- Shopping Bag ( 0 items )
Bridge the gap from existing regional wireless standards to the next generation global communication system with these selected readings, a companion text to Software Radio Architecture, by Joseph Mitola III. This single-source reference summarizes major trends in software radio technology by combining classic publications, cutting-edge overviews and promising research from leading experts in the field.
The role of software and digital signal processing in radio has grown to encompass reconfigurability at all levels of the protocol stack. Research has ignited wider commercial interest in these concepts, which have been endorsed by all segments of industry, from software applications providers to component manufacturers. The software radio approach will increase software content for commercial wireless systems and is a potential solution to the need for increased efficiency in service delivery.
This volume covers:
SOFTWARE RADIO TECHNOLOGIES is an essential resource for researchers and practicing engineers seeking an in-depth treatment of technical issues raised by software radio architecture evolution.
Without mathematical foundations, systems architecture discussions quickly devolve into whether "my" approach is better than "your" approach. With mathematical foundations, both "you" and "I" have an outside reference point, some quantitative criteria by which to assess the quality of both of our designs. Structured programming abolished the GOTO statement and set criteria for module size, for example. The banishment of the GOTO was not just good engineering practice. Mathematicians proved that one could express any Turing-computable expression without the GOTO, so there was no need to clutter up software with random branches when the "structured" primitives were easier to understand and were provably capable of computing whatever is needed. A software radio contains a lot of software, so the software design principles apply. In addition, however, the software radio supports isochronous processes, tasks that have to be accomplished within a short time window that can be specified tightly in advance. Speech signal processing tasks in a software radio, for example, cannot process more than a few hundred milliseconds of data in a block without introducing time delays that the speakers find objectionable. Software radio code is therefore isochronous code. This means that the code must run to complete within a short time window that may be tightly specified in advance. The second paper shows that this constraint actually rules out the use of Turing-complete constructs in isochronous radio software. Although this sounds like a limitation, it really is not. This paper shows that it is an advantage because in this case, one can write an algorithm to answer the general question "Is a given program an algorithm-that is, will it terminate?" This classic problem is provably unsolvable in the Turing-computable case, per the famous Goedel incompleteness theorem. But this question is provably computable for isochronous software. Not only that, but the algorithm yields a tight upper bound on how long the program will require to run to complete. This algorithm can then be used to determine automatically whether software radio plug-and-play modules will support isochronous streams even if they have never been used together before. The mathematical foundations of the software radio, then, offer a relatively untapped reserve of tools to constrain and manipulate the performance of the software radio.
In general, architecture is about functions, components, and design rules. A good architecture prescribes design rules for applying a family of hardware and/or software components to achieve a specified set of functions within specified constraints that may also be considered high-level design rules. The design rules include interface specifications but also include performance constraints and management-oriented rules. The mandatory reuse of existing hardware or software components is an example of a management-oriented design rule that is typically part of a software radio architecture. The mathematical foundations of architecture employ the methods of point-set topology. The first paper introduces topological techniques, and the second extends and applies these techniques to plug-and-play goals of software radio. The third paper emphasizes the need for performance management of the signal processing resources, particularly of multithreaded multiprocessing pooled processors. These architecture considerations also have an impact on the business case, which is also introduced in the third paper. The two papers also offer precise definitions of (ideal) software radio, software-defined radio (SDR), and programmable digital radio (PDR).
The hallmarks of software radio-multiband multimode capability, wideband RF, high-speed ADCs and DACs, and software realization of signal processing functions-imply additional foundations besides mathematics. Engineering practical software-defined radios therefore require cutting-edge analog, ADC/DAC, frequency standards, timing, and related engineering subsystems. The last three papers emphasize the engineering foundations.
Software Radios Survey, Critical Evaluation and Future Directions
J. Mitola III, Chief Scientist, Electronic Systems E-Systems, Melpar Division 11225 Waples Mill Road, Fairfax, VA 22030
A software radio is a set of Digital Signal Processing (DSP) primitives, a meta-level system for combining the primitives into communications systems functions (transmitter, channel model, receiver ...) and a set of target processors on which the software radio is hosted for real-time communications. Typical applications include speech/music, modems, packet radio, telemetry and High Definition Television. Low cost high performance DSP chips promote delivery of enhanced communications services as software radios. Time to incorporate a new service into a product is reduced dramatically using this approach. Low costs and new services will continue to increase demand for software radio tool sets and CAD environments.
