you the grocery list and reminds you that the lawn needs mowing—well, maybe that

isn’t such a good feature! How about doing instant messag­

ing one better and send instant pic­tures. The miniature cameras used recently in cellular phones are about 96×96 pixels; that is less than 10 kbits. So, a picture could be sent in less than 30 seconds simultaneously with a voice contact.

In a sense, this voice and data ca­pability is like DSL: you can talk over the same channel while data are transmitted—although not at DSL speeds in this mode. The data you can send through this channel are limited only by your imagination. For in­stance, what do you think about mix­ing voice and next-generation APRS?

VoIP and D-STAR VoIP voice communication is of

course possible in the digital data mode because it does not matter what information is carried in the data. However, VoIP is not a very attractive method of communication via Ama­teur Radio today. It often suffers from poor voice quality scores due to the very long latency from the intensive signal processing and because the Internet does not give priority to voice packets.2 These voice quality problems would certainly not be helped by the 128 kbps data rate of D-STAR.

Finally, VoIP on D-STAR is spec­trally inefficient, requiring 130 kHz of bandwidth compared to less than 6 kHz for the highly compressed D-STAR digital voice mode. Still, for applications that require higher-speed data simultaneous with voice, inven­tive amateurs may find solutions to these problems.

Registration The D-STAR proposal currently

keeps a list of amateurs (call signs) who have accessed the system. So, if you want to call me, KC7YXD, you don’t need to know the linking re­peater. The system simply finds the repeater I last accessed and automati­cally routes your call to me. A logical extension of this capability is that if my radio is on the repeater frequency, the system can poll it and automati­cally register me onto that repeater.

This feature could be extended to keep a database at each repeater of each registered amateur’s interests. How would our hobby change if you could call “CQ Collins radio collector” and automatically link to someone on the other side of the country or per­haps the other side of the world?

Roaming Another feature that hams could

add to the system is roaming. What if,

46 Nov/Dec 2003

gibbs.pmd 10/1/2003, 12:44 PM47

when driving through an area, the re­peater could download into the radio memory the frequencies and call signs of nearby D-STAR repeaters? Then as I drive away from the repeater, the radio is all set to access the next re­peater. Never again sit down with a repeater book and program the radio before the next trip!

Of course, if we were to add GPS capability and the D-STAR repeater database held the footprints, calls and frequencies of adjacent repeaters, the radio could automatically switch re­peater sites as you drive through an area!

Trunking “Trunking” is a land-mobile-radio

term for a system that uses multiple repeaters to support many contacts at once. Most trunking systems use a “home channel” for calling, then the system assigns a clear repeater fre­quency to complete the contact. The radios then automatically go to the assigned frequency. The basic advan­tage of trunking is that the system can support many more users simulta­neously than with individual systems. Effectively, it lets one listen to all re­peaters in an area by only monitoring

the home channel for a call. Since the D-STAR system sends call signs digi­tally, it is easy to envision a simple computer program that would moni­tor the home channel and alert me when I am being called.

Conclusion Clearly, Amateur Radio is at a

crossroads today. Technical and regu­latory forces are pushing us out of our well-proven but inefficient ways. The possibilities that digital radio brings to our hobby are truly limited only by our imagination.

I hope this article has stirred your imagination and stimulated your interest in the possibilities of digital voice and high-speed data in Amateur Radio today. Perhaps you will be inspired to try the D-STAR sys­tem and maybe even develop applica­tions or variations of the D-STAR system.

Recommended Reading Visit www.dvsinc.com to read more about

AMBE and to hear voice samples at vari­ous coding rates.

D.W. Griffin and J.S. Lim, “Multiband Excita­tion Vocoder,” IEEE Transactions on Acoustics, Speech and Signal Processing, Vol 36, No 8, August 1988, pp 1223-1235.

Notes 1Wireless Digital Communications: Design

and Theory, Tom McCermott, N5EG, Tuc­son Amateur Packet Radio Corporation, 1996.

2The D-STAR digital voice mode addresses this problem by giving real-time data, such as voice, priority over repeater links. At age 3, John exhibited early talents

in electronics by “helping” his dad fix a TV. He plugged the speaker into a wall socket! Despite this traumatic start, he spent his youth building Heathkit and EICO equipment, repairing vacuum­tube radios and TVs and designing and building numerous homebrew projects including a Morse decoder high-school project built with resistor-transistor logic in the mid-1960s.

With BSE and MSEE degrees in control and communication theory, he has worked for Hewlett-Packard in the fields of spectrum and network analy­sis and frequency synthesis. He is currently the research department en­gineering manager at ICOM America, where his primary interests are digi­tal communications and DSP. John has eight patents and is currently ap­plying for four more.

An Extra class license holder, John usually is found on the HF bands, pri­marily operating PSK31.

Nov/Dec 2003 47