IEEE Communications Magazine • December 2009 210163-6804/09/$25.00 © 2009 IEEE

INTRODUCTION

The development of the Aloha Systemproject at the University of Hawaiibeginning in the late 1960s provides his-torical insight into a variety of wirelessdata network applications at the start ofthe 21st century. In 1971 the Aloha Sys-tem established and operated a UHFterrestrial data network (AlohaNet)within the state of Hawaii. In 1973 theAloha System used a VHF transponderin an experimental NASA satellite(ATS-1) to demonstrate an internation-al satellite data network (PacNet) con-necting NASA in California and fiveuniversities in the United States, Japan,and Australia. Also in 1973 the AlohaSystem pioneered the unconventionaluse of a conventional commercial Com-sat channel to link the AlohaNet andPacNet to ARPANet in the continentalUnited States.

AlohaNet within Hawaii and PacNetcovering the Pacific were both based onthe use of a random access channelarchitecture now generally referred toas an ALOHA channel. Today ALOHAchannels are utilized in all major mobilenetworks and in almost all two-waysatellite data networks.

Every time you power up yourmobile phone or use that phone toestablish a voice, SMS, or Internet con-nection, the very first packet transmit-ted is sent via an ALOHA randomaccess channel. The importance of theALOHA channel within all wirelessnetworks has increased as the utiliza-tion of such networks for SMS andInternet traffic has increased. In addi-tion, the initial Aloha System researchled directly to the work on carrier sensemultiple access (CSMA) at the Univer-sity of California at Los Angeles(UCLA) which provided the theoreticalfoundation for both CSMA with colli-

sion detection (CSMA/CD) in Ethernetand CSMA with collision avoidance(CSMA/CA) in WiFi and WiMAX.

This anecdotal account of the devel-opment of the Aloha System is basedon an invited paper [1] first publishedin a special issue of IEEE Transactionson Information Theory edited by JimMassey in March 1985 [2].

BACKGROUNDIn the late 1960s a number of effortswere in progress to use the existingworldwide telephone network to pro-vide remote access to computer systemsfor terminals and, in some more ambi-tious cases, to provide connectionsamong large information processingsystems for resource sharing. The term“resource sharing” at that time wasoften taken to mean a sharing of hard-ware that would be considered primi-tive by today’s standards. Nevertheless,in 1968 it was becoming apparent thatthe existing circuit-switched telephonenetwork architecture was not well suit-ed to the rapidly emerging data net-working needs of the 1970s. Indeed, itwould have been surprising had such anetwork architecture, shaped by therequirements of voice communicationsat the end of the 19th century [3], beencompatible with the emerging require-ments of data communication networksat the end of the 20th century. The orig-inal goal of the Aloha System was toinvestigate the use of radio communica-tions as an alternative to the telephonesystem for computer communicationsand to “determine those situationswhere radio communications are prefer-able to conventional wire communica-tions” [4].

At that time the University ofHawaii was composed of a main cam-pus in Manoa Valley near Honolulu, a

four-year college in Hilo, Hawaii, andfive two-year community colleges onthe islands of Oahu, Kauai, Maui, andHawaii, all within a radius of about 300km from Honolulu. In September 1968a faculty group in the Department ofElectrical Engineering at the Universityof Hawaii began to plan for an experi-mental radio-linked computer networkto connect all of these locations togeth-er in order to permit sharing of thecomputer resources on the main cam-pus. Even after the decision to build aradio data network, however, a numberof basic design decisions dealing withthe choice of network architecture hadto be addressed. From the beginning itwas clear that the nature of the radiochannel provided new system designoptions not available in systems usingconventional point-to-point telephonechannels. As the network planning pro-ceeded, this key point assumed greaterand greater significance. It would begratifying to be able to report that thiskey difference between radio channels,with broadcasting and multiple accesscapabilities, and conventional point-to-point wire (or microwave) channels wasappreciated as we began the project.Unfortunately, such an appreciationdeveloped among the members of theproject only with time. As is the case inmany real-world situations, our fore-sight was not as clear as our hindsight.

