Guest Editorial software radios - Selected Areas in ... - IEEE Xplore

IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 17, NO. 4, APRIL 1999

509

Guest Editorial Software Radios I. INTRODUCTION

T

HE Guest Editors of this issue of the IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS (J-SAC) are pleased to present this issue dedicated to the Software Radio. Our vision for this issue is that it serves as a surrogate graduate-level text on this important topic. Although virtually unheard of five years ago, the software radio now figures prominently in plans for third generation (3G) wireless services, architectures, and products. Multiband, multimode, mobile vehicular radios are being acquired for the military, while multimode commercial handsets are the focus of commercial interest. In June of 1998, for example, the European Commission and the Software-Defined Radio Forum (formerly the Modular Multifunction Information Transfer Systems or MMITS Forum) sponsored the First International Workshop on Software Radios. This was held in conjunction with the European Advanced Communications Technology and Services (ACTS) Mobile Communications Summit. A. Urie, Product Strategy Director for Alcatel’s Mobile Communications Division, warned during the closing panel session that “the next six months are critical to the software radio.” He and H. Huomo, Vice President of Nokia, agreed that the normalization of the U.S. and European recommendations for IMT-2000 [for code division multiple access (CDMA)], is essential to the development of software-radio handsets in the near term. Current differences in chip rate and channel spacing in U.S. and European standards perpetuate hardwareintensive implementations. Normalization to a family of chip rates from a single clock would allow one handset architecture to embrace the diverse needs of the global marketplace tailored by software. The related economies of scale may be essential to 3G success. Service providers like M. Swinburne (Orange, U.K,) and S. Blust (BellSouth Wireless, U.S.), also speaking at the International Workshop, underscored software radio technology for service delivery. At least one service provider had to recall “literally millions of handsets” to correct a software error in erasable read-only memory that did not become apparent until deployed. A software-defined handset could have been updated without the expensive recall. Orange Communications, similarly, would like to download qualityenhancement features such as vocoders customized to the subscriber’s native language. The development of a trusted download capability is thus a key thrust of the SDR Forum’s current work. In this context, this issue comprehensively reviews emerging software radio technology, attending to the performance and flexibility goals of handset, fixed infrastructure, and mobile vehicular applications. The software radio concept has evolved considerably since the landmark special issue of the IEEE COMMUNICATIONS MAGAZINE in May 1995. Prior to that time, Publisher Item Identifier S 0733-8716(99)02981-9.

wideband antennas and RF subsystems, high performance analog to digital conversion (ADC), and digital signal processing (DSP) technology were so expensive that its applications were limited to cutting edge one-of-a-kind military applications. SPEAKeasy, the military software radio technology program, was described in that issue. The only commercial application described was an adaptive array antenna with a high capacity multichannel DSP architecture for a software radio base station. But there was no significant commercial investment at that time. To the careful reader, Wepman’s paper explained why: ADC’s did not yet have the combination of sampling rate and two-tone spurious-free dynamic range needed for commercial applications. GSM base stations, for example, are tested to a 90 dB near–far ratio, while the ADC technology offered only expensive or inferior approaches. A 16-bit ADC could attain the near–far performance with a sampling rate of 2 MHz, requiring a dozen parallel ADC channels for a 25MHz GSM uplink band. Or one could sample the entire band with about 70 dB of near–far ratio. Neither approach worked for the commercial sector. SPEAKeasy demonstrated that one could move the ADC to intermediate frequencies (IF’s) but it did not address the hard product-delivery issues faced by the commercial sector. II. FOUNDATIONS How has the software radio concept evolved? What is a software radio? With the intense interest and excitement comes an inclination to stretch the definitions. So this issue offers an authoritative set of definitions in the lead paper by Mitola, “Software Radio Architecture: A Mathematical Perspective.” An expanded reference model of radio communications highlights the software aspects without minimizing the significant analog RF aspects. This paper also introduces important mathematical properties of the software radio including computability. Perhaps a little surprisingly, software radios do not need full Turing computability. In fact, a subset of the total recursive functions is more appropriate to software radio code because the algorithms must terminate within a short predefined time window using limited computational resources. The isochronous nature of the communications streams, therefore, greatly constrains the computational model. Since such models of computation have not been studied extensively, the goal is to stimulate interest among computer science researchers. The computational geometry of the software radio also suggests a layered reference model for the system and software. Concrete complexity and computational geometry are both necessary to address the stability of the software in over-the-air downloading of new capabilities. The service providers are depending on such downloads for the commercial success of 3G services.

