The Wireless Revolution Rapid multinational progress will soon make global wireless communication a ubiquitous reality.

Theodore 5. Rappaport

Theodore Rappaport is U

faculty member of the Virginia Polytechnic Institute, the State Uniwrsit) in

B l a c k s b q , virgmia. and at tlie Mobile and Portable Radio Research Group (.MPRG).

ver the past three years, the inter- est in wireless communications has been nothing less than spectacular. Cellular radio systems around the world have been enjoying 33 percent to 50percentgrowth rates. Manypag-

ing services have been gaining customers at a rate of 30 percent to 70 percent or more per year, and within the last two years there has been intense corporate research and development aimed at com- mercializing new wireless communication services called PCS (personal communications services). Meanwhile, new digital cellular technologies have been installed in Europe, and developing nations are beginning to install cellular infrastructure.

The first wide-scale adoption of a wireless per- sonal communications system was in citizens band (CB) radio during thelate 1960s and early 1970s. Although it was a victim of its own success due to a rapid and uncontrolled saturation of the radio spec- trum, and suffered severely from lack of traffic man- agement and services, the staggering acceptance of CB radio was a clear indication that consumers wanted an inexpensive, portable means of com- municating. A more modern wireless device that has enjoyed widespread popularity is the cordless telephone. By the time you read this article, it is like- ly that more than 65 million cordless telephones will have been sold in the United States alone (although it is estimated that almost halfof the cord- less phones sold have been discarded or are not used). The convenience of portable telecommunications offered by a device a s simple as t he cordless phone is clearly in demand. Although the design solu- tions for a ubiquitous wireless communications network will be tremendously complex, all mar- ket indications show that consumers wish to have a small device, similar in size to a cordless phone, that would allow them to make and receive phone calls wherever they are. The idea behind wireless personal communications networks (PCNs) is to make communications truly personal, so that any- one can place a call to anyone else. Much like the stereo Walkman'" or thelaptopcomputer, PCNwill permit truly ubiquitous access regardless of the loca- tion of the user at the time of access. Today, the terms PCN and PCS often a re used interchangeably. PCN refers to a concept by which a person can use a single communicator anywhere in a large geographic area. PCS refers to a service that may not embody all of the PCN concepts, but is more per- sonalized (i.e., lightweight terminal, better per-

formance, more flexibility, user options, etc.) than a present-day cellular telephone.

Current Demand The premise that wireless personal communi-

cations is emerging as a key, wide-sweeping tech- nology that will dramatically impact our society is supported in numerous sources, including world- wide t r ade jou rna l s and government agency reports . As a n example, in t h e United States there were more than 6.3 million cellular tele- phone users as of September 1991 [18]. This com- pares with 25,000 users in 1984, and 2.5 million U.S. users in la te 1989 [ l ] . It is clear t o most industry experts that t he Cellular Telephone Industry Association's (CTIA) 1989 projections of 10 million United States cellular users by 1995 will be exceeded in late 1992, and cellular radio carri- ers are enjoying exponential increases in service sub- scriptions. This demand for mobileiportable telecommunications is a worldwide trend and is par- ticularly acute in Europe. For example, cellular tele- phone in Sweden already enjoys a 6.6 percent adult market penetration [18], and this figure has been increasing by more than 0.1 percent per month. Finland, Norway, and France have been experiencing similar growth rates. In Hong Kong, more than 50percent of the adult population own or have oper- ated cellular telephones. Although no spectrum has yet been allocated to emerging PCS in the United States, industry analysts already are pro- jecting annual U.S. revenues to be between $33 billion and $55 billion by the year 2000 [17].

In the United Kingdom, viewed by many as the leading country for PCS initiatives, three major companies are investing hundreds of millions of dollars t o install an infrastructure that may eventually allow citizens to use small 10 mW portable terminals to place and receive calls in populated areas throughout the country. At worst, PCS will pro- vide relief for users who operate in congested 900 MHz United Kingdom cellular markets. At best PCS will offer customers the ability to use a single wireless communications unit for home, office, or automobile, thereby obviating the need for a tra- ditional wired phone to the home. Two of the three companies involved, Microtel and Unitel, are actually consortia consisting of many leading telecommunications companies. The third United Kingdom PCN service provider, Mercury, is the lead- ing non-wireline service provider of England's

52 0163-68O4/91/$Ol .OO 199 1' IEEE IEEE Communications Magazine November 1991

present analog cellular system. These PCN com- panies are expected to become some of the biggest advertisers in the United Kingdom throughout this decade, as they strive to pioneer and market the revolutionary PCN service [19].

The pervasiveness of PCN will be made available by an immense infrastructure of low-power, suitcase- sized base stations which will provide portable subscribers with wireless access to the local telephone loop in populated areas. PCN hopes to be able to offer wireline communications quality using radio as a transmission medium. Base stationswill be placed on lamp posts, roofs, and ceilings of buildings and concourses, and in other locationswhere people con- gregate. Backbone links that connect PCN base stations to other PCN base stations and to the public switched telephone network (PSTN) will be supplied by one of two methods: either via the existing telephone wire or fiber plant, or via line- of-sight microwave links. Of course, the wide- scale deployment of such an extensive high-grade wireless telephone system will require engineer- ing tools and techniques and antenna designs that allow rapid and accurate propagation prediction and system design. The radio coverage of each base station will intentionally be limited by low transmitter power, so that the same frequencies can be re- used many times within a few city blocks. Depend- ingon regulatory decisions by the British Post Office, PCN could compete directly with wired residen- tial telephone service in the United Kingdom. Thus, it is conceivable that the customers could bypass the telephone company and use a single PCN ter- minal for communications at home or in the office.

In the United States, more than 65 experimen- tal licenses have been issued within the last year to regional bell operating companies (RBOCs), non- wireline cellular service providers, manufacturers, cable companies, and new start-up companies hoping to pioneer PCN service. Experimental FCC licenses tha t allow limited use of radio transmissions for portable FAX service, wireless spread spectrum PABX systems, and microcellu- lar and personal communication services have been granted and trials are in various phases.

To increase the competition and development time of new wireless technologies and services, the FCC created a special incentive, called the pioneer’spreference, that offers the exclusive use of reallocated radio spectrum to companies that first demonstrate new technologies or concepts for PCN or other new wireless services. Many of the UnitedStatesPCNexperimenters havefiledpeti- tions for rulemaking to the FCC requesting new ded- icated or shared spectrum for PCN services, and hope to secure an advantageous position through pioneer’s preference. Industry and government experts view the demand for wireless personal communi- cations to be sogreat that the FCC has indicated that underutilized portions of existing radio spectrum could be subject to reallocation in order to accom- modate consumer demand.

Several good tutorial articles that describe the impetus and technological challenges behind cel- lular radio and PCN can be found in the August 1990 and September 1990 issues of the IEEE Commu- nications Magazine. The different second generation standards that have emerged around the world are described in several invited papers in the May 1991 Issue of the IEEE Transactions on Vehicular

Technology (VT) which features digital cellular tech- nologies. As indicated in the VT special issue, it is conceivable that 50 million to 75 million sub- scribers could be using wireless systems for vari- ous types of personal communications by the mid-1990s. This is corroborated by a recent Mor- gan Stanley report that predicts cellular and PCN systems will achieve at least 12 percent market penetration in many developed countries by the end of this century.

