Selecting the right software defined radio solution for your application - Embedded.com

Selecting the right software defined radio solution for your application

SDRs are available in a wide range of performance, form factor and cost, so choosing the right SDR for your application can be a challenge.

After the technological boom of digital electronics, software defined radios (SDRs) became a popular state-of-the-art technology in radio applications, creating a niche that continues to expand. Within the SDR market, there is an immense variety of products, ranging from small portable devices for the consumer enthusiast to precise and robust equipment for large-scale organizations and critical applications. Thus, it is common to find SDRs varying from the size of a USB dongle to entire racks of SDRs, often with the form factor being proportional to the performance. Moreover, with the evolution of semiconductor technology, electronic components are now cheaper and smaller, allowing for the development of SDR products with optimal balance between performance, form factor and cost. Modern SDR manufacturers such as Per Vices are offering all these benefits combined with easy system integration and interoperability.

Product selection is not an easy task in the SDR field, and to guide you through this decision, we will discuss the basic concepts of SDR, applications, common challenges during system design, and how the right SDR can solve them.

What is an SDR?

Radio communications was for a long time dominated by analog circuitry. However, analog radio is very limited in terms of robustness, integration, flexibility, modulation/demodulation and frequency tuning. Also, analog upgrades or modifications can only be performed by changing the hardware, which usually means replacing the a lot of circuitry. The rapid development of digital technology allowed radios to be defined by software instead of analog hardware. Modern SDRs are designed to achieve high levels of flexibility, stability and configurability, using field programmable gate arrays (FPGA) as the heart of computation. FPGAs allows implementation of different modulation protocols, bandwidth capture, error correction, DSP algorithms and even artificial intelligence in the same device, and upgrades can be performed periodically by simply reprogramming the device. An SDR is divided into two primary elements: the radio front-end (RFE) and the digital backend (Figure 1).

click for full size image

Figure 1: A high-level overview of SDR architecture. (Source: Per Vices)

The RFE is responsible for all the essential analog radio functions, such as pre-amplification, mixing and anti-aliasing or anti-imaging filtering. The RFE contains both receive (Rx) and transmit (Tx) channels connect to or from an antenna via SMA connectors. In high-end SDRs, a state-of-the-art RFE can operate over a very high tuning frequency range and large instantaneous bandwidth. High-performance SDRs like Per Vices’ Cyan SDR can reach 3GHz of instantaneous bandwidth and up to 18GHz of tuning frequency, that can be upgraded up to 40 GHz. Both Tx and Rx channels are interfaced with the backend via ADCs and DACs, and one SDR can have multiple independent channels (up to 16).

The digital backend, on the other hand, performs all the signal processing, generation, data packetization and flow control, and application-level algorithms necessary for operation. It contains an FPGA with on-board DSP capabilities that is synchronized to several functional peripherals, such as clocks, power controllers, ADCs/DACs, communication interfaces and displays.

Users of SDR

Considering the configuration described in the last section, there is an immense variety of SDR products in the market. This is because each component varies greatly according to performance, size, cost and functionality. The main performance parameters that define the performance of a typical SDR are:

  • RFE: tuning range, RF performance and number of channels
  • ADC/DAC: sample-rate, bit resolution, linearity and noise
  • Clock: latency, stability, clock rate
  • DSP: support for basic processing functions, such as multiply-accumulate (MAC), floating point unit (FUP), modulation, decimation and interpolation
  • Software support: UHD, GNU Radio, Matlab, Python, C++

Due to this variety of possibilities, SDRs can be classified into three main categories:

  • hobbyist level SDRs
  • mid-range performance SDRs
  • very high-end performance SDRs.

With the evolution of small and powerful commercial off-the-shelf (COTS) components, a fourth category is arising, which combines the convenience of mid-range devices with the performance of high-end SDRs.

Hobbyist level SDRs became popular among technology enthusiasts with the advent of cheap and user-friendly digital platforms. Websites such as rtl-sdr.com constantly publish new amateur SDR projects that are pushing the boundaries of homemade electronics. Some examples are amateur radio astronomy, VCR decoding, and plane tracking. The SDR used in these projects may not be very powerful, but this shows a range of applications that can be done with an SDR the size of an USB dongle.