Even at the beginning of the project,however, it was understood that theintermittent operation typical of inter-active computer terminals was a con-vincing argument against theassignment of point-to-point channelsin a conventional frequency-divisionmultiple access (FDMA) or time-divi-sion multiple access (TDMA) manner.Some more efficient form of sharing acommon communication channel

HISTORY OF COMMUNICATIONSEDITED BY MISCHA SCHWARTZ

The following article provides a concise, clearly written dis-cussion of the history of the well-known Aloha protocol by Prof.Norman Abramson, the key individual responsible for its inven-tion, development, and initial deployments in a variety of appli-cations involving joint use of a given medium by potentiallyinterfering systems. He discusses the use of this protocol and itsvarious incarnations, continuing to this day, in such disparateapplications as wired Ethernet, satellite communications, andmobile wireless systems, among others. What is, in my mind, so

amazing about this scheme is its basic simplicity, yet profoundability to find application in such a widespread range of commu-nication systems.

The protocol, in its most concise form, says transmit at will. Ifinterference is detected, retransmit some random time later. Whatcan be simpler? Yet, despite its apparent simplicity, such a proto-col was not at all obvious at the time of its invention and initialdeployment. Enjoy reading this article!

—Mischa Schwartz

INTRODUCTION BY EDITOR

THE ALOHANET — SURFING FOR WIRELESS DATANORMAN ABRAMSON, UNIVERSITY OF HAWAII

LYT-HISTORY-Abramson-Dec 11/19/09 12:32 PM Page 21

22 IEEE Communications Magazine • December 2009

resource appeared necessary. The clas-sical spread spectrum techniques ofdirect sequence spreading or frequencyhopping seemed to be one way of shar-ing channel resources, and a spreadspectrum architecture was consideredfor the system. But the implementationof a spread spectrum format in eachuser terminal at a useful data rate andat a reasonable cost seemed aformidable task in 1969, given the exist-ing state of technology.

SYSTEM DESIGNThe first year of the project was spentin the study of the above questions, thespecification of a number of basic sys-tem parameters (e.g., frequency band,bandwidth, data rate), and the organi-zation of a laboratory for both the radioand digital circuit parts of the system.In addition to me, the faculty involvedin the project at that time includedThomas Gaarder, Franklin Kuo, ShuLin, Wesley Peterson, and Edward Wel-don. The key decision to use the directform of transmitting user informationin a single high-speed packet burst in ashared channel, now known as anALOHA channel, was made at a meet-ing of the project participants in 1969.The operation of the channel was notwell understood at that time, and thederivation of the throughput of anALOHA channel [4] was to come onlyseveral weeks after the decision to usethat channel. Nonetheless, the criterionof design simplicity which led to thatdecision should not be underestimatedas a reliable guide to designing datanetworks that work.

The remaining system design deci-sions were easier to make. In retro-spect, the only one that was crucialfrom the point of view of getting somesort of acceptable system operation wasthe choice of frequency band. That deci-sion was made on the basis of somehelpful advice from Dr. David Braver-man of Hughes Aircraft Company. Alo-haNet was assigned two 100 kHzbandwidth channels at 407.350 MHzand 413.475 MHz, and, after the usualunexpected software delays, the firstALOHA random access user terminalswent into operation in June 1971. For-matting of the ALOHA packets as wellas the retransmission protocols used inthe AlohaNet were implemented by aspecial-purpose piece of equipment(designed by Alan Okinaka and DavidWax) called the terminal control unit(TCU) (Fig. 1). A user terminal wasattached to the TCU by means of astandard RS232 interface, and the userwas connected to the central University

of Hawaii IBM computer at a data rateof 9600 bs anywhere within range of theradio system (about 100 km for the Alo-haNet).