0733–8716/99$10.00  1999 IEEE

510


In addition, the economic viability of the components has evolved. The wideband ADC is one of the critical components of the software radio. ADC’s with 14 bits (nominally 84 dB) of dynamic range with 70 Ms/s sampling rates are virtually commodity products, while 16-bit devices are on the horizon. Walden’s paper “Analog-to-Digital Converter Survey and Analysis” characterizes the critical parameters of this pivotal technology including trends in ADC product and technology evolution. The rate of progress in simultaneously increasing sampling rate and usable dynamic range of ADC’s has slowed in recent years. Pushing timing uncertainties below 1 ps is critical to near-term progress. Fortunately, there are about four orders of magnitude before we encounter the Heisenberg uncertainty limit. Given a wideband ADC, one must filter that incoming stream efficiently, selecting the subscriber signal and converting it to baseband. Although there are a number of programmable down-converter applications-specific integrated circuits (ASIC’s) (“channelizers”) on the market, continued improvements in computational efficiency are needed. Oh et al. describe this in “On the Use of Interpolated Second-Order Polynomials for Efficient Filter Design in Programmable Downconversion.” They formulate the design of the programmable down converter as an optimization problem over normalized constraints in the pass band and stop band, comparing the results to the latest product from the Harris Corporation. Although the resulting pointdesign is important, it is the sophistication of the design process, introducing degrees of freedom through a novel sequence of cascaded integrator-comb filters, decimators, and modified half-band filters that is most instructive. If the number of baseband channels is relatively small (e.g., in the CDMA standards), the linear growth of the hardware with the number of channels is not an issue. But for narrowband standards such as the digital advanced mobile phone system (D-AMPS), with hundreds of baseband channels, the large number of programmable down converters drives base station cost and complexity. Zangi and Koilpillai attack this issue further in “Software Radio Issues in Cellular Base Stations,” extracting all channels at once in the spirit of the transmultiplexer. Their analysis explicitly takes into account the quantization noise of the wideband ADC and the computational complexity of the parallel channelizer. A reduction in complexity of up to 50 times is derived for their subsampled filter bank channelizer. These papers reflect the evolution of the software radio from a research topic to a focus of infrastructure product development. And product focus continues to evolve. Four years ago, the early adopters of software radio technology were the military. Since DSP’s and ADC’s consume high power compared with analog components (albeit with more flexible functionality), hand-held software radios would have low battery life. But military vehicular radios with fewer power constraints offered promising applications. Military radio designers had to address self-generated interference including the problem of “screaming in your own ears” that occurs when a multichannel radio attempts to transmit and receive on subscriber channels

that are within the same ADC channel. Active cancellation remains a subject of research. But the commercial sector, in forums such as the European Telecommunications Standards Institute, the ACTS Mobile Summit, and the GSM MoU committee, have set a different goal for the software radio. They see the software radio as a handset design for cost effectively delivering 3G services in partial-band deployments. The wideband CDMA modes of 3G may be allocated part of a first or second-generation’s spectrum in a deployment area. As a result, the handset must be backward-compatible with prior generation infrastructure. If there were a single wideband IF with a chip-rate that is compatible with both current CDMA and 3G CDMA, then the handset hardware could have a single hardware-intensive clock, despreader, and digital converter, with essentially all other modes implemented in software (and/or firmware). Low cost, low power DSP’s are needed to address wideband CDMA challenges posed by this product focus. Gunn therefore describes lowpower DSP design strategies for advanced channel modulation and protocols in “A Low Power DSP Core-Based Software Radio Architecture.” This paper illustrates the design of low power ASIC’s which are based on a DSP core. Several DSP manufacturers now offer this approach as a way of minimizing packaging and interconnect while delivering specialized circuits needed for 3G CDMA clocks and rake receivers. The section of this issue on foundations contains the papers cited above. In addition, there are sections on Systems, Smart Antennas, and Applications. The Systems section describes integrated systems including virtual radios, channel modulation testbeds, and SPEAKeasy II, the latest military software radio. Smart antennas warrant a section that includes a tutorial with traffic capacity implications. Joint space–time techniques for enhancing received CIR and increasing the effective data rate are also included. Rapidly deployable networks based on smart transmitting antennas concludes that section. The issue concludes with applications of software radio technology to trellis coding, proximity sensors, and airborne systems. Highlights of these papers are as follows.