Recent Events in Wireless InlargeUnitedStatesmarketssuch asLosAnge-

les and New York, where hundreds of thousands of users can access the cellular radio spectrum, the 832 cellular analog-FMvoice channels are unable to accommodate the number of users, and meth- ods to improve capacity and cellular system design a re desperately needed. For those not familiarwith cellular radio, in each market (i.e., city) there are two cellular service providers: the radio common carrier (non-wireline provider), called the A channel provider, and the wireline, o r B channel , provider . Each of t he two service providers are allocated 416 duplexvoice channels in a 25 MHz spectrum allocation. Each voice chan- nel is comprised of a 30 kHz base-to-mobile link and a 30 kHz mobile-to-base link.

Over the past four years, numerous standards have been proposed for digital cellular radio communication interfaces throughout the world. Digital modulation offers improved spectral effi- ciency and simultaneouslyoffers better speech intel- ligibility for a given carrier-to-interference ratio (CII). More importantly, digital modulation accommo- dates powerful digital speech coding techniques that further reduce the spectrum occupancy of voice users. Using digital transmission formats , service providers will be able to offer customers addi- tional features, such as dynamically allocated data services, encryption, etc.

In early 1990, the CTIA and the Telecommu- nications Industry Association (TIA) approved Inter- im Standard 54, which specifies a dual-mode cellular radio transceiver that uses both the analog FM (the present-day United States Advanced Mobile Phone System, or AMPS, standard) and a linearized n/4 Differential Quadrature Phase Shift Keying mod- ulation format with a Code Excited Linear Pre- dictive (CELP) speech coder (called the U S . Digital Cellular, or USDC standard) [20].

T h e U S D C standard offers roughly th ree times the capacity improvement over AMPS bypro- viding three voice channels that are time-division multiplexed (TDMA) on a single AMPS 30 kHz FM voice link. With further speech coding improvements, six times capacity is likely to be achieved by 1994. The dual mode equipment will allow a graceful tran- sition from analog FM to digital cellular, since cellular operators will be able to change out ana- log channels t o digital channels depending on their geographic capacity demands. In this man- ner, customers with analog phones will be assured service in any market until some announced time in the future.

A rural cellular carrier that does not suffer great capacity demand would likely stay with AMPS for as long as possible, probably several years (until the mid-l990s), while a metropolitan oper- ator would likely change out AMPS for USDC more

- In l a s e United States markets, the 832 cellular ana log- FA4 channels are unable to accommo- date the number of users.

IEEE Communications Magazine November 1991 61

- In contrast with USDC, the Euro- pean di@tal cellular sys-

tem (Group Special Mobile, or GSM) was developed for a new spectrum allocation in the 900 MHz band.

rapidly (within two years). It is interesting (and also a bit troubling to those who invested agreat deal of time and money into developing t h e IS-54 standard and who plan to abide by it) that since the introduction of the USDCstandard IS-54, numer- ous vendors have introduced their own compet- ing standards that are incompatible with IS-54[20]. Now virtually all major cellular radio service providers are conducting field trials to evaluate new, competing standards, so it isunclear if US. digital cellular radio systems will be completely compatible through- out the country, even though this was the major goal of USDC.

In contrast to USDC, the European digital cellular system (called Group Special Mobile, or GSM) was developed for a brand new spectrum allo- cation in the 900 MHz band. That is, GSM was devel- oped to ensure that a single access and equipment standard would be used throughout the Euro- pean continent.

Unlike USDC, which hoped to make a seam- less transition of America’s analog system to digi- tal, GSM was developed from scratch and made pan-European compatibility its primary objective. Today, in most European countries, the cellular sys- tems and standards are unique to the individualcoun- tries. In fact, at the time of this writing, a cellular phone that works in France cannot be used in England. GSMoperates in new spectrum at 900 MHz dedicated for use throughout Europe. Details of the GSM specification are given in [21,25] and equip- ment will be available to the pan-European com- munity by the time this article is published.

GSM is the world’s first TDMA cellular sys- tem standard, and uses a constant envelope mod- ulation format to gain power efficiency (constant envelope modulation allows more efficient class- C power amplifiers to be used) over spectral effi- ciency (constant envelope modulat ion has a larger bits per hertz of R F occupancy than does linear modulation). GSM uses an equalizer and slow frequency hopping to overcome multipath effects that cause intersymbol interference and, thus, irreducible bit error rates. As the world’s first dig- ital cellular radio standard that has been adopted by a large market, GSM is viewed as a front-run- ner for early implementation of PCN throughout the world. Depending on the success of United King- dom 1800 MHz PCN initiatives, GSM equipment could be implemented on a worldwide scale if spec- trum allocations are made available in other countries.

In February 1990, just after CTIA adopted IS-54, QUALCOMM, Inc. proposed to CTIA the use of spread spectrum and sophisticated base station signal processing to offer capacity improvements ten t imes g rea t e r t han AMPS (see t h e pape r by Gilhousen, et. al. in the May 1991 IEEE Trans. VT special issue on Digital Cellular Technolo- gies). Major cellular radio service providers and man- ufacturers have steadily supported QUALCOMM as they develop prototype spread spectrum cellu- lar phones that will be ready by mid-1992 [22]. Work conducted two decades ago [26], and more recently [29], confirms that spread-spectrum holds great promise for accommodating huge capac- itywith simple frequency management, although the capacity is highlydependent on radio path losswith- in the service area. Of course, higher capacity means higher revenues and less churn (loss of cus- tomers) for cellular service providers, so the QUAL-

COMM proposal received immediate attention. Today, many United States companies (e.g., American Personal Communications, PCN Amer- ica, and Omnipoint Data) are lookingat theviability of overlaying TDMA, FDMA, and spread spectrum mobile radio systems on existing point-to-point microwave links, used by utility companies, banks, and public safety offices, in the 1850 MHz - 1990 MHz band. There are many others who are aggressive- ly conducting research and experimentation to deter- mine overlay possibilities, and their experimental results may be obtained a t t he FCC office in Washington, D.C. Recent FCC experiment reports filed by American Personal Communications [36] and Telesis Technology Laboratories [37] indi- cate that even in the largest urban United States mar- kets, it may be possible to break the existing 1850 MHz to 1990 MHz band into 5 MHz segments which could be used simultaneously by line-of-sight microwave users and PCS providers.

The concept behind overlay is that low-power PCN service could be offered directly on top of the existing terrestrial microwave band. The rea- soning is that since existing microwave links use direc- tional antennas, and thus do not radiate over a large geographical area, it could be possible to inter- twine PCS frequencies over a geographical area such that existing microwave links are not interferedwith by overlaid PCS. Recent experiments and analysis seem to suggest that existing microwave links have not been engineered throughout large markets to take full advantage of the spectrum.

References [23] and [24] discuss the concept behind overlay, and some techniques that could be used to minimize interference between exist- ing and new users. Recent propagation measure- ments to test the levels of interference caused by PCN subscriber units to existing fixed microwave users were presented in [24] and more recently have been detailed in FCC reports [37, 381 as part of experimental PCN licenses.