The design of the original TCU wasconstrained by the need to provideextensive debugging tools in this firstrandom access packet interface and bythe technology of the times. Someunderstanding of the distance we havetraveled since then may be obtained byconsidering how one key decision in thedesign of the original network wasaffected by the cost of memory. It wasgenerally understood in the project thata full duplex mode of operation for theTCU would be desirable from the pointof view of both system protocols andsimpler hardware design. Full duplexoperation, however, required the use oftwo independent buffers to store thepackets flowing into and out of theTCU, and at that time the cost of mem-ory for an additional packet buffer of88 bytes was about $300. We decidedthat full duplex operation was notimportant enough to justify that expen-diture, since we were interested in anetwork that might eventually containhundreds of user stations. As a point ofcomparison, I recently purchased aUSB memory stick with 16 Gbytes stor-age for about $40 — a decrease in theprice per bit of storage by a factor ofmore than 1 billion in 40 years.

During 1971 and 1972, as additionalTCUs were built and a network cameinto existence, our understanding of thedistinction between point-to-point chan-nels and random access/broadcast chan-nels deepened. It became clear that thekey innovation of the AlohaNet was notjust the first use of wireless communica-tions for a data network, but at a deeper

level it was the use of a randomaccess/broadcast communications archi-tecture for network communications.Furthermore, we began to understandthat the advantages we were seeing inour network could be obtained in anynetwork with intermittent (bursty) trans-mitters as long as the bursty networktraffic flowed in a random access chan-nel. The imminent launch of domesticcommunication satellites by the UnitedStates (Westar I was launched in 1974)suggested the use of a random accessarchitecture for satellite networks, andwe turned our attention to the use ofALOHA channels in satellite systems atabout that time [5, 6].

SATELLITE NETWORKSAs one step in this process, it was decid-ed to obtain a satellite link to connectthe AlohaNet in Hawaii to the rapidlyexpanding ARPANet packet-switchednetwork on the U.S. mainland. TheAloha System funding at about thistime was transferred to the AdvancedResearch Projects Agency (ARPA),under the direction of Dr. LawrenceRoberts, and the connection to theARPANet seemed sensible for a varietyof reasons.

When the initial contacts with possi-ble carriers were made in order to leasea 56 kb/s satellite channel for the con-nection from the AlohaNet to a node ofthe ARPANET at NASA AmesResearch Center in California, it wassuggested that we lease 12 satellitevoice channels under existing tariffs forvoice grade channels, interface each ofthese channels with a modem operatingat 4800 bs, and multiplex the outputs toobtain the data rate desired. Such aprocedure had obvious economic bene-

HISTORY OF COMMUNICATIONS

Figure 1. ALOHA TCU 1971.

LYT-HISTORY-Abramson-Dec 11/19/09 12:32 PM Page 22

23

fits to those providing the 12 leasedsatellite channels, but it also requiredthat ARPA lease 12 satellite voice chan-nels to send only 56 kb/s. The leasing of12 voice channels seemed wasteful whenwe pointed out that the standard COM-SAT technique for transmitting a singlevoice channel at that time was to trans-form the analog voice signal into a 56kb/s pulse code modulation (PCM) datasignal at the Earth station. In otherwords, the existing tariff structure wouldhave had us convert our 56 kb/s datasignal into 12 4800 b/s signals for trans-mission to the satellite Earth station onthe north shore of Oahu, where thesechannels would have been transformedinto 12 56 kb/s signals, transmitted tothe other Earth station in California as12 56 kb/s signals, and transformedback into 12 4800 b/s signals for deliv-ery to the network. These signals wouldthen be multiplexed to form the 56 kb/sdata stream desired. It is a pleasure toreport that common sense finally pre-vailed, and after some discussions withRoberts, a new 56 kb/s tariff was intro-duced by COMSAT (which showed upin the COMSAT annual reports for sev-eral years as the “ARPA tariff”) allow-ing the lease of digitized 56 kb/schannels as data channels.

