SDRs Definitively Defined

Publish Date: Apr 21, 2015 | 3 Ratings | 4.33 out of 5 | Print | Submit your review

Overview

Software Defined Radios (SDRs) are not new or novel. In fact, SDRs were first introduced in 1991 by Joseph Mittola who published the 1992 seminal research paper that presented the world to the concept of a “software” defined radio (1). Mittola’s concept outlined a vision where a radio would transmit and receive wireless signals using one specific set of parameters, and that users could re-program, “on-the-fly,” the underlying hardware to instantly change the profile to different modulation, bandwidth, filtering, or frequency to communicate with another device or SDR – making the possibilities endless. Industry observers may argue the relevance and impact of SDRs over the last 20+ years; however, the recent acceleration of SDR adoption is quite relevant, progressing in-step with the advancements in software, processing technologies, and radio transceivers. Perhaps, SDRs are just now coming of age. Why the gap of more than 20 years from nascent concept to commercial adoption and deployment? Understanding this transition and time span is key to definitively defining SDRs. Basic definitions do not tell the whole story.

Some Background

Ever since Guglielmo Marconi commercialized wireless communications in the early 1900s, radios have for the most part remained fixed function, dedicated devices. Even today, consumers must purchase a new cell phone or wireless device to take advantage of a new wireless technology. SDRs have indeed been commercialized to the benefit of few.

The United States Armed Forces standardized on SDRs as a core technology for their communication systems spanning all services (Army, Navy, Air Force, Marines, and Coast Guard). This program, known as JSTARs, is the cornerstone of the communication infrastructure deployed today. The SDR architecture enables each service to communicate with one another, but more significantly, enables each transmitter and receiver to avoid jamming or inevitable interference in emergency or combat scenarios.

Conceptually, SDRs suit military applications quite well, but history may offer a slightly different view, as the early SDRs were difficult to implement. Initially, it was imperative to ensure inter-operability and reliability among many different parties, spanning multiple vendors of various components and several customers. Another potential unforeseen consequence for the JSTARs program in general was that the SDR systems were physically large, taking up space—a precious commodity in mobile vehicles.

When the military began adopting SDRs, other industries including cellular and wireless LAN grew exponentially, while largely ignoring the SDR concept. Juxtaposing the military adoption of SDR with the cellular industry’s lack of acceptance, a clear point arises. The benefits of SDR for military outweighed the higher costs and other drawbacks whereas commercial wireless industries such as cellular and WLAN had to focus on achieving cost and size targets to entice billions of users. The JSTAR decision was purpose-driven, but imagining cell phones measured in square feet rather square inches may have inhibited the utility, and ultimately attraction, of these now indispensable communication devices.

Consumer markets really did not migrate to SDRs until the mid 2000s. In 2004, Matt Ettus started a small company in the bay area called Ettus Research that began to change the minds of researchers all over the world concerning the potential of SDRs. Using open source software (GNU Radio) and off-the-shelf, commercially available hardware, Ettus created the first low-cost, widely deployed SDR called the Universal Software Radio Peripheral (USRP™). The combination of plentiful, low-cost hardware and software options made it possible for researchers to quickly prototype their ideas for communication. Hobbyists and even amateur researchers (i.e. makers and hackers) cou ld finally use an SDR and apply it to their applications. Commercial companies even started to build complete networks using the USRP. This expanded the number and capabilities of SDR developers with open hardware and software architecture, low-cost, and a thriving community of software collaborators.

Defining the SDR

A Google search today reveals the following definition from the Wireless Innovation Forum’s website

"Radio in which some or all of the physical layer functions are software defined" (2)

This overall definition rings true, but as with many descriptions that seem simple on the surface “the devil is in the details.” Let’s review the USRP architecture and why the USRP has been adopted by thousands of SDR developers around the world.

The SDR Architecture

Figure 1 below presents a modern SDR architecture that is common to the USRP and many other SDR systems commercially available today. Unfortunately, the early SDR innovators could not take advantage of some of the more modern components described in Figure 1 because the components were not feasible, not capable, or not available until recently.  

The modern SDR architecture functionally mirrors real mobile devices and/or cellular base stations by employing parallel processing architectures, and commercial, off-the-shelf RF front ends. The RF portions used in SDR are largely comprised of direct up/direct down topology also known as homodyne. From a performance perspective, RF purists may note that homodyne architectures are inferior to super heterodyne radios for which all wireless systems were based until the early 2000s. Homodyne architectures do suffer from DC offsets, high-carrier leakage, and I/Q skew—all of which negatively impact the wireless link. However, homodyne radios are smaller and cheaper than super heterodyne alternatives. Developers using homodyne radios can also improve the RF performance using software and digital signal processing techniques. Because of these attributes, silicon vendors universally have adopted homodyne architectures.