Presently, major telecommunication compa- nies are researching the effectiveness of overlay systems from a capacity standpoint. If a sufficient grade of service could be offered through spec- t rum sharing between new wireless service providers and existing line-of-sight microwave licensees, then the FCC would be able to instant- ly accommodate market demand for wireless cel- Iularb’CNwithout a major new spectrum allocation. The fear , of course, is that there could be an unsatisfactory degradation of service to both the existing microwave users and the overlayed PCN users. The overlay concept is being tested bynumer- ous companies in the United States under the FCC experimental license program. At present the FCC has not made a decision on their position on the matter.

The cable industry also has been watching the sudden growth in cellularPCN throughout the world. Cable companies have amassive RFnetworkinstalled throughout populated regions, and it is obvious that telecommunication services could be offered over the existing cable plant. In fact, PCN base stations could be installed easily in residential areas by splicing existing cable runs and tapping on base stations mounted on lamp posts or inside buildings. Indeed, several United States cable companies are conducting experiments to determine the feasibility of PCS using the cable plant. When

62 IEEE Communications Magazine November 1991

one realizes that United States cable operators already have spectrum allocated for microwave feeds in the 2 GHz and 13 GHz bands, the possibility of utilizing the cable spectrum for PCN services instead of point-to-point feeders presents a lucrative new business opportunity for the cable industry.

For local loop, or premises applications, which merge voice and da ta , Bellcore [15] and the European Technological Standards Institute (ETSI) have proposed digital TDMA standards that offer between 400 kilobits per second (kbis) and 1100 kbis data rates in office and residential environments. Although the widespread deployment of such standards likely will rely on new dedicated spectrum for the services, significant engineering manpow- er has been devoted to develop the standards, and a great deal can be learned from the research.

Bellcore’s system exploits the slow time-vary- ing nature of indoor channels and uses antenna polar- ization diversity to improve link performance between a base station (port) and mobile terminal (portable). Bellcore’s proposal limits the data rate to 450 kbis based on an extensive measurement program that determined worst case multipath channels in a large number of buildings, houses, and cities. Bellcore’s approach minimizes the complexity and battery drain of the portable, since power- hungry adaptive equalizers can be avoided. Also, Bellcore’s system uses a novel oversampling demod- ulation technique that allows areceiver tolockcoher- entlyonto the incoming modulationwith only a couple of bits of overhead. Reference [15] provides addi- tional details about the Bellcore system, and [21, 25,34, and 351 provide additional information about ETSI and the DECT standard.

Funding in Wireless The burgeoning wireless communications indus-

try has created an interesting problem for wire- less manufacturers and service providers throughout the world. Because the field of cellular radio andper- sonal communications is changing so rapidly, and since the field involves system concepts seldom taught at universities, wireless companies are having dif- ficulty finding entry-level graduates with suffi- cient education to make an immediate contribution in research or design. In particular, a large num- ber of universities presently do not offer under- graduate or graduate courses on the topics of mobile radio propagation or wireless communication sys- tem design. Consequently, recruiters are forced to raid competing companies for more senior personnel, and resign themselves to spending six months to two years to train new graduates in the art and science of mobile and portable radio.

In an informal survey of some of the largest cellular radio companies in the United States, the author has learned that engineers with knowl- edge of mobile radio communications change jobs often, and are in extremely high demand. Particularly in the past two years, since cellular radio has enjoyed a50percent annual growth rate and new digital systems have been proposed and tested, engi- neers with experience in cell-site design or computer simulation expertise in mobile radio propagation, traffic modeling, antenna design, and digital sig- nal processing have been highly sought after by the wireless industry.

The National Aeronautics and Space Admin-

istration (NASA), through its research programs in space communications and mobile satellite communications experiments (MSAT-X), and more recently through its Advanced Communications Technology Satellite (ACTS) program, has provided the bulk of United States federal research fund- ing in the mobile communications area. Although mobile satellite systems do not promise nearly the same capacity (users per MHz) as do land- based microcellular and PCS systems, there is remark- able commonality between the technologies used in satellite and cellular mobile systems. Factors such as modulation, coding, multiple access, antenna design, network control, and propagation are fun- damental to both types of wireless communica- tion systems, as are the techniques used to design and analyze such systems.

Through the NASA MSAT program alone, at least 11 United States universities and numerous companies have been able to conduct multiple- year research projects that have produced both howl- edgeable graduates and innovative technologies. Today, many graduates of NASA-funded com- munications research programs are working in the satellite or cellular industries. Examples of two different satellite systems for personal com- munications are described in [38] and [39].

Until recently the National Science Founda- tion has shown little interest in funding experimental or theoretical work applied to wireless communi- cations. It is ironic that an NSF-sponsored proj- ect at Purdue University in the mid-1970s provided the first study of aspread spectrum cellular radio sys- tem [26], a study that created intense interest in spread spectrum access approaches for cellular radio communicationswhich now (17years later) are being extensively commercialized by numerous compa- nies. This summer, however, NSF awarded Rut- gers Universitywith a cooperative UniversityDndustry Center. The Rutgers Wireless Information Network (WINLAB), founded by Professor David Goodman in 1988-1989, isworking on network and access solu- tions for third generation wireless personal com- munication systems, and was the first United States academic laboratory formed for wireless per- sonal communications education and research. Steadily, more universities are recruiting faculty interested in the wireless communications field.

The Defense Advanced Research Projects Agen- cy (DARPA) has funded projects to develop small, low-powered, wireless devices, and is sup- porting research that will lead to technology development for rapidly deployable local area communication systems. New design and fabrica- tion technologies range from advanced silicon integrated circuits, to enhanced speech coding algo- rithms and circuit fabrication, to software tools for propagation prediction and installation. The per- sonnel development, knowledge base, resulting tech- nologies, and system design tools from these projects will not only improve the United States militarycapa- bility, but also has relevance to the U.S. con- sumer wireless personal communications industry.

Virginia Tech’s Mobile and Portable Radio Research Group (MPRG) is a new group within the university’s Bradley Department of Electrical Engineering. Founded in the spring of 1990, it iscon- ducting basic and applied research in the areas of radio propagation measurement and prediction, communication system design using measure-

- NASA has provided the bulk of United States federal research finding in the mobile communica- tions area

IEEE Communications Magazine November1991 63

- In Canada, numerous initiatives are under way to encourage academic participation in research for wireless persona 1 communica- tions.

ment-based propagation models, and simulation of various digital modulation, diversity, and radio equalizer techniques. The group’s mission is to help the United States wireless industry through the development of analysis tools and computing techniques for emerging wireless personal com- munication systems, while providing quality research opportunities for graduate and undergraduate students.