The ARPA/Aloha System link wasthe first commercial satellite link toemploy a single satellite voice channelpermitting a user to transmit 56 kb/s ofnetwork data. However, this channelwas not operated in a random accessmode by COMSAT, but in a conven-tional point-to-point mode. The firstnetwork to utilize random access packettransmission in a satellite channel wasput into operation in the Aloha Systemin 1973, using the ATS-1 satellite in anexperimental network that included theUniversity of Hawaii, NASA AmesResearch Center in California, the Uni-

versity of Alaska, Tohoku University inSendai, Japan, the University of Elec-tro-communications in Tokyo, and theUniversity of Sydney in Australia. Thisnetwork, called PacNet, operated at9600 b/s in an ALOHA channel usinglow-cost satellite Earth stations to showthe potential of data networks with largenumbers of small Earth stations [7].

Financial support for the researchon the Aloha System while the Alo-haNet and PacNet were being built, andthese two networks were connected tothe ARPANet in the continental Unit-ed States was provided by the Informa-tion Processing Techniques Office ofARPA, under the direction of LarryRoberts. In addition to this financialsupport of the Aloha System, Robertsalso contributed to the success of theproject in a way not ordinarily obtainedfrom funding agencies. In a real sense,Roberts acted as another member ofthe research staff of the project, con-tributing a number of major technicalresults (including the first derivation ofthe capacity of the slotted ALOHAchannel and the first analysis of captureeffect in ALOHA channels [8]).

Sometime in 1972, I was visitingRoberts’ office in Washington for dis-cussions dealing with both technical andadministrative matters in the Aloha Sys-tem when he was called out of his officefor a few minutes to handle a minoremergency. 1972 was a year of rapidgrowth for the ARPANet as the inter-face message processors (IMPs) thatdefined the nodes of the network wereinstalled in the first network locations.While waiting for Roberts’ return, Inoticed on the blackboard in his officea list of the locations where ARPA wasplanning to install IMPs during the nextsix-month period, together with theinstallation dates. Since I planned tobring up the question of installation of

an IMP at the Aloha System laboratoryin Hawaii to be used with the satellitechannel discussed above, I took thechalk and inserted “ the Aloha System”in his list and beside it placed the dateof December 17 (chosen more or less atrandom). After Roberts’ return, wecontinued our discussion but, becauseof the rather long agenda, we did notdiscuss the installation of an IMP inHawaii, and I forgot I had inserted aninstallation date of December 17 for usin the ARPA schedule on his black-board. I never did get the opportunityto discuss the installation of an IMP inour laboratory with Roberts. Instead,about two weeks before the December17 date, we received a phone call fromthe group charged with the responsibili-ty of installing the IMPs asking us toprepare a place for the equipment. OnDecember 17, 1972, an IMP connectingthe ALOHA system and AlohaNet tothe ARPANet by means of the firstsatellite channel in the ARPANet wasdelivered and installed.

By 1974, with the advent of the firstmicroprocessors by Intel, it was clearthat much of the logic of the TCU couldbe handled by a microprocessor, andseveral new terminal controllers, basedon the INTEL 8080, were designed byChristopher Harrison and put intooperation. Because the network proto-cols could now be implemented in soft-ware, and this software could besimultaneously modified for all operat-ing units by having the central stationbroadcast new protocol parameters oreven a completely new protocol, thesenew controllers were called pro-grammable control units (PCUs) (Fig.2). Once the basic protocols of theALOHA channel were analyzed anddemonstrated in the AlohaNet UHFradio channels, it was not long beforeother groups began to look into thepossibility of using ALOHA randomaccess in other media as well. One ofthe first and most successful of theseefforts was begun in the doctoral disser-tation of Robert Metcalfe at HarvardUniversity published as a project MACreport at the Massachusetts Institute ofTechnology (MIT) [9].