This paper relates performance of enabling hardware technologies to software radio requirements, portending a decade of shift from hardware radios toward software intensive approaches. Such approaches require efficient use of computational resources through topological consistency of radio functions and host architectures. This leads to a layered topology oriented design approach encapsulated in a canonical Open Architecture Software Radio model. This model underscores challenges in simulation and computer aided design tools for radio engineering. It also provides a unified mathematical framework for quantitative analysis of algorithm structures, host architectures and system performance for radio engineering CAD environments of the 90's.
An ideal software radio transceiver is illustrated in Figure 1. D/A and A/D converters at the transmit/ receive antenna and at handset allow all radio transmit, receive, signal generation, modulation/ demodulation, timing, control, coding and decoding functions to be performed in software.
This "software" radio of course includes many non DSP hardware components like RF conversion, RF distribution, anti-aliasing filters, power handling, etc. But the increased performance and continually dropping costs of the enabling technologies of A/D and D/A converters, high speed digital signal distribution, DSP chips and embedded computing are facilitating a shift toward software intensive approaches especially in large scale telesystems applications.
A. Software Radios Expand Multimedia Services
The ideal software radio interoperates with any communications service in its RF preselector band and A/D bandwidth. By running a different algorithm, the software radio instantly reconfigures itself to the appropriate signal format. This opens interesting possibilities for expanded radio services. A future software radio might autonomously select the best transmission mode (Personal Communication Network, Mobile Cellular Network, etc), send probing signals to establish a link, explore communications protocols with the remote end and adapt to the remote signal format. It could select the mode for lowest cost, service availability or best signal quality. The software radio reconfigures itself on the fly to support the required services.
This kind of flexibility opens opportunities for reduced costs and improved services for military as well as civilian applications. Prior generation military radios used single signal formats. Comms centers require different radios for different modes: SINGCARS for voice, TACFIRE for data, etc. Radios under development embed computing resources for a wider range of data formats. The Commanders' Tactical Terminal Hybrid (CTT-H), for example, will interoperate with several signal families in UHF, providing voice, data and imagery data relay . CTT radios are multi-media capable today using a separate image compression unit. In the future image compression services may be embedded in a software radio service of CTT. Such software radio technology will also offer expanded services. Instead of calling target coordinates to a missile battery by voice, a future forward observer might uplink a video frame of the target from his night vision goggles to an airborne relay and thence to the terminal guidance of a missile. Later, he might send a battle damage assessment video frame to headquarters using the same airborne relay but a different protocol. Such advanced services require flexible signal generation, wideband Intermediate Frequency (IF) Analog to Digital (A/D) conversion, adaptive signal processing in the radio relay and data-driven routing.
B. Software Radio Telesystems Architecture
Such advanced services are on an evolutionary path which began in the early 1980's. The data links, mobile radios and LAN's of Figure 2 illustrate the system architecture of software radio oriented wide area telesystems.
A regional service center provides central control while local service centers provide statistical multiplexing, bandwidth management and ancillary data. These telesystems have used message passing for distributed remote control since 1982. Layering and message passing have provided a robust, extensible applications architecture through several generations of hardware and operating systems. Figure 3 shows how telesystems connectivity services are layered according to the International Standards Organization/ Open Systems Interconnect (ISO/OSI) model (a) within a service center, (b) on line of sight radio data links and (c) on wide area remote satellite links. With this approach 44 thousand Lines of Code (KLOC) allocated to connectivity layers insulates over half a million lines of software radio applications code from changes in the implementations. This open architecture now extends to server nodes based on the VME backplane (Figure 4).