M P R G also is providing opportunities for technical interchange. An E E graduate student lecture series in the fall of 1989 featured key researchers from the cellular radio industry, and the First VirginiaTech Symposium on Wireless Personal Communicationswas held June 1991 in Blacksburg, Virginia. The symposium featured 20 invited speak- ers and panel discussions by industry experts over a three-day period. It was attended by 175 people from 22 states and nine countries. The sympo- sium, now scheduled as an annual June event on the Virginia Tech campus, provided an opportu- nity for students and engineers to learn both fun- damentals and trends in wireless communications. The necessity for academic research groups like M P R G becomes clear when o n e realizes the enormous, yet sudden, activity in the wireless field, and the growing demand for young gradu- ates who can make immediate contributions to a dynamic field.

I t is worth noting that in contrast with the United States government agencies, the Euro- pean, Canadian, and Japanese governments have made substantial funding commitments to research laboratories and university programs focusing on wireless personal communications and related tech- nologies. Fo r example, in Europe, the R A C E program (Research and Development in Advanced Communications in Europe) has committed over $100 million per year to the European communi- ty during the period 1987-1992 [27].

A significant portion of those funds has been spent on collaborative industryluniversity research in wire- less communications, with the goal of concurrent- ly expanding the knowledge base and pool of technical experts. RACE appears to be yielding large divi- dends. The European community is widely recog- nized as the world leader in creating and accepting new digital cellular radio system techniques (many European countries presently enjoy more than a 5 percent cellular market penetrat ion, comparedwith approximately3 percent in the Unit- ed States), and it is where the concept of personal communications was first put into wide scale practice.

Casual conversationswith researchers across the world indicate that more than 50 European Ph.D. students are tackling dissertations dealingwith anten- nas and propagation for emerging wireless per- sonal communication systems. This does not include other areas of PCN, such as modulation, coding, diversity, and system design.

In Canada, numerous initiatives are under way to encourage academic participation in research for wireless personal communications. The Telecom- munications Research Institute of Ontario (TRIO) program is enhancing the technological competi- tiveness of Canadian industry through university/industry partnerships in telecommuni- cations research. TRIO has grown steadily since it was founded in 1987.

This year, the Ontario province and Canadian industry will provide more than $6 million for university communications research in Ontario. A significant portion of those funds is being used for mobile and satellite systems research for wire- less personal communications. The mobile and satel- litesystemsprogram supportsmore than 50graduate students at five universities, and involves approxi- mately 20 Canadian communications companies. Other Canadian provinces also are providing research support for regional universities active in wireless personal communications.

On a federal level, The Canadian Institute for Telecommunications Research (CITR) was estab- lished in 1989, and is providing research grants to universities throughout Canada. CITR’s budget is more than $4 million annually (overhead is waived on CITR funding) and is focusing on two major areas: broadband and wireless communications. Examples of CITR projects in the antennas and propaga- tion area include propagation and diversity tech- niques for indoor wireless communications at millimeterwaves, fading issues for 20 GHz to 30 GHz personal satellite communications systems, and new cellular system design techniques. Universities involved with CITR wireless research include Carleton University. Concordia, Ecole Polytechnic, Lava1 University, McGill, Queens, Simon Frazier University, the University of British Columbia, the University of Calgary, t he University of Ottawa, the University of Toronto, the University of Victoria, the University of Waterloo, and the Uni- versity of Western Ontario.

Japanese research programs in wireless personal communications abound, and there are numerous universities active in the area that are making fundamental contributions. As an example, Kyoto University has been a major contributor in the wireless communications field, and has an active graduate program in propagation and communi- cation system design. Federal funding for aca- demic research in Japan often involves interaction with industrial laboratories [27]. It seems clear that most federal governments view the tremendous impact that wireless communications will make on the world’s economy as agolden opportunity, and are hoping their investments will result in new technologies and abody of experts who can engineer and develop new wireless personal communica- tion systems.

Early efforts to obtain government funding from the National Science Foundat ion failed, but industry response has been excellent and after one year, 15 major wireless companies (including many regional Bell Operating Companies, major radio manufacturers, and computer manufacturers), DARPA, and the FBI have provided a funding base of over $1 million. Collaborative research projects with Purdue University, Northeastern Uni- versity, and the University of California at San Diego have provided cross-fertilization of ideas and knowl- edge, and are helping to make the United States approach to wireless communications research more synergistic.

At MPRG, students receive an educational expe- rience that provides them with a solid under- standing of the theory and practice of mobile radio communications and emerging personal com- munication systems. At the same time, they are tasked with developing andvalidating analyses, models, and

64 IEEE Communications Magazine November 1991

research tools for the wireless industry. These tools help transfer knowledge from the MPRG laboratory into industry and academia.

In 1990, the first year of MPRG, five M.S.E.E. studentsgraduatedwith expertise in RFfilter design, indoor radio propagation measurement and pre- diction, adaptive noise-cancellation techniques, and urban radio propagation prediction. Presently, there are 18 graduate and six undergraduate MPRG students pursuingdegreeswith an emphasisonwire- less communications. As a result of student the- ses, several analysis and simulation software tools have been developed for internal use, and are also being used for research and development by a number of companies and universities. An elec- trical engineering graduate course dedicated to the topics of cellular radio and personal commu- nications was offered in the spring of 1991, and enjoyed an enrollment of 34 students, making it the most popular graduate course in the EE cur- riculum at Virginia Tech that term. Senior elec- tive courses on r ad io wave propagat ion and satellite communications also are popular EE courses at Virginia Tech.

Although the research and educational mis- sion of MPRG concerns itself with more than just propagation, the group received its first research contracts in the area of propagation measure- ment and prediction, and continues to maintain an active program in this area. There is extreme inter- est in and demand for propagation measurements and models for the proper design of emerging wireless services such as PCN, and more powerful site-specific channel modeling techniques and tools are needed.

The United States wireless industry, however, often finds conducting their own measurements and propagation research to be a time-consuming and expensive task, and many industrial players view it as an expensive luxury that involves high-priced consultants. MPRG has emerged as a sensible and cost-effective alternative, since it has an established equipment arsenal and provides research expertise in the area of propagation measure- ment and prediction. By pooling resources in an industrial affiliates program, MPRG is generating useful tools and basic propagation models that can be shared by all affiliate members, and the result- ing value of the research is much greater than the cost to an individual member. All results are pub- lished so the entire research community benefits from the knowledge base.

Here, we briefly present some of the basic param- eters used to describe multipath channel charac- teristics, and show results f rom propagat ion measurements in urban, microcellular, and indoor channels. These measurements involve wideband characterizations, where avery short RFburst is sent and the echoes are received, as well as narrow- band (CW) characterizations that measured sig- nal fading over large temporal and spatial spans. As is subsequently shown, antenna experiments have revealed that polarization can dramatically reduce the multipath time delay spread and the fluctua- tion of delay spread in mobile channels. Resultspre- sented in this article are not meant to be interpreted as definitive work in mobile radio propagation. On the contrary, it is hoped that this articlewill spark interest and ongoing discussions in propagation mod- eling and prediction for personal communications.

Propagation Measurement A large part of MPRG research has dealt with

measuring, and then statistically modeling, the path loss and time dispersion of multipath radio chan- nels. Measurements and models have been made in many different environments: traditional urban cellular radio channels with base station antenna heights exceeding many tens of meters [l, 21; urban cellular radio channels with lower antenna heights, on the order of 15 to 20 meters [2]; and in-building channels within sports arenas, facto- ries, and office buildings [3, 41, and open plan office buildings [5]. Also, impulsive noise mea- surements have been made inside many types of buildings at three bands between 900 MHz and 4.0 GHz [28], includingtwoof the license-free Indus- trial, Scientific, and Medical (ISM) bands.