III. SYSTEMS Systems papers embrace research and commercial and military systems architectures. “Virtual Radios” by Bose et al. describes the “purest” implementation of the software radio concept thus far reported in the literature. Hosted on a general purpose DEC Alpha computer, the virtual radio environment employs minimal high speed hardware to deliver a sampled IF stream to a software environment which achieves realtime performance for low capacity radio streams without the design overhead associated with DSP implementations. The SpectrumWare software environment encompasses the tools needed for related research. Turletti et al. extends the virtual radio to the GSM base station and addresses the quantification of processing demand and available computational capacity in depth. Their paper, “Toward the Software Realization of a GSM Base Transceiver Station,” explains the quantitative basis for sizing engineering alternatives to software radio


GSM base stations. Correal et al. then describe a testbed for “A DSP-Based DS–CDMA Multiuser Receiver Employing Partial Parallel Interference Cancellation.” The description of the algorithm and the analysis of its complexity shed light on the power of the software radio architecture for the deployment of algorithms that enhance receiver performance. In “A Smart Software Radio: Concept Development And Demonstration,” Patti et al. describe the FLIP-WAVE channel modulation and related software-based modem which was developed using their software radio testbed. Cook and Bonser provide “Architectural Overview of the SPEAKeasy System,” emphasizing the SPEAKeasy Phase 2 system. In addition to additional background on the SPEAKeasy program, this paper tabulates the functions of the applications programmer interface (API) in which this system was implemented. This API forms the basis for industry deliberations on open architecture standards in the Software Defined Radio Forum. Although the concise treatment merely introduces the API, the details are available from the SDR Forum. IV. SMART ANTENNAS Razavilar et al. open the section on smart antennas with a tutorial, analysis of computational complexity, and implications on traffic capacity of a cellular system. Their paper, “Software Radio Architecture with Smart Antennas: A Tutorial on Algorithms and Complexity,” introduces the key concepts and illustrates the significance of the smart antenna to traffic engineering. Lee and Samueli then describe “Adaptive Antenna Arrays and Equalization Techniques for High Bit-Rate QAM Receivers.” Joint adaptive beamforming and equalization are shown to yield enhanced bit error rate performance indoors and outdoors. The analysis of concrete complexity illuminates the way in which such smart antennas may be implemented in a software radio architecture. This section concludes with the paper by Evans et al. who, in their paper, “The Rapidly Deployable Radio Network,” describe aspects of the DARPA global mobile (GloMo) program that emphasizes smart antennas in infrastructure that is mobile This paper addresses much more than smart antennas, and it is unique in its treatment of a software radio transmitter architecture which includes an efficient implementation of an adaptive transmit beamforming network.

511

DSP architecture to deliver quality enhancement differentially by subscriber need. They consider bit-interleaved coded modulation for bandwidth-efficient transmission, addressing iterative decoding, hard-decision feedback, punctured convolutional codes, Hamming-distance, diversity order, Euclidean distance, and multiphase and multilevel modulation. Their approach yields performance gains over conventional trelliscoded modulation over channels with a wide variety of fading statistics with a simple mechanism for variable-rate transmission. Smith et al. apply software radio techniques in their paper “Code-Division Multiplexing of a Sensor Channel: A Software Implementation.” The only hardware used is a front end gain stage consisting of two operational amplifies and a microcontroller. The CDMA and TDMA modems of a proximity sensor are implemented entirely in software. Cummings’ paper “Software Radios for Airborne Platforms” concludes the issue. The U.S. Department of Defense plans to acquire a new generation of radios based on a programmable modular communication concept. This paper describe software radio configurations useful in airborne applications. This paper identifies organizational roles in software radio development, characterizes the need for software radios in airborne applications, and highlights those configurations found to be attractive. The methodology, CONOPS and conclusions are summarized. The high intensity of interest across the global wireless industry—military and civilian—clearly underscores the timeliness of this issue of J-SAC.

JOSEPH MITOLA, III, Guest Editor MITRE Corp. McClean, VA 22102 USA

VANU BOSE, Guest Editor Massachusetts Institute of Technology Cambridge, MA 02139 USA

BARRY M. LEINER, Guest Editor Corporation for National Research Initiatives Sunnyvale, CA 94087 USA

V. APPLICATIONS “Broadband Interference Excision for Software-Radio Spread-Spectrum Communications Using Time–Frequency Distribution Synthesis” by Lach et al. introduces the Applications section. They introduce a new method for interference excision in spread-spectrum communications that is conducive to software radio applications. Spare processing capacity in the receiver permits the use of time–frequency techniques to synthesize and cancel nonstationary interference in the time–frequency domain. “Trellis-Coded Modulation with Bit Interleaving and Iterative Decoding” by Li and Ritcey continues the theme of using the software radio’s flexible