The Digital Multimedia Workstation is a typical VME node. It provides remote voice, video and data acquisition and control including full motion video and facsimile services. Earlier generations were limited to voice and data. But increased DSP processing capacities of Commercial Off The Shelf (COTS) boards have made it practical to integrate video, extending the ISO model to multimedia. Earlier systems also required dedicated PCM backbones for signal routing and separate audio intercoms. Current systems integrate these functions on a COTS FDDI network.
Each VME node has a signal flow architecture. But closer examination reveals the VME Open Architecture Myth. Although the multiple DSP boards conform to VME at the physical level, that is where open-ness ends. With no organizing paradigm at the applications layer, the integration of useful primitives like digital filters into radio systems is highly labor intensive. The system developer manually structures the data sets and signal and control flows. SPOX provides non-application specific support, but lacks a radio engineering paradigm for applications layer interoperability. The Software Services Backplane developed by E-Systems provides interoperability for signal processing data bases between VAX/VMS-ORACLE and Sun/UNIX-Sybase. But the Fast Fourier Transform (FFT) output from a DASP/ Fast Fourier Transform (FFT) board is not structured for the TMS320 MIMD board. If VME board and software vendors followed a common signal flow model, the open-ness of the VME architecture would begin to extend to software radio applications. The open architecture software radio offers insights into necessary applications layer interoperability.
The Celltap node illustrates the reductions in product development time realized through software radios. This law enforcement node was conceived in 1990. An initial AMPS based product was delivered in a laptop computer with a COTS DSP board in 1991. Product extensions from AMPS to TACS to Nordic Mobile required only a few months of software development with a small Independent R&D staff. The product is a flexible receiver rather than a complete transceiver. This simplification contributed to product success as a "pure" software radio. Such software-only evolution of new communications services is possible if (1) New services are in the preselector RF band, (2) New instantaneous bandwidths are within the A/D - D/A bandwidths and (3) The new signal, data and service complexities are within the capacities of the embedded processing. Such software radio applications will grow as the enabling technologies increase in instantaneous bandwidth and capacity of embedded processing.
II. Enabling Technologies
Technologies which support A/D, DSP/embedded computing and high speed digital interconnect define feasible applications and evolution paths for software radios.
A. A/D and D/A Conversion
Placement of the A/D converter in the software radio architecture drives sampling rate and dynamic range. Baseband A/D samples a single subscriber bandwidth exclusive of hopping or spreading. Agility band IF A/D samples the predetected, hopped or spread spectrum bandwidth. Wideband IF A/D samples the instantaneous bandwidth of a network of users. RF A/D samples filtered antenna signals without RF conversion. RF/IF-A/D needs greatest dynamic range because of the "near-far" situation in which strong nearby signals and weak distant signals must be processed by one A/D. Table 1 shows the relationship between potential software radio applications and representative A/D converter requirements. HF-IF requires large dynamic range due to propagation, fluctuating noise and small coherence bandwidth, Dynamic range requirements diminish with increasing frequency and increase for multiple users due to the near-far problem.
Figure 5 shows dynamic range as a function of sampling rate for representative A/D converters. Phillips described a 650 MHz 7.8 bit A/D with input signal bandwidths up to 150 MHz. A 1 GHz optical A/D with 1.8 bits of noise free dynamic range has been demonstrated. The 8 GHz 8 bit converter goal was announced by DARPA. Resolution in dB is not equal to linear dynamic range. Generally, A/D resolution is 3 to 6 dB greater than spurious free noise free linear dynamic range. For example, the Burr Brown 10 MHz 12 bit model ZPB1603 has a spurious free dynamic range of 72 dB, but a Signal to Noise Ratio of 67 dB.
Comparing A/D requirements to capabilities shows the feasibility of software radio applications in speech, IF and agility band VHF and UHF; and wideband applications of limited instantaneous dynamic range (e.g. microwave QAM with constellations < 64 and symbol rates < 30 MHz).
Critical Assessment: The noise free spurious free linear dynamic range of A/D converters is limited by aperture uncertainty and device noise of the sample and hold circuit. The aperture uncertainty tolerated by a B bit converter is approximated for a signal V(t)= A cos(wt), dV/dt = Aw sin(wt) and [dV/dt]*Dt = A/2**(B+1). Thus, aperture uncertainty Dt is approximately 1/(w*2**(B+1)).