For typical urban mobile radio channels, it has been reported that the coherence bandwidth (the bandwidth over which the received signal strength willlikelybewithin90percent of anyother frequency from the same source) is between 10 kHz to 500 kHz [ 6 ] . Consequently, to sufficiently resolve multipath components (in the time domain) that cause frequency selective fading, a channel sounder for urban channels should possess an R F bandwidth several times larger than the maxi- mum coherence bandwidth. This thinking led us to use a500nsprobingpulse (4MHzRFbandwidth) in [I, 21. For indoor measurements, we have used probes that have durations on the order of 5 ns, there- by providing measurements that span over 400 MHz and which resolve individual multipath echoes to about 1.6 m separation distance. (The baseband pulse is DSB modulated so there is a bandwidth expansion factor of two at RF. Thus, the baseband complex envelope response has a 200 MHz resolution bandwidth). Time domain techniques rather than swept frequency techniques are used since it is easy to identify the location and intensity of reflecting objects in the channel.

This information is vital for successful devel- opment of site-specific propagation models that can recreate the channel impulse response. Our mea- surement systems are easy to assemble, test, and deploy, and provide instantaneous channel mea- surementswith excellent time delay resolution. Post- detection integration is employed to improve the signal to noise ratio of the measured power delay profiles. Our approach, though, requires more peak power than the direct sequence systems used in [7,8, and 91 for a specifiedcoverage distance. Broad- band discone, yagi, and helical antennas have been used to ensure no pulse spreading is attribut- ed to impedance mismatch.

Important parameters that indicate multipath dispersion in mobile radio channels are illustrat- ed in Fig. 1. Therms delay spread (oT) measures how spread of the channel power delay profile about the centroid, and the excess delay spread (XdB) indi- cates the maximum excess delay at which multi- path energy falls to X dB below the peak received level. These parameters are useful measuresfor com- paring different multipath channels and have been used to determine approximate bit error ra tes for digital modulation schemes without equalization.

Historically, time dispersion parameters (e.g., for ionospheric channels) were computed by using a

- A large part of MPRG research has dealt with measuring, and then statistically modeling, the path loss and time disper- sion of multipa th radio channels.

IEEE Communications Magazine November 1991 65

- Particularly in indoor channels, individua 1 multipath components fade ve y lit- tle between two f i e d terminals or terminals moved along a small area.

, - E 1- Mean Excsss Deley = 45.05 n6 = 7

~P , -30 I , , , , , , , , , , -50 0 50 103 150 zoo 250 300 350 400 450

Lppp Excess Delay (ns) Figure 1. Example of a power delay profile illustrat- ing important channel parameters temporal average of the channel impulse response during a time period overwhich the channel appeared wide-sense stationary. In mobile systems, howev- er, the time variation of the impulse response is due primarily to motion, so the parameters may be computed over a spatial average duringwhich the channel appears wide-sensc stationary [9, 301.

Particularly in indoor channels, individual multipath components fade very little between two fixed terminals or terminalsmoved along asmall area [31. 321. Statistical processing on an exten- sive indoor propagation data base showed that individual multipath components fade in a log- normal sense over small temporal and spatial intervals, with a standard deviation of only a cou- ple of dB. Simultaneous C W measurements showed that the narrowband fading between two fixed terminals is Ricean, but the CW fading can be either Ricean, log-normal, or Rayleigh when the receiver is moved over a small area. Deep fades of individual multipath components arc pri- marily due to shadowing a s a terminal is moved. or results from the phasor sum of unrcsolvable multipath components within a resolution cell.

Knowledge of the channel time dispcrsion to temporal resolutions much greater (smaller dura- tion) than the bit durations of a communication signal, and how thc time dispersiveness changes over space, is important because these factors deter- mine the instantaneous bit error floor that occurs because of data smearing. By performing the time convolution of transmitted data bits with accurate spatially (time) varying impulse response models. i t becomes possible to predict burst errors and conduct real-timc system design expcr-

f I

i

I I

1

I

1.0 10.0

T-R Separat!on. Ikilometersi ~ -2 ~

~

Figure 2. Scatter plot of wideband path loss measured in cellular and microcellular channels in Europe[2] imcnts using computer simulation instead of pro- totype hardware [MI.

Historically, path loss has been found to be closc- ly l inked to the separat ion distance between transmitter and receiver, so a simple model for the path loss at some distance r>ro from a trans- mitter can be expressed as

P(r) = P(r,) 6 0 irIn (1) The exponent 11 in Equation (1) represents the

best-fit (in a mean square sense) average power law at which signal power decays with respect to a free space measurement at ro, the close-in reference distance. Measurements have shown that field mea- surements are generally log-normally distributed about the average distance-depcndent power law given in ( I ) . independent of r.

Different time dispersion and path loss results for urban microcellular measurements reported in [2] are shown in Fig. 2 and Table 1. The mea- surements were made throughout several existing ccllular markets in Germany using a 500 ns prob- ing pulse and existing cellular base station anten- na heights that ranged from 20 m to 93 m in height [ 2 ] . Data in Fig. 2 and Table 1 assume ro is 100 m. Using the assumption that path loss is log-normally distributed about the mean power law, the standard deviation 0 in dB completcly speci- fies the distribution of received power as a func- tion of distance. For a best-fit path loss exponent, owill be minimized over the entire scatter plot.

Reference [29] shows that the selection of the

66 IEEE Cornmunications Magazine Novemher 1991

I I ,

a) Excess Delay lnsl

1 ' ' ' ' ' ' ':'n=4

:'n=J

n=2.1

. . . . . . ,

I . . . * . - I . 3 . , L . U " " t ' 1 100 1.5 2 3 4 5 6

l o T-R Separalion (m)

CIM. SIRCIMprovides the user with datafiles that 'igure 3 (These waveforms are examples produced bq ., contain amplitudes, phases, and time delays of the channel impulse response, and computes fading statistics, large-scale path loss, best fit exponent and standard dev. User can specify building type, topography (LOS or OBS), and T-R separation distance. SIRCIM is based on extensive measurements made in more than 10 different buildings. close-in reference distance is important for sys- tem design considerations and capacity analysis. The close-in reference distance is a leverage point from which a best-fit mean path loss exponent extends. Historically, a reference distance of 1 km was used as early land mobile systems strived to pro- vide coverage over tens of kilometers. Emerging microcellular systems, however, will cover delib- erately smaller regions so that spectrum re-use can be employed more extensively in a specific market. By diminishing the size of the radio cells, more users can be accommodated within a given spectrum allocation; consequently capacity can be increased at the expense of more base stations and infrastructure. Thus, path loss models must use a close-in reference distance that is several times smaller than the distance of the most distant user within a coverage zone.

Using field measurements, it has been shown that the value of the path loss exponent can change depending on the free space reference distance cho- sen, which indicates that simple dn path loss models are subject to interpretation of the close- in leverage point selected. For microcellular and PCN systems, reference distances on the order of 1 m to 100 m are appropriate.