COMMERCIAL APPLICATIONSALOHA in Cable and Fixed WirelessNetworks: Bob Metcalfe was probablythe first person to appreciate the signif-icance of the distinction between thecommunications architecture of point-to-point circuit-switched and packet-switched data networks and the randomaccess/broadcast architecture of theAlohaNet. In his doctoral dissertation

HISTORY OF COMMUNICATIONS

IEEE Communications Magazine • December 2009

Figure 2. ALOHA PCU 1973.

LYT-HISTORY-Abramson-Dec 11/19/09 12:32 PM Page 23

24 IEEE Communications Magazine • December 2009

he coined the term “packet broadcast-ing” to emphasize this distinction,although the term “random access” istypically used to describe that commu-nications architecture today. When hisdissertation was complete, he spent sev-eral months with the Aloha System atthe University of Hawaii, working withRichard Binder who had developed andimplemented the ALOHA software andnetwork protocols [10]. After leavingHawaii, Metcalfe joined the Xerox PaloAlto Research Center, where his devel-opment of Ethernet (with David Boggs)demonstrated the effectiveness of a ran-dom access architecture on a cable-based medium. The first 3 Mb/sEthernet implemented at Xerox wasinitially called the Alto ALOHA Net-work [11]. Metcalfe then worked withDEC, Intel, and Xerox (the DIX group)to establish an open CSMA/CD stan-dard for a 10 Mb/s Ethernet and com-mercialized this standard after leavingXerox to found 3COM Corporation.

The UHF frequencies used in theALOHANet were obtained as a tempo-rary experimental license authorized bythe FCC as part of the ARPA programfunding our university-based research. Inthe early 1970s after we had built the net-work, experimented with a wide variety ofnetwork applications (including packetrepeaters, sensor networks, and internet-working), and developed a theoreticalfoundation for many of these applications[12], it was not at all clear how to con-vince the appropriate regulatory authori-ties to provide the frequencies necessaryfor the commercial development of wire-less ALOHA technology.

In the late 1970s Michael Marcus, astaff member at the U.S. Federal Com-munications Commission, proposed theestablishment of unlicensed spread spec-trum bands in the United States [13].After a good deal of discussion, on May9, 1985 the Commission establishedrules for the unlicensed use of spreadspectrum systems to operate in theindustrial, scientific, and medical (ISM)bands at 902–928, 2400–2483.5, and5725–5850 MHz on a noninterferingbasis to other authorized users of thesebands [14]. Other countries followed thelead of the U.S. FCC in authorizingthese bands for unlicensed transmissionof data. In 1997 the IEEE 802.11 stan-dard for WiFi was established usingCSMA/CA for wireless access [15] usingthe ISM bands. The issue of frequencyavailability was not an issue in the designof random access for cable-based datanetworks. The cable-based DOCSISstandard [15, 16] uses a slotted ALOHAchannel for both Internet and voice

packets. The initial random access pack-et in DOCSIS is used to set up a tempo-rary reservation channel as in GeneralPacket Radio Service (GPRS).

ALOHA in Satellite Networks: Thefirst use of ALOHA random access inan operational commercial satellite net-work took place shortly after the Ether-net effort in the COMSAT Marisatsystem in 1976 [17]. The request channelused to allocate voice and telex channelsin Marisat required some techniquecapable of sharing a single low-speed(4800 bs) channel among hundreds ofpossible users, and an unslotted ALOHAchannel was chosen. Much of the theo-retical foundation for both the cable-based and satellite-based systems wasdeveloped at about the same time byKleinrock, Tobagi, and Lam at UCLA[18–20]. Once these developments tookplace, it was astonishing how rapidly theacceptance of ALOHA-based mediumaccess control (MAC) protocols pro-ceeded. By 1977, at least one book hadbeen published that referred to theunslotted ALOHA channel as the “clas-sical ALOHA channel” [21, p. 587].

