Software Defined Radio: What Every RF Engineer Should Know

I've collected some of the highlights of their remarks, and have also included some links below to some popular SDR items appearing on the site this year.

RFDL: What would you most like RF designers to understand about SDR?
Ditore: SDR does not eliminate the need for RF design. In fact, SDR will in all likelihood place new challenges in the path of RF designers to create wide band very linear RF front ends for SDR platforms. It is also important for RF designers to start moving out of their comfort zone to develop competencies in digital signal processing and firmware development for FPGAs. One can no longer design RF or DSP in isolation because SDRs, particularly cognitive SDRs, require a high degree of integration between RF and DSP.
Brannon: There are various implementations of SDR, many of which are already being adopted into mainstream designs. For example, many communication system designers are already designing wideband transceivers that can be used with a wide variety of waveforms. Often these are not changed on the fly as the classic definition of SDR goes. Instead, one system is designed and manufactured for a wide variety of waveforms to gain economies of scale. During alignment, calibration, or even in the field, the waveforms for the particular standard are 'installed'. These waveforms are determined by a combination of FPGA configurations and DSP code that may evolve over time. Because the basic transceiver does include the flexibility to deal with a wide range of waveforms, as the waveform evolves or in fact changes, these may be uploaded as a combination of firmware and software changes to accommodate the new waveforms. Of course the down side is that the transceiver must be designed to allow for the widest dynamic range.
Bailey: The power, space and cost tradeoffs involved in a true SDR solution. While semiconductor vendors, like TI, are working to reduce these implications, we still see these as immediate challenges to a truly software defined radio card RF design.
Hosking: RF designers often struggle deciding where to draw the line between analog and digital processing tasks when architecting a specific radio system. SDR still requires prudent design of the analog RF sections before the signal is digitized to ensure that subsequent digital signal processing stages can successfully extract the transmitted signal. Successful allocation of tasks mandates a careful tradeoff analysis of costs and performance benefits by experts in both RF and DSP domains.

RFDL: What are the remaining technical/market challenges for SDR?
Sathappan: The main technical/market challenges facing SDR today is that the framework and middleware defined to support SDR systems (SCA framework and CORBA middleware) is considered to be very resource intensive and, as such, does not fit well into the commercial communications market where the demand for highly cost-efficient systems is high. So what the SDR community needs to drive is to focus and define the specs for less resource intensive framework and middleware that can still help users achieve maximum software defined capabilities in their radio systems.
Hosking: Adoption of industry-wide software standards for SDR can help ensure customer acceptance by yielding compatible and competitive solutions that are available from a range of vendors. At the same time, innovative software solutions developed by leading vendors must be protected in order to provide some return on investment. Finding a business model to create a balance point between these two conflicting issues is the greatest challenge for SDR.
Ditore: SDR has already been deployed in military tactical radios and has been fielded successfully. Many commercial radios have also moved much of the radio functionality into DSP (embedded DSP and FPGA/ASIC). The current challenges are moving into the areas of cognitive radio. Being able to detect the operating environment and switching between different communication PHYs on the fly allows a radio to function anywhere at any time. Also possibly being able to optimize the needed signal BW for a given link would allow for efficient use of available spectrum.
Brannon: From a technical point of view, the bottle neck is usually the ADC in the receive path. In the transmit path, the limitation is often the DAC and modulator, however, this is dependant on the waveforms necessary to be transmitted. From a marketing point of view, the challenge of a full SDR system is the increased cost associated with additional FPGA and DSP capacity. Dedicated ASICS are always cheaper and the flexibility always comes with a cost. In addition, a system designed with SDR in mind often costs more because the RF design must meet the worst anticipated requirements cause over design for all others. This poses a challenge to meet both the performance requirements as well as the cost objectives

RFDL: What will the "next-generation" of SDR look like?
Brannon:Next gen SDR will largely be transparent to the end user. Just as many systems today incorporate many SDR techniques, the user is often unaware of this fact. A well designed system will therefore be transparent to the end user. From the operator point of view, the ultimate cost of operation should be much lower in that one piece of hardware may be used for a much longer period of time with sufficient resources to support field upgrades in a manner that will extend the usefulness of the equipment.
Hosking: SDR technology will move even closer to the antenna. Obviously, higher speed A/D converters and faster DSP engines will allow wideband software signal processing. But the key enabler for this trend will be innovative designs for programmable RF filters preceding the wideband SDR digitizers.
Sathappan: The next generation SDR is a full software controlled radio whereby even the frequency of operation is programmable to a wide range with modulators, mixers and tuners capable of covering wide range of frequencies. There is an increasing trend toward zero IF (or direct conversion) radio which will become standard. This will be followed with cognitive radios that can sample the wireless interfaces and adapt the radio to the frequency and modulation scheme with the best efficiency of operation.
Ditore: The holy grail of SDR is to have a DC to daylight analog front end with digital down-conversion and sampling of the signal at the antenna. While we are not there yet, the next generation of SDR will have higher dynamic range A/D and D/A converters operating at GSPS rates for mA of current consumption. These will be incorporated into single chip designs with all of the filtering, demodulation, decoding, and equalization contained onboard.

Design Features

• Basics of Software Defined Radio, Part 1
• SDR Basics Part 2: Receivers
• SDR Basics Part 3: Transmitters
• High-speed ADC technology paves the way for software defined radio
• Putting FPGAs to Work in Software Radio Systems
• The new software defined radio development process -- yours for the taking

Products

• Pentek's multiband SDR transceiver delivers 10 dB A/D performance improvement

News

• Software Defined Radio Can Help Public Safety Groups Use 700 MHz Band
• SDR Forum launches second Smart Radio Challenge