MPRG research has focused on measuring indoor

and microcellular channels and developing mod- elsfor such channels, since it is our belief that indoor environments and street-level systems will serve the largest number of wireless users in the com- ing decades. Extensive indoor propagation mea- surements have been and continue to be made with the goal of deriving site-specific modeling approaches and installation tools based on physi- cal descriptions of building interiors [S , 131.

Along the way, we have used statistical model- ingprocedures [lo] to reproduce, on apersonal com- puter, extensive propagation measurements given in [ 131 so that research can focus on indoor radio communication system design using realistic com- puter-generated impulse responses. Also, more recent measurements [3,4] and measurements reported in the literature [ 141 have been used togen- erate (on a computer) impulse response and path loss measurements in traditional-partitioned and soft-partitioned (Herman-Miller office parti- tions) office buildings. The statistical channel models are useful for determining, through simu- lation, irreducible bit error rates, modulation per- formance, diversity implementations, and robust equalization methods.

The propagation simulator, called SIRCIM (Sim- ulation of Indoor Radio Channel Impulse response

- MPRG research has focused on measuring indoor and microcellular channels and develop- ing models for such channels.

IEEE Communications Magazine November 1991 67

-~~ All Soft Par t i t ioned Envi ronments

130 7-p

m Elo0 (0

4 90

5 0 80

a~ 70 U

3

; 60 a

b Office Building 1 4 th Floor 1 0 Office Building 2 2nd Floor 1 0 Office Building 1 51h Floor

--/ ~ -~ Soft Porlltlon - 1 39 dE Concrete = 2 38 dE o = 4 l d B

~

1

~

40 I /’ i l 30 1

30 40 50 60 70 BO 90 100 110 120 130 Measured P a t h Loss (dB)

Figure 4. Best f i t line for a simple two-parameter model used to predict signal loss due to obstructions

Measurements). is a valuable tool for M P R G communications research. and also is being used by 40 companies and universities. The models used in SIRCIM are detailed in [ I O ] . Although similar in nature to the SURP simulation pro- gram for urban radio propagation [11], SIRCIM is based on measurements made over small scale dis- tances. and thus allowssynthesisof the phase of indi- vidual mul t ipa th componen t s based on t h e Doppler shift and a random scattering model. By computing the spatially varying phasor sum of multipathcomponents, SIRCIMrecreatcsCWfad- ing envelopes identical in nature to those mea- sured in the field. MPRG plans to update the software as more measurements become available. and as user

160 50

I 90 80 I

I I

Figure 5. Measured and predicted signal strength contours based on site-specific model [S]

experiences dictate. A useful result from [ 3 ] is that propagation

characteristics are very similar at both 1.3 GHz and 4.0 GHz, which means the SIRCIM models (based o n measurements at 1.3 GHz) will hold up to at least 4.0 GHz, and probably at somewhat higher frcquencies. Typical examples of the data pro- duced bySIRClMareshownin Fig.3. Datafiles that contain the amplitudes,phases, time delays, and path loss for individual multipath components are pro- duced by SIRCIM and written to disk for later use in bit error simulation. These files may then be used to test bit error rates and system designswith- out prototype hardware.

Narrowband measurements have shown how the

80

path loss exponent, and the dcviation about the best- fit average path loss model. can be affected by build- ing type, o r location within a building. Table 2, extracted from [ 5 ] , indicates that in different buildings, the floors can offer different values of attenuation.

Measured a t tenuat ion factors for various obstacles in indoor environments are presented in Tables 3 a n d 4 [4, 51. I n [ S I , a simple two- paramcter statistical model was developed to modcl the loss due to each partition or each wall encoun- tered between a transmittcr and receiver locatcd within a building.

A scat ter plot of p a t h loss measurements made on three different floors of two different office buildings, and the predicted path loss based on the simplc model in [ 5 ] is shown in Fig. 4. The agreement between nieasured and predicted path loss is very good for the most par t , and has an overall standard deviation of 4 dB. A standard deviation much greater than 10 dB usually results when only distance (and n o site-specific informa- tion) is used to predict signal s t rength from a data base of several different types of buildings.

For measurements in Fig. 4. the transmitter anten- nawasmounted 1.8 m abovc ground, and the receiv- e r was located at desk height. slightly shadowed by movablc office cubicles (soft partitions) and obstructed by concrete walls. Using a simple model that assumes 1.4 d B loss for each cubicle wall and 2.4 dB for each concrete wall (the walls did not span the ent i re floor), and free space propagation cverywherc else, it was possible toclose- Table 2. Attenuation values for different buildings andjloors

ly predict signal strength contours. Figures 5 , 6, and 7 illustrate how the simple

site-specific attenuation model can accurately predict coverage throughout the work space. A schematic of a typical 59 m x 59 m open plan office building with movable cloth partitions and concrete walls is shown in Fig. 6. Dark lines in Fig. 6 denote concretewalls that extended from floor to ceiling. Lighter lines represent 2.0 m tall soft partitions. Fig. 5 and 7 can be overlayed on Fig- ure 6, and show measured and predicted signal s t rength contours based o n the si te-specific model in [5]. A simple distant-dependent path loss model would provide circular contours of constant radius about the transmitting antenna.

Although this modeling work is preliminary, it demonstrates that simple descriptions about the building topography could be used to predict coverage areas and interference zones with much better accuracy than models used today. MPRG researchers are continuing measurements that will help to develop accurate site specific models. These models will then be incorporated into an auto- mated system design tool t ha t will optimally locate base stations for minimum interference and consequently maximum capacity. MPRG iswork- ing to exploit knowledge of the propagation envi- ronment to improve and automate system installation without measurements.

Work in [4] has shown that antenna polariza- tion can play a big par t in reducing the delay spread (i .e. , improving the bit e r ro r perfor- mance) . In [15], Cox descr ibes the Bellcore UDPC system as using polarization diversity to open the eye in digital modulation techniques. Our work [4] shows that, indeed, polarization diversity can be used to select the best channel at a partic- ular location. Our work also shows that circularly polarized (C-P) directional antennas, when used in line-of-sight channels, can provide a much lower delay spread than linear polarized anten- nas with similar directionality [4].

Figure 8 shows how the instantaneous rms delay spread changes as a mobile receiver is moved over a 2.5h track. The identical trackwas tra- versed with a receiver using omnidirectional and directional linear polarized antennas, and direc- tional circular polarized antennas. Note that the

I Figure 6. Blueprint of open plan building measured in Fig. 5

ITEM

In Office Buildings

Concrete Block Wall

Loss From One Floor

Loss From One Floor and one Wall

Fade observed when transmitter turned a right angle corner in a corridor

Light Textile inventory

Chain link fenced in area 20 ft. high which contains tools, inventory, and people

Metal Blanker-I2 square feet

Metallic Hoppers which hold scrpa metal for recycling-I 0 square feet

Small Metal Pole-6 in. diameter

Metal Pulley System used to hoist metal inventory4 sq. ft.

Light Machinery c 10 sq. ft.

General Machinery - 10-20 square feet

Heavy Machinery > 20 sq. ft.

Metal catwalkktairs

Light Textile

Heavy Textile inventory

Arerea where workers inspect metal finished products for defects.

Metallic inventory

Large I-beam - 16-20 in

Metallic inventory racks - B square feet

Empty Cardboard inventory 3 0 x 8 s

Soncrete Block Wall

Ceiling Duct

LOSS (dB)

13

20-30

40-50

10-15

3-5

5-12

4-7

3-6

3

6

1-4

5-1 0

10-12

5

3-5

8-1 1

3-1 2

4-7

8-1 0

4-9

3-6

13-20

1-8

"able 3 Measured signal loss due to common obstructions in buildings* * These data were collected by comparing signal strengths on either side of the obstruction

- Narrowband measure- ments have shown how the path loss exponent can be affected by building type, or location within a building.

IEEE Communications Magazine November1991 69

- Site-specific propagation models that predict, with good accuraq, the shadowing losses and the difiuc- tion effects in urban canyons are needed for system design.

I 60 60 70 80 90 1

Figure 7. Predicted contour plot of signal strength using best fit model of Fig. 3. Agreement with mea- sured data is better than simply using T-R separation alone [S] C-P helical antenna offers much less delay spread, and smaller delay spread variability, than do the L-P omni o r yagi antennas. Also, directional antennas reduce the delay spread when com- pared with omni antennas. We also have seen this on cross-campus links that illuminate several buildings at a time. In outdoor links especially, it appears that when aligned off-axis, directional C- P antennas offer much more multipath resistance than linear polarized antennas. The multipathreduc- tion is likely due to cancellation of odd-bounce mul- tipath, and offers a significant performance gain since it reduces the time dispersion of the chan- nel. Further, this finding indicates that an accu- rate propagation prediction tool must consider polarization effects.

Obstacle Description I Attenuation 1

2.5 m storage rack with small metal parts

4 m metal box

5 m storage rack with paper products

10-1 2

5 m storage rack with paper products

5 m storage rack with large metal parts

Semi-automated Assemblv Line

12-1 4

Stainless Steel Piping for Cook-Cool Process

8-1 5 Concrete floor I 10 - 1

Table 4 Shadowing eflects of common factory equip- ment

MPRG is conducting additional measure- ments up to 30 GHz around campus to provide data for building penetration loss, floor-to-floor loss for different-shaped buildings, the correlation of sig- nal strengths over small distances, and the impor- tance of antenna pattern and polarization on system design.

Referr ing back t o Fig. 2 , t he re is a large amount of scatter about the best fit, indicating that surrounding buildings have a large impact on the measured path loss between a transmitter and receiver. Site-specific propagation models that predict, with good accuracy, the shadowing loss- es and the diffraction effects in urban canyons are needed for system design, and good progress is being made by numerous researchers through- out the world. A recent paper shows the viability of ray tracing and shadowing for accurate propa- gation prediction for microcellular systems [ 161. Additionally, unpublished work at MPRG shows that in fact only a few rays and simple diffraction methods can be used most of the time to get sur- prisinglygoodprediction (within3 dB) of measured signals in microcellular environments.

Con clusion The wireless personal communications age is

coming soon. This paper has attempted to put inper- spective some recent trends andevents that areshap- ing the wireless personal communications revolution. The paper also gave insight into the research and educational activities at Virginia Tech, and point- ed out some tools and techniques used to charac- terize radio propagat ion. New models for propagation prediction will lead to appropriate spec- trum allocations and robust system designs.

One area of research that is of great impor- tance but was not mentioned is environmental safety. As important as research aimed at developing wireless personal communications, good, objec- tive in-depth experiments must be performed jointly and immediately by engineers and medical scientists to determine the health risks associated with continuous and pulsed microwave radiation. Although worldwide standards exist, it is unclear to many if those standards actually represent safe levels for humans. If a cause and effect relation- ship exists between cancers and microwave radia- tion, this must be made known immediately and the physical mechanisms must be learned to combat radiat ion effects in the future . A universal , responsible approach to this potential problem, and the public's perception of the problem, should be undertaken immediately by government, indus- try, and academia. Within a decade, low-level R F radiation at microwave frequencies will be near our bodies and all around us. In 20 to 30 years, wireless communications probably will be as per- vasive as utilitylines and housewiring are today. We must be certain that the wireless personal com- munications age will be an environmentally safe age, as well.

References [ l l T. S. Rappaport, S. Y. Setdel. and R. Singh. "300 MHz Multipath

Propagation Measurements for US. Digital Cellular Radiotele- phone," /E€€ Trans. Veh. Technol., pp. 132-39. May 1930.

121 S. Y. Seidel. eta/ . . "Path Loss, Scattering, and Multipath Delay Statistics in four European Cities for Digital Cellular and Micro- cellular Radiotelephones," lEEE Trans. Veh. Technol., vol. 40, no. 4. Nov. 1931.

70 IEEE Communications Magazine November 1991

[3] D. A. Hawbaker and T. S . Rappaport, "Indoor Wideband Radiowave Propagation Measurements at 1.3 GHz and 4.0 GHz " IEEE ElectronicsLetters, vol.26,no.21,pp. 1800-02,OctoberlO. 1990.

[4] D. A. Hawbaker, "Indoor Radio Propagation Measurements, Mod- els, and Communication System Design Issues." Masters Thesis in Electrical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA, May 1991. Also see paper in Globecom '91

[51 S. Y. Seidel and T. S. Rappaport, "900 MHz Path Loss Measure- ments and Prediction Techniques for In-Building Communica- t i o n System Design." 1991 I E E E Vehicular Technology Conference, St. Louis, MO, May 21, 1991. To be published in IEEE Trans. Antennas Propag. early 1992

[61 R. W. Lorenz "Impact of Frequency-Selective Fading on Digital Land Mobile RadioCommunicationsat Transmission Ratesof Sev- eral Hundred kbitJs," IEEE Trans. Veh. Technol.. vol. VI-35, no. 3, pp. 122-28. Aug. 1987.

[7] R. J. C. Bultitude, "Propagation Characteristics on Microcellular Urban Mobi le Radio Channels at 910 MHz," IEEEJ. in Se/. AreasCommun.. vol. 7, no. 1. pp. 31-39. Jan. 1989

[E] D. M. 1. Devasirvatham, "Multipath Time Delay Spread in the Digi- tal Portable Radio Environment," IEEE Commun. Magazine. vol. 25. no. 6, pp. 13-21, June 1987.

[91 D. C. Cox. "Time and Frequency-domain Characterizations of Mul- tipath Propagation at 910 MHz in a Suburban Mobile-Radio Environment,"RadioSci., MI. 7, no. 12, pp. 1069-81, Dec. 1972.

[ l o ] T. S . Rappaport, S. Y Seidel. and K. Takamizawa, "Statistical Channel Impulse Response Models for Factory and Open Plan Building Radio Communication System Design " IEEE Trans on Commun. vol. 39, no. 5, pp. 794-807. May 1991.

[ l 11 H. Hashemi. "Simulation of the Urban Radio Propagation Chan- nel," /E€€ Trans. Veh. Techno/., vol. VI-28. pp. 213-25, Aug. 1979.

