Feature Story

More feature stories by year:


Return to: 2007 Feature Stories


Sept. 1, 2007: ECNmag.com


Figure 1. This generalized block diagram of a software-defined radio transceiver shows the wide use of digital technologies in all but the analog RF front-end circuits.

Evolving communications technologies require systems that can support many existing and emerging standards. By implementing communication functions in software, designers can build such "future-proof" communications, called software defined radios (SDRs). Software tools must let developers work at a high level of abstraction so they can port reusable code to other systems. But even though SDRs depend on software for their agility, they still require hardware in the form of switch fabrics, high-speed interconnects, and heterogeneous processor architectures that include FPGAs.

What is a Software Defined Radio?

A software defined radio can receive or transmit signals in the radio frequency (RF) spectrum, but its signal-modulation methods depend on software loaded into the radio. Today, SDRs rely mainly on traditional circuits to process RF signals; but day by day, software gets closer to the antenna. A typical SDR comprises RF front-end circuits that connect to ADCs on the receive side and DACs on the transmit side (Figure 1). These converters connect to a signal processing subsystem that contains general-purpose or reconfigurable processors.

The processors' software implements wireless standards, or "waveforms," such as GSM, CDMA or the Single Channel Ground and Airborne Radio System (SINCGARS). As long as the RF front-end circuits and the ADCs and DACs operate with a wide enough bandwidth, designers can modify the radio's capabilities simply by updating its software.

Because government agencies or service providers can reconfigure SDRs on the fly, users can operate one radio in different environments and applications. A globe-trotter can download new software to her cell phone when visiting countries with different communications standards. And base stations can support many wireless standards without the need for extra expensive hardware. SDR technology also lets public-safety organizations talk to each other over non-uniform communications channels at an emergency site. International military forces can use each other's waveforms to communicate during joint operations.

Figure 2. Wideband data-link and satellite-communication terminals that operate with 274-Mbps (data rate) waveforms can use a 6U-size ECV4-RFT Echotek wideband remote fiber-transceiver board that provides high-speed ADCs and DACs as well as three FPGAs. Engineers can purchase many SDR subsystems on boards that comply with an industry-standard bus specification. Courtesy of Mercury Computer Systems.

Additionally, an SDR supplier can provide new features and services to a system in the field to extend its life without the need for an expensive trip to the factory or to a repair depot. SDR techniques also let engineers build radios that can sense RF-spectrum conditions and respond appropriately.

A Brief History of SDR

The US Air Force's Integrated Communications Navigation and Identification Avionics (ICNIA) system offers an example of an early SDR developed in the late 1970's. The system used a reprogrammable digital signal processor (DSP) to operate multi-function multi-band airborne radios in the 30 MHz to 1,600 MHz spectrum. The ICNIA technology provided the foundation for many military-radio programs.

In the 1990s, a US government program specified the first fully programmable SDR system, Speakeasy; the so-called "PC of the communications world." Speakeasy used open hardware and software architectures to support a family of voice, multimedia, and networking waveforms in the 2 MHz to 2 GHz frequency range.
Another government program, the Joint Tactical Radio System (JTRS), uses Speakeasy technology in a family of interoperable, multi-band, networked SDRs. The JTRS program aims to replace existing US-military radios with equipment that vendors can upgrade by downloading new software. This program relies on the Software Communications Architecture (SCA) open standard.

Even the Free Software Foundation has an open-source SDR project underway. GNU Radio, a free software toolkit, lets engineers build and study SDR. The source code, available under the GPL, provides implementations of broadcast and narrow-band FM radios as well as an Advanced Television Systems Committee digital HDTV transmitter and receiver.

SDR Technology Meets System Design

Figure 3. The MTI-203 Advanced Mezzanine Card (AMC) provides three DSP chips and an FPGA, which lets it easily handle WiMAX applications in commercial and defense equipment. Boards based on the AMC standard plug into MicroTCA slots. Courtesy of Mercury Computer Systems.

An SDR system should maximize flexibility, so engineers aim to design generic hardware that will let software control waveforms and radio functions. The software should take advantage of methods that enhance code reuse and portability

Engineers must first choose the best mix of processor technologies. Most SDRs use a combination of DSPs and FPGAs in addition to general-purpose microprocessors (MPUs). These types of ICs meet the high-bandwidth, low-latency, and high-performance processing requirements of a variety of current and future waveforms. But as SDR technology moves from narrowband radios to wideband data links and satellite communications, hardware designs will rely more on FPGAs.

Hardware based on standards such as VME, AdvancedTCA, and VPX a VME-based standard that supports switched-fabrics lets engineers mix and match boards to create a flexible and heterogeneous system (Figure 2). MicroTCA boards, derived from the AdvancedTCA standard, shows up in many SDR-based WiMAX base stations (Figure 3). The use of standards, such as those noted above, lets engineers take advantage of new technologies. Some sophisticated SDR systems already include the new 64-bit Cell Broadband Engine, developed by Sony, Toshiba and IBM.

Because waveforms can require different bandwidths and latencies, system designers try to maximize the bandwidth between processors in a symmetric manner. This pool-of-processors approach lets software quickly allocate resources based on communication requirements. Switch fabrics provide an effective way of handling this type of communication. New SDR systems may use high-bandwidth interconnect technologies such as Serial RapidIO and PCI-Express for processor-to-processor communications.

Figure 4. The upper signal flow represents a general SDR receiver. The lower signal flow shows how an SDR might divide tasks between FPGAs and a general-purpose processor (GPP). Mercury Computer Systems calls this the Component Portability Infrastructure and adopting it improves waveform-code portability in FPGA- and DSP-based systems.

SDR designers also aim for software reusability and portability, so many military SDR programs mandate the use of the SCA standard mentioned earlier. The SCA defines a common operating environment and APIs for waveforms. The SCA spec addresses waveform software written for general-purpose software environments that can support a significant subset of the Portable Operating System Interface (POSIX) standard and a middleware technology called the Common Object Request Broker Architecture (CORBA). The POSIX standard minimizes the cost of porting waveform software because it provides an abstraction layer for operating system-specific methods. CORBA provides a level of transparency and program-language independence. Use of such technologies in defense programs also shows the US government's commitment to adopting commercial standards as a way to reduce program risk and lifecycle costs.

Many of the new high-bandwidth waveforms demand processing power and I/O bandwidth that exceeds that provided by MPUs alone, so new SDRs code will likely run on systems based on FPGAs. Next-generation middleware products such as Mercury's Component Portability Infrastructure (Figure 4) extend the SCA concepts into the realm of DSPs and FPGAs and improve code portability and simplify system integration.

The standard software architectures and component-based design methodologies also allow third-party vendors to offer design and modeling tools for waveform applications. Radio engineers should consider adopting tool-based design methods that improve debugging, reduce human errors and increase code reusability. These tools' automatic code-generation features reduce development time but do not affect code performance.

SDRs' Future Looks Bright

SDR technologies let engineers build smarter radios that can make decisions about communication conditions and needs. The next step in SDR technology leads to cognitive radio equipment that can sense an operator's behavior and the current spectrum use and change operating conditions to increase communication efficiency. Cognitive radio technology can let unlicensed users occupy a frequency band, provided their use does not interfere with licensed users' communications. When a cognitive radio senses the presence of a licensed user, it moves to an unoccupied frequency band. This approach will lead to the efficient use of the radio spectrum.

Sidebar: What is the Software Defined Radio Forum?

SDR Related Websites

Return to: 2007 Feature Stories