Figure 1: This figure depicts a modern SDR architecture that includes a multiprocessor subsystem, RF front end, host connection, and baseband converters.

The baseband uses analog I and Q converters where the physical layer and signal processing task are handled using an FPGA or dedicated digital signal processor (DSP). To handle control and application interfacing, microcontrollers or General Purpose Processors (GPPs) are typically used. The connection between the FPGA/DSP and the GPP/uC must match the bandwidth of the I/Q sample rate, and feature low latency to ensure real-time performance in order to respond to an external radio and sustain a communication link. This architecture reflects the USRP architecture and has been adopted by many commercial applications. The overall architecture exhibits a certain beauty in its simplicity; however, to enhance an understanding of SDRs, the focus must turn to challenges of turning this architecture into a usable system.

A Simple Architecture?

The SDR hardware architecture does look rather simple, so one may erroneously conclude that developing SDR applications is easy. The ultimate system challenge does not lie in the hardware but in the software design. Of particular note, the computational partitioning of signal processing blocks among parallel heterogeneous computational engines (for example, GPP and FPGA) that comprise today’s wireless systems can be difficult and even daunting to tackle. The software developed for the FPGA/DSP uses different tools and perhaps even different languages than the software running on a GPP. Not only is there a steep learning curve from mastering diverse tools and languages, if software blocks are not available then the SDR developer must build these from scratch. The Ettus Research use of GNU radio cannot be emphasized enough in driving rapid adoption of SDRs via an open source repository. GNU radio provides an open and comprehensive set of libraries for building communications links. And since GNU radio is open, developers gain access to the source code, which provides an optimum starting point.

The Key to SDRs is the S(oftware)

GNU radio provided a big step forward for SDRs; however the open source library did not address many other challenges in building SDR systems. GNU radio and the USRP provide a simple model where the FPGA is open/programmable but access is restricted from many developers as the tools to program the FPGA are difficult to use, expensive, or a combination of the two. The “real-time” aspect of a USRP developed system relies almost entirely on the computational capability of the host computer using a base FPGA image. For those savvy enough to delve into FPGA programming, the actual system development and integration can take time as it necessitates the use of multiple, disaggregated software development tools.

In 2010, NI acquired Ettus Research in a strategic step to help customers unlock the huge potential of SDRs. Although Ettus Research continues to operate independently and support the open source community, NI’s vision for SDRs builds on the Ettus Research strategy by focusing on the most challenging aspect of developing an SDR—the software. With the introduction of the NI USRP RIO, the vision is clear and applications that were previously difficult, expensive, or for other reasons unachievable, are now possible with LabVIEW (and especially the upcoming LabVIEW Communications software) and the USRP RIO.

To make SDRs more accessible to a wider audience, LabVIEW directly addresses the following points:

  1. Tight hardware integration with driver-level abstraction
  2. Software tools that offer simulation, target compilation, and compatibility with multiple languages/tools
  3. Heterogeneous multiprocessing capabilities to simplify algorithm deployment

SDRs Defined Definitively

The Wireless Innovation Forum’s definition of SDR really does not tell the whole story. In fact, the definition can apply to many types of hardware platforms available today, but does not fully explain the challenges of developing and deploying SDRs widely. While SDRs promise software re-configurability, this process is still effort-intensive and herein lies the opportunity to further refine the definition. SDRs with software tools that enable true software re-configurability align with Mittola’s vision outlined in the early 90’s. Ettus Research pioneered the way using open source tools and technologies. NI embraced the Ettus Research strategy and took it a step closer to realizing the full potential of SDRs using the base hardware architecture to then simplify the development with LabVIEW and increase access to the FPGA. It is not only the software deployed on the SDR, but the software development tool used to develop SDRs. In that, the definition should be expanded to the following:

Radio in which some or all of the physical layer functions are software defined effortlessly, in a functionality equivalent manner, and in a commercially relevant timeframe. 

 

Stay up-to-date with NI and SDR technology at ni.com/sdr.

 

 —James Kimery, Director of Product Marketing: RF/COMMS/SDR, NI

 

  1. Software Radio: Survey, Critical Analysis and Future Directions" 1992, Mittola
  2. http://www.wirelessinnovation.org/assets/documents/SDRF-06-R-0011-V1_0_0.pdf

 

Back to Top

Bookmark & Share


Ratings

Rate this document

Answered Your Question?
Yes No

Submit