Today satellites in geostationary orbitprovide data transmission services toseveral million small Earth stationsaround the world. In almost all cases,whenever these earth stations are usedfor two-way data, network access is pro-vided by ALOHA or slotted ALOHA.Many of these networks are private net-works for companies in retailing, pub-lishing, and the financial industry or forgovernments; other satellite data net-works using ALOHA are public net-works providing general Internet accesssuch as HughesNet and WildBlue. BothHughesNet and WildBlue use slottedALOHA for traffic in the multiple accesschannel. More recently both ViaSat andCerona have used Spread ALOHA [22]for more advanced satellite networks.

ALOHA in Mobile Networks: It isconvenient to characterize the introduc-tion of ALOHA channels in wirelessmobile networks by recalling the stateof technology, and of telecommunica-tion regulation, during each of the pastfew decades. In the early 1970s bothALOHANet and PacNet went intooperation, providing a clear demonstra-tion of the practicality and efficiency ofthis type of wireless network architec-ture for user generated bursty data.Later in the 1970s Ethernet and theComsat Marisat network provided thefirst examples of ALOHA randomaccess in commercial networks.

Although the basic concepts of cellu-lar telephone technology had beenunder development at Bell Labs (and in

Europe and Japan as well) in the 1970s,it was not until 1982 that the U.S. FCCallocated frequencies for establishingsuch a system [23]. At the beginning ofthe 1980s there were limited optionsavailable for wireless transmission ofdata. Two of the most important net-works of that era were ARDIS andMobitex, both operating in the 800–900MHz range and both providing an ini-tial data rate of 8 kb/s. ARDIS was ajoint venture by IBM and Motorolabased on an unslotted ALOHA channeland initially utilized primarily by IBMservice technicians; Mobitex was devel-oped by Ericsson and Swedish Telecombased on a slotted ALOHA channelimplemented in Europe and (by RAMMobile Data) in the United States [24].But by the end of the 1980s it wasbecoming clear that effective develop-ment of wireless data networks shouldbe included within the burgeoningworldwide cellular mobile phenomenon.

The first commercial deployment ofa first-generation (1G)mobile networkin the United States was AMPS inChicago in 1983. AMPS was based onan analog architecture, but the requestchannel for AMPS was implemented asan unslotted ALOHA channel for initi-ating call connections [25]. Similar net-works were deployed in that time periodin both Europe and Japan. Rememberthat the set of applications suitable fora pervasive data network in the early1980s was limited. The IBM PC wasannounced in August 1981. In Decem-ber 1981 I purchased one of the firstIBM PC’s sold in Hawaii including anIBM modem board and some primitive(but open) software to allow me to con-nect to the Hawaiian telephone net-work at 300 b/s.

The 2G mobile networks that startedto appear in the 1990s (GSM in Europe,IS-54 and IS-95 in the United States)were digital networks which initiallyused ALOHA only for the requestchannel — slotted ALOHA in allcases.1 The use of ALOHA packets for

HISTORY OF COMMUNICATIONS

1 It should be noted that the spread spectrumIS-95 system employed what can be called aspread spectrum ALOHA channel in that bothspread spectrum and ALOHA properties wereused in the request channel for IS-95. Thisspread spectrum ALOHA channel should bedifferentiated from the Spread ALOHA channel[1, 22] in that IS-95 still uses different spreadingsequences to separate different requesting trans-mitters in the random access channel while theSpread ALOHA channel employs the samespreading sequence for all transmitters in therandom access channel.

LYT-HISTORY-Abramson-Dec 11/19/09 12:32 PM Page 24

25

requests and signaling applications with-in the mobile voice network suggestedthe use of ALOHA packets for usertraffic as well, so by the beginning ofthe new century user traffic was han-dled in a variety of 2G or 2.5G net-works by GPRS [23] using a slottedALOHA random access channel com-bined with a version of the ReservationALOHA scheme first analyzed by agroup at BBN [26].