THIERRY TURLETTI, Guest Editor INRIA Sophia Antipolis 06902 France

DAVID TENNENHOUSE, Guest Editor Massachusetts Institute of Technology Cambridge, MA 02139 USA

A. BUSH, J-SAC Board Representative

512


Joseph Mitola III (M’74) received the B.S. degree in electrical engineering (highest honors) from Northeastern University, Boston, MA, in 1972 and the M.S.E. degree in stochastic optimal control from Johns Hopkins University, Baltimore, MD, in 1974. He is a consulting scientist with the MITRE Corporation, where he provides technical advice to the U.S. Department of Defense, the Federal Administration, and the Executive Office of the President on key technical issues in telecommunications and information systems. He is on an executive exchange program between the Department of Defense and the MITRE Corporation where he is a Program Manager with the Defense Advanced Research Projects Agency. He teaches a software radio seminar at the graduate level in the United States and Europe. Prior to joining MITRE in 1993, he was Chief Scientist of Electronic Systems of E-Systems Melpar Division. He has been Chief Scientist of SIGINT Systems with the Harris Corporation, General Manager of Eastern Operations with Advanced Decision Systems (ADS) and Manager of Digital Systems with ITT’s Electrophysics Laboratories. He began his career in 1967 with the U.S. Department of Defense. He has published many articles and reports on telecommunications and information processing science and engineering. Mr. Mitola was Guest Editor of the special issue of the IEEE COMMUNICATIONS MAGAZINE on software radios in May 1995. In 1996, he was elected Chairman of the Modular Multifunction Information Transfer Systems (MMITS) Forum.

Vanu Bose received the B.S. degrees in electrical engineering and in mathematics from Massachusetts Institute of Technology (MIT), Cambridge, in 1988 and the M.S. degree in electrical engineering from MIT in 1994. He currently is a Ph.D. candidate in the MIT Department of Electrical Engineering and Computer Science. His current research is in the area of software radios, focusing on the signal processing and software architecture.

Barry M. Leiner (S’69–M’74–SM’82) received the B.E.E.E. degree in 1967 from Rensselaer Polytechnic Institute, Troy, NY, and the M.S.E.E. and Ph.D. degrees from Stanford University, Stanford, CA, in 1969 and 1973, respectively. He currently is Special Assistant to the President at the Corporation for National Research Initiatives, working in a number of areas of distributed system and networking technologies. From 1996 to 1997, he was Vice President of Microelectronics and Computer Technology Corporation (MCC), responsible for its West Coast laboratories. From 1992 to 1996, he was a Senior Scientist with the Universities Space Research Association. He spent the last two years on loan to the Advanced Research Projects Agency where he was Assistant Director of the Information Technology Office. Prior positions included Director of Research with Advanced Decision Systems (1990–1992), Assistant Director of the Research Institute for Advanced Computer Science (1985–1990), and Assistant Director, Information Processing Techniques Office, ARPA (1980–1985). He has spent the last 20 years working in the area of packet switched networking technology and its applications. Dr. Leiner is a member of Eta Kappa Nu, Tau Beta Pi, ACM, and the Internet Society.


513

Thierry Turletti (M’97) received the Ph.D. degree from INRIA, Sophia Antipolis, France. He is a Research Scientist in the High-Speed Networking Research Group at INRIA. From September 1995 to September 1996, he was a Postdoctoral Associate in the Telemedia, Networks and Systems Group at the MIT Laboratory for Computer Science. He has been working in the SpectrumWare project headed by D. Tennenhouse. He has designed and implemented in software part of a GSM base station which is currently integrated into the VuSystem. His previous research focused on designing, implementing, and evaluating multimedia applications over the Internet. His current interests include distribution of multimedia flows over heterogeneous receivers and software radio applications. He has developed IVS.

David Tennenhouse (M’87) received the B.A.Sc. and M.A.Sc. degrees in electrical engineering from the University of Toronto, Canada. He received the Ph.D. degree in 1989 from the University of Cambridge, U.K. He joined the faculty of Massachusetts Institute of Technology (MIT) in 1989. He is currently a Senior Research Scientist at MIT’s Laboratory for Computer Science and Sloan School of Management. His research has focused on systems issues related to networks, distributed computing, software radio, media processing, and the impact of information technology on organizations. Dr. Tennenhouse is a member of the ACM, has served on the Visiting Committee on Advanced Technology of the National Institute of Standards and Technology, and is Chairman of the Technology and Policy Working Group of the President’s National Information Infrastructure Task Force.