1121 J. I. Smith, "A Computer Generated Multipath Fading Simulation for Mobile Radio." IEEE Trans. Veh. Technol.. vol. 24, no. 3, pp. 39-40. Aug. 1975.

[131 T. S . Rappaport, "Characterization of UHF Multipath Radio Chan- nels in Factory Buildings."lEEE Trans. AntennasandPropag.,vol. 37, no. 8. pp. 1058-69, Aug. 1989.

[141 D. Molkdar, "Review on Radio Propagation Into and Within Build- ings,"lEEEProc.-H. vol. 138. no. 1. pp. 61-73, Feb. 1991.

[151 D.C.Cox."ARadioSystemProposalforWidespread Low-PowerTeth- erless Communications," IEEE Trans. Comm., vol. 39, no. 2, pp. 324-35. Feb. 1991.

I161 F. Ikegami, T. Takeuchi, and S. Yoshida, "Theoretical Prediction of Mean Field Strength for Urban Mobi le Radio." IEEE Trans. Antennas Propag., vol. 39, no. 3, pp. 299.302. March 1991.

[171 D. Bishoo. "Personal Communications Service," Mobi le Radio Technol., p. 4. July 1991.

Bus., PP. 79-81, October, 1991. 1181 "RSA Report and International Cellular Penetration," Cellular

[191 J. Loeber; "Personal Communications in the UK," 1991 Eastern Communications Forum, Engineering Communications Forum, Washington. D.C., April 30, 1991.

(201 Electronics Industry Associatioflelecommunications Industry Ass@ ciation Interim Standard IS-54"Cellular System Dual-Mode Mobile Station-Base Station Compatibility Specification," May 1990.

[211 A. Maloberti. "Radio Transmission Interface of the Digital Paneu- ropean Mobile System," 1989 IEEE Vehic. Technol. Conf. Proc.. Orlando, FL, pp. 71 2-1 7, May 1989.

I221 A. Salmasi. "Cellular and Personal Communications Networks EasedontheApplicationof CodeDivision MultipleAccess(CDMA)," Proc of First Virginia Tech. Symp. on Wireless Personal Com- mun., Blacksburg,VA, pp. 10.1-.9.June5, 1991.

[231 R. L. Pickholtz, L. B. Milstein. and D. L. Schilling, "Spread Spec- trum for MobileCommunications."lEEETrans. Veh. Technol., vol. 40, no 2, pp. 313-22. May 1991.

I241 L. B. Milstein, R. L. Pickholtz, and D. L Schilling, "Shared Spec- trum ConsiderationintheDesignofa Direct SequencePCN," Proc. of FirstVirginia Tech. Symp. on Wireless Personal Commun.. Blacks- burg, VA. June 5, 1991, pp. 12.V.2, June5, 1991

[251 K. Raith and 1. Uddenfeldt, "Capacity of Digital Cellular TDMA Systems," IEEE Trans Veh. Tech., vol. 40, no. 2, May 1991. pp. 323-32, May 1991

[261 G. R. Cooperand R. W. Nettleton. "Aspread-SpectrumTechniquefor High-Capacity Mobi le Communications." IEEE Trans. Veh. Tech.. vol. 27, no. 4, pp. 264-75, Nov. 1978.

[271 IEEE Spectrum Special Issue on Research and Development, Octo- ber 1990.

1 - 00

80

60

40

20

. . . . . . . ,. . . . . . . . ........................... o Omni-Vertical :

Omni-Horizontal : 0 Directive -Vert . j

' . m Directive - Hor

OBS Topography - 32 6 m mean 1-R separation

0 2 5 0 5 0 0 7 5 1 00 1 25 1 50 1.75 2.00 2.25 2.50 Distance (wovelengths)

Fimre 8 Variation of RMS delay spread along a 2.5 ;1 track for omni and direc- - - - tional antennas. Gain of omni antenna was 1.5 dei, gains of linear directional and CP directional antennas were approximately 10 dBi. Frequency = 1.3 GHz. L281 K. L. Blackard. T. S. Rappaport, and C. W. Bostian, "Radio Fre-

quency NoiseMeasurementsand Modelsfor Indoor Wireless Com- munications at 918 MHz. 2.44 GHz. and 4.0 GHz," Proc. of 1991 IEEE Int'l Commun. Conf., Denver, CO, pp. 19.3-.5, June 23, 1991.

L291 1. S. Rappaport and L. B. Milstein, "Effects of Path Loss and Fringe User Distribution in CDMACellular Frequency Reuse," Proc. of 1990 IEEE Global Commun. Conf., San Diego, CA, pp. 500- 06, Dec. 2, 1990.

[301 T. S. Rappaport, "Indoor Radio Communications for Factories of the Future," IEEE Commun. Magazine, vol. 27, no. 5, pp. 15- 24, May 1989.

[311 T. S . Rappaport and C. D. McGillem. "UHF Fading in Factories," IEEEJ. Se/. Areas Commun., vol. 7, no. 1 , pp. 40-48, Jan. 1989.

[321 R. Ganesh and K. Pahlavan "Statistics of Short Time Variations of Indoor Radio Propagation," Proc. 1991 IEEE Int'l Commun. Conf., Denver, CO, pp. 1-6. June 22, 1991.

I331 V. Fung and T. 5. Rappaport, "Bit-Error Simulation of d 4 DQPSK in Flatand Frequency-Selective Fading Mobile RadioChannels with Real Time Applications," Proc. 1991 IEEE Int'l Commun. Conf., Denver, CO, pp. 19.3.1-3.5, June23, 1991.

[341 H. Ochsner, "DECT-Digital European CordlessTelecommunications." 1989 IEEE Veh. Technol. Conf. Proc.. Orlando, FL. pp. 718-721. May 1989.

[351 D. J. Goodman, "Trends in Cellular and Cordless Communica- tions," IEEE Commun. Magazine. vol. 29, no. 6, pp. 31-40, June 1991.

1361 "Report on Spectrum Sharing in the 1850-1990 MHz band Between PCS and Private Operational Fixed Microwave Ser- vice," American Personal Communications, vols. I-11, filed at FCC, Washington, D.C., July 1991.

[371 "Experimental License Progress Report." Telesis Technologies Lab- oratory, filed at FCC. Washington, D.C., August 22, 1991

[381 P. Estabrook, e t al., "A 20/30 GHz Personal Access Satellite Sys- temDesign."lEEEICC89.Boston,MA,pp.7.4.1-4.5,Junel2.1989.

[391 B. Bertiger. "The Iridium System," Proc. of First Virginia Tech. Symp. on Wireless Personal Commun.. Blacksburg. VA, pp. 4.1-.9. June 5, 1991.

Biography Theodore Rappaport ISM '911 received a Ph D in EE from Purdue Uni- versity in 1987 He is an assistant professor in the Bradley Department of Electrical Engineering at Virginia Polytechnic Institute and State University in Blacksburg, Virginia Dr Rappaport i s a faculty member of the Mobile and Portable Radio Research Group (MPRG)

IEEE Communications Magazine November 1991 71