The first 3G mobile networks wereintroduced in Japan in late 2001 and inKorea in early 2002. The first 3Gmobile networks in Europe began com-mercial operation in 2002, while servicein the United States began in late 2003.Both wideband code-division multipleaccess (W-CDMA) and cdma2000 pro-vide the higher bit rates and increasedemphasis on packet mode trafficrequired in the networks of today [23,27]. ALOHA provides a natural choiceof network architecture for this kind oftraffic, and the designs of both these3G networks reflect an increasing useof ALOHA random access for userpacket data as well as for signaling andcontrol purposes.

As this is written on the threshold ofthe 2010s, the shape of the implementa-tion of 4G mobile wireless is not clear,but it does seem clear that an increasedemphasis on packet data services,always connected response times, andhigher data rates will lead to increaseduse of ALOHA random access in 4Gwireless mobile networks.

ACKNOWLEDGMENTSSupport for Aloha System research wasprovided by the Information ProcessingTechnology Office of ARPA within theU.S. Department of Defense from 1968to 1974. After that date major supportwas provided by the U.S. National Sci-ence Foundation, IBM Research Labo-ratories, NEC, and Fujitsu. Additionalsupport was provided by members ofthe ALOHA System Industrial AffiliatesProgram at the University of Hawaii.

REFERENCES[1] N. Abramson, “Development of the Alo-

haNet,” IEEE Trans. Info. Theory, vol. IT-31,no. 2, Mar. 1985, pp. 119–23.

[2] J. Massey, “Guest Editorial,” IEEE Trans. Info.Theory, vol. IT-31, no. 2, Mar. 1985, pp.117–18.

[3] I. Dorros, “Retrospective — 25 Years Later,”IEEE Commun. Mag., June, 2009, pp. 14–20.

[4] N. Abramson, “The ALOHA System — Anoth-er Alternative for Computer Communica-tions,” Proc. 1970 Fall Joint Comp. Conf.,AFIPS Press, vol. 37, 1970, pp. 281–85.

[5]. N. Abramson, “Packet Switching with Satel-lites,” AFIPS Conf. Proc., Nat’l. Comp. Conf.,New York, June 1973, vol. 42, pp. 695–702.

[6] N. Abramson, “The Throughput of PacketBroadcasting Channels,” IEEE Trans. Commun.,vol. Com-25, no. 1, Jan. 1977, pp. 117–28.

[7] N. Abramson, “Satellite Data Networks forNational Development,” Telecommun. Policy,vol. 8, no. 1, Mar. 1984, pp. 15–28.

[8] L. G. Roberts, “ALOHA Packet System With andWithout Slots and Capture,” Comp. Commun.Rev., vol. 5, no. 2, Apr. 1975, pp. 28–42.

[9] R. M. Metcalfe, “Packet Communication,”MIT, Cambridge, MA, Report MAC TR-114,July 1973.

[10] R. Binder et al., “ALOHA Packet Broadcast-ing — A Retrospect,” AFIPS Conf. Proc.,1975 Nat’l. Comp. Conf., AFIPS Press, Mont-vale, NJ, May 19–22, 1975.

[11] R. Metcalfe, Xerox PARC memo to AltoAloha Distribution on Ether Acquisition, May22, 1973.

[12] N. Abramson, “The ALOHA System FinalTechnical Report,” ARPA, contract no. NAS2-6700, Oct. 11, 1974

[13] M. Marcus, “Wi-Fi and Bluetooth — ThePath from Carter and Reagan-Era Faith inDeregulation to Widespread ProductsImpacting Our World,” keynote talk at theInfo. Economy Project Conf. on the Genesisof Unlicensed Wireless Policy, George MasonUniv., Apr. 4, 2008; http://www.iep.gmu.edu/UnlicensedWireless.php

[14] K. J. Negus, and A. Petrick, “History ofWireless Local Area Networks (WLANs) in theUnlicensed Bands,” George Mason Univ. LawSchool Conf., Info. Economy Project, Arling-ton, VA, Apr. 4, 2008.

[15] A. Tanenbaum, Computer Networks, 4thed., Prentice Hall, 2003.

[16] J. Martin, “The Impact of the DOCSIS1.1/2.0 MAC Protocol on TCP,” 2nd IEEEConf. Consumer Commun. and Networking,2005, pp. 302–06.

[17] D. W. Lipke et al., “MARISAT — a MaritimeSatellite Communications System,” COMSATTech. Rev., vol. 7, no. 2, Fall, 1977.

[18] L. Kleinrock, Queuing Systems, Volume II:Computer Applications, Wiley, 1976, Secs.5.11 and 5.12.

[19] L. Kleinrock and F. A. Tobagi, “PacketSwitching in Radio Channels: Part I — Carri-er Sense Multiple-Access Modes and TheirThroughput-Delay Characteristics,” IEEETrans. Commun., vol. COM-23, Dec. 1975,pp. 1400–16.

[20] L. Kleinrock and S. S. Lam, “Packet Switch-ing in a Multi-Access Broadcast Channel:Performance Evaluation,” IEEE Trans. Com-mun., vol. COM-23, Apr. 1975, pp. 410–23.

[21] J. Martin, Future Developments in Telecom-munications, 2nd ed., Prentice-Hall, 1977.

[22] N. Abramson, “VSAT Data Networks,” Proc.IEEE, Special Issue on Satellite Communications,vol. 78, no. 7, July, 1990, pp. 1267–74.

[23] M. Schwartz, Mobile Wireless Communica-tions, Cambridge Univ. Press, 2005.

[24] K. Pahlavan and A. Levesque, “WirelessData Communications,” Proc. IEEE, vol. 82,no. 9, Sept. 1994.

[25] B. Stavenow, “Throughput-Delay Character-istics and Stability Considerations of theAccess Channel in a Mobile Telephone Sys-tem,” Proc. 1984 ACM SIGMETRICS Conf.Measurement and Modeling of Comp. Sys.,pp. 105–12.

[26] W. Crowther et al., “A System for BroadcastCommunication: Reservation — ALOHA,”Proc. 6th Hawaii Int’l. Conf. Sys. Sci., Hon-olulu, HI, Jan. 1973, pp. 371–74.

[27] M. Karim and M. Sarraf, W-CDMA andcdma2000 for 3G Mobile Networks ,McGraw-Hill, 2002.

BIOGRAPHYNORMAN ABRAMSON [F] received an A.B. in physicsfrom Harvard College in 1953, an M.A. inphysics from UCLA in 1955, and a Ph.D. in elec-trical engineering from Stanford in 1958. Hewas an assistant professor and associate profes-sor of electrical engineering at Stanford from1958 to 1965. From 1967 to 1995 he was pro-fessor of electrical engineering, professor ofinformation and computer science, chairman ofthe Department of Information and ComputerScience, and director of the ALOHA System atthe University of Hawaii. He is now professoremeritus of electrical engineering at the Univer-sity of Hawaii. He has held visiting appoint-ments at the University of California at Berkeley,Harvard, and MIT (1965, 1966, 1980). He is afellow of the IEC. He is also the recipient of theIEEE Sixth Region Achievement Award, the PTC20th Anniversary Award, the IEEE Koji KobayashiComputers and Communications Award, theIEEE Information Theory Society Golden JubileeAward for Technical Innovation, the EduardRhein Foundation Technology Award, and theIEEE Alexander Graham Bell Medal.

HISTORY OF COMMUNICATIONS

IEEE Communications Magazine • December 2009

LYT-HISTORY-Abramson-Dec 11/19/09 12:32 PM Page 25