Aperture uncertainties increase linearly with frequency and exponentially with the number of bits (Table 2). Thus, an 8 GHz 8 bit converter requires aperture uncertainty < 100 fs. Hundred femtosecond stabilities have been reported in conjunction with optoelectronic RF sampling circuits with bandwidths of 275 GHz. Thus an 8 GHz x 8 bit A/D is within reach in terms of aperture uncertainty. Conversion of the sample and hold signal depends on switching speeds. An electron-gas GaAs Schottky diode with 3 THz cutoff frequency has been reported. Such switching device technology is sufficient to evolve toward 30p GHz A/D rates but aperture uncertainty requirements will limit dynamic range. There are many untapped software radio applications within this growing envelope or evolving A/D technology.
Excerpted from Software Radio Technologies Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
Chapter 1: Foundations.
Software Radios: Survey, Critical Evaluation and Future Directions.
The Software Radio Architecture.
Technical Challenges in the Globalization of Software Radio.
Software-Defined Radios: Facets of a Developing Technology.
Software Radio Architecture: A Mathematical Perspective.
Beyond Software Radio, Towards Re-configurability Across the Whole System and Across Networks.
Chapter 2: Enabling Technologies.
CMOS Wireless Transceivers: The New Wave.
Analog-to-Digital Converters and Their Applications in Radio Receivers.
Analog-to-Digital Converter Survey and Analysis.
Power Consumption of A/D Converters for Software Radio Applications.
An Overview of Sigma-Delta Converters.
The Theory of Bandpass Sampling.
RSFQ Front-end for a Software Radio Receiver.
On Sampling Rate, Analog Prefiltering, and Sufficient Statistics for Digital Receivers.
Digital IF Filter Technology for 3G Systems: An Introduction.
The DSP Bottleneck.
VLSI Design and Implementation Fuels the Signal-Processing Revolution.
Digital Signal Processors in Cellular Radio Communications.
Recent Developments in Enabling Technologies for Software Defined Radio.
FPGA in the Software Radio.
The Flexibility of Configurable Computing.
Chapter 3: Systems and Architectures.
Programmable Channelized Digital Radio.
Speakeasy: The Military Software Radio.
Advanced Digital Receiver Principles and Technologies for PCS.
Broadband RF Stage Architecture for Software-Defined Radio in Handheld Terminal Applications.
Trends in Silicon Radio Large Scale Integration: Zero IF Receiver! Zero I & Q Transmitter! Zero Discrete Passives!
Advanced Base Station Technology.
Advanced Software Radio Architecture for 3[superscript rd] Generation Mobile Systems.
A Soft Radio Architecture for Reconfigurable Platforms.
Software Radio Issues in Cellular Base Stations.
A Low-Power DSP Core-Based Software Radio Architecture.
DSP-Based Architectures for Mobile Communications: Past, Present and Future.
Architectural Overview of SPEAKeasy System.
An Architecture for Radio-Independent Wireless Access Networks.
Software-Defined Radio Architectures for Interference Cancellation in DS-CDMA Systems.
Chapter 4: Software-Defined Radio Emerging Technologies.
Direction Finding and "Smart Antennas" Using Software Radio Architectures.
Software Radio Architecture with Smart Antennas: A Tutorial on Algorithms and Complexity.
The Rapidly Deployable Radio Network.
Sample Rate Conversion for Software Radio.
Mobile Middleware for the Reconfigurable Software Radio.
Engineering the Embedded Software Radio.
Cognitive Radio: Making Software Radios More Personal.
Chapter 5: Software Defined Radio Applications and Economics.
Mode Switching and Software Download for Software Defined Radio: The SDR Forum Approach.
Toward the Software Realization of a GSM Base Station.
Real-Time Implementation of a Reconfigurable IMT-2000 Base Station Channel Modem.
Code-Division Multiplexing of a Sensor Channel: A Software Implementation.
Software Radios for Airborne Platforms.
Receiver Dimensioning in a Hybrid Multicarrier GSM Base Station.
Software Radio Economics.
About the Editors.