High-end SDRs, on the other hand, are implemented in much more critical and precise applications, which requires expensive components and software. Some examples of these applications are:

  • Radar: Applies electromagnetic radiation to locate and track the position, speed and angle of moving objects. The SDR provides flexibility for different radar applications, such as general pulse radar, synthetic aperture radar (SAR) and Doppler radar, as well as interoperability with legacy radar technologies. Radar methods that require powerful and fast processing, such as SAR, can benefit significantly from SDR technology. Figure 2 shows the basic operation of an SAR radar.

click for full size image

Figure 2: A diagram of SAR Radar operation. (Source: Per Vices)

  • GPS/GNSS: Sending/receiving GNSS/GPS signals/data and ground station tracking. SDR provides low latency, accuracy, precision and high sensitivity to improve signal uplink/downlink.
  • Low Latency Links: Uses SDR in high frequency trading (HFT), point-to-point links and distributed networks. Fast processing, low latency, and response are essential in these networks.
  • Spectrum Monitoring and Recording: Applied to monitor radio activity in restricted areas, such as prisons, military facilities and airports. Also used to enforce spectrum policy and detect illegal use of licensed bandwidth. In these applications, SDR with large bandwidths, large tuning ranges, high throughput and excellent resolution are required, to catch the weakest signals at any given frequency over a large bandwidth.
  • Medical Applications: Several medical devices, such as MRI, NMR and bioimpedance instruments, require MIMO waveform generation and signal measurement to operate. SDRs can be implemented as the base unit in these systems. An SDR like Per Vices’ Crimson TNG SDR is very convenient for these applications, providing four channels operating up to 6.8 GHz in a small form factor.
  • 5G Networks: SDRs can be applied as both part of base stations (receiving and transmission) and part of the research of new telecom algorithms, providing simulation, or test and development platforms for 5G networks.
  • Test & Measurement: In this application, the SDR is used to both generate and measure radio frequencies to test devices (in terms of EMI and RF performance) and simulate electromagnetic conditions. SDRs with high RF performance must be implemented to ensure precision and robustness. For test flexibility, large tuning ranges, bandwidth, number of channels and DSP resources are required.

Naturally, the class of the SDR will affect greatly its performance: there is an enormous gap between a USB dongle costing less than $200 and precise equipment that costs tens or even hundreds of thousands of dollars. Firstly, a high-end instrument will provide a very large instantaneous bandwidth, which requires very fast ADCs/DACs for signal conversion. Configurability is essential in high-end devices, so FPGAs supporting custom DSP are used, with several parallel channels. To manage the massive amount of data in applications like smart-grids and defense systems, high-throughput data links are required in modern SDRs: the highest throughput SDRs use qSFP ports and optical cable transceivers to stream data up to 100Gbps. Furthermore, multiple channel operation with excellent phase coherence and stability is only found in high-end MIMO SDRs.

Other differences between high-performance and hobbyist SDRs are very low noise figure, high SFDR/dynamic range, multiple management ports, large tuning range, onboard waveform storage and high-quality RFE components. Figure 3 shows an example of high-end Tx/Rx channel, using top-tier components.

click for full size image

Figure 3: Tx/Rx chains contain high quality components. (Source: Per Vices)

Although the market has been more focused on either hobbyist level SDRs and top tier expensive devices, there is a significant demand for instruments in between, combining high performance with affordable prices. This class can be called optimum range SDRs, represented by SDRs like Per Vices’ Chestnut SDR designed specifically to fulfill this demand. The Chestnut SDR provides high bandwidth, extended tuning range and high digital throughput in a 4 Rx/Tx architecture. Figure 4 shows a comparison between three typical SDRs in the market.

click for full size image

Figure 4: Comparison of various classes of Per Vices SDRs. (Source: Per Vices)

Challenges of R&D

Researchers and engineers often have a hard time developing RF systems due to limitations both internal and external to the project. The most typical and frustrating obstacle is lack of funding. This problem is typically encountered in government laboratories, which do not encourage the development of new specialized hardware and software for experiments. Also, the lack of funding prevents the acquisition of modern high-end devices, so engineers are often stuck with obsolete technology and older equipment to continue their projects.

Another issue is the so-called SWaP (Size, Weight and Power) limitation. For example, a radar system on a plane needs to be designed to operate reliably with a limited power supply, small size and weight, so that it can fulfill its function without compromising balance and space. Similarly, portable devices and small base stations for telecommunications, such as 5G networks, need to provide high data throughput and low latency using very low-power radio systems. However, the processing capabilities of the device are directly proportional to the power consumption, size and weight, so there is a strong trade-off between SWaP compliance and performance.

