In research, simulation is king. It’s predictive, repeatable, can be performed with a standard computer at the lab or at home, and can be shared between peers. This holds true for research related to OFDM: you can simulate different delay spreads to observe inter-symbol interference effects, or run modulation schemes through pre-distortion algorithms to quantify peak-to-average power ratio improvements. However, when you require empirical results, you have to generate actual analog signals, and to do that you need specialized hardware.
Characteristics of Vector Signal Generators
One option that’s commonly considered is to use a vector signal generator (VSG). Offered by companies such as Rohde & Schwarz and Agilent, these products offer out-of-the-box OFDM signal generation. With the touch of a button, you can, for example, output sequences of WiMax or LTE bursts in RF. You can also easily change modulation types and add RF impairment (for example, Rician channel, Doppler effect, and so on). Since they come from the instrumentation world, they provide excellent analog performance. As you would expect, however, their high cost reflects their instrumentation-grade quality. Another major drawback is the limited functionality that’s available. If you need to output different sequences of bursts, or generate different variations of the standard modulations, VSG systems will be of no help, since their core functions are encapsulated in the firmware, and little to no access is provided.
SDR Offers a Flexible Alternative
A more versatile approach is to use software defined radio (SDR) systems, since some of them have the capability to play back signals, acting like an arbitrary waveform generator. Using the playback functionality, you can pre-generate baseband sequences (I&Q signals), which are derived from standard simulation tools like MATLAB®, and stock them in memory. From there the SDR reads them, converts them to