Unfortunately, high-performance radio applications require strong processing power. The main problem here is that high processing capabilities means expensive processors. In high-end RF applications, there are several examples: SAR radars require huge amounts of signal processing (including data compression and complex algorithms for artifact adjustments) in a very short period, beamforming/beamsteering requires highly synchronized and independent phase coherent channels, and spectrum monitoring applications need to process big chunks of data almost in real-time. Moreover, the system may need to be able to change the process scheme for security and functionality, such as channel/frequency hopping for jamming prevention and modulation switching in GPS/GNSS.

Even with a powerful embedded processor, a typical desktop computer system may not be able to handle all the data captured in certain applications. In this case, a server-like host system should be implemented. However, this introduces another problem: it is challenging to design a host system capable of capturing, storing and processing data in real-time. In very low-latency applications, this is extremely problematic, as data overflow and underflow can easily occur.

How SDRs and Custom Host Systems Can Help

Concerning the host system problem, custom systems can be developed to better comply with extremely high real-time data capture SDR applications. This usually involves building server-like host systems with a large numbers of NVMe drivers, SD cards, several GBs of RAM and proper Network Interface Cards (NICs) and/or FPGA based accelerator cards. There are also ways of reducing latency between an SDR, and the optical links to the host system for the host system, by switching to a real-time operating systems (RTOS) or applying NICs optimized for low latency transfer. In extreme cases, such as HFT networks, the SFP+ connectors can be designed to work with custom interface protocols.

Because almost every important aspect of the SDR is defined by the software, it is also very easy to integrate solutions. For instance, they can be used as simulation platforms for preliminary experiments, as they are capable of working with several different protocols and techniques before field testing. Furthermore, they can be shared between different experiments, as an SDR can be connected to a network, reducing the costs for all projects involved. Network connection also facilitates the integration and syncing of point-to-point links or the use of an SDR as an external reference for other equipment.

SDRs are also built in a modular way, which makes the SWaP more flexible for the intended application. In such cases, the SWaP can be customized by only adding the required peripherals, eliminating unnecessary components or reorientating boards into a different chassis, thus, reducing volume and weight. The portability provided by the reduced SWaP enables their transportation to difficult areas, such as observatory telescopes, antenna in mountains and deployed in space craft systems, while also acting as a storage solution for data accumulated insuch applications.

Customization also enables better compliance with the application requirements, for example, by optimizing sensitivity and dynamic range. Moreover, the FPGA can reconfigure its entire system at any time, so that the SDR can operate with different protocols, algorithms, data formats and waveforms without any hardware modification, using the programmable internal logic block interconnections (Figure 5). The FPGA can also be customized to reduce the amount of processing and memory required from the host computer.

click for full size image

Figure 5: An FPGA contains logic blocks and input/output blocks with programmable interconnect. (Source: Per Vices)

Although an SDR application is based on code, only basic knowledge of software development is required to configure and apply the device. Predeveloped code examples are incredibly useful to begin working with your SDR, and further custom software can be easily programmed in Python, C++ and GNU Radio for a variety of functionalities. GNU Radio also provides several DSP tools designed for test and measurement, modulation/demodulation protocols. Figure 6 shows the implementation of PSK demodulation), error correction and spectrum analysis.

click for full size image

Figure 6: GNU Radio can be used for PSK demodulation. (Source: https://wiki.gnuradio.org/index.php/File:Stage1.png)

Conclusion

Selecting the correct SDR and/or complete system is fundamental to get the most out of an RF-based application. However, this process can be challenging and very costly, but the availability of affordable, high performance SDRs makes this task easier. Suitable SDR models provide solutions for all application classes, from the tech enthusiast to large companies, all with state-of-the-art performance – and working with a vendor able to provide custom development for your projects lets you more easily translate your design from paper to real-life.


Brendon McHugh is the Field Application Engineer and Technical Writer at Per Vices. He possesses a degree in theoretical and mathematical physics from the University of Toronto.
Kaue Morcelles is an electrical engineer, with emphasis on electronic design and instrumentation. He currently works with biomedical research, developing instrumentation devices for tissue engineering. Learning and writing about cutting-edge technologies is one of his passions.

Related Contents:

For more Embedded, subscribe to Embedded’s weekly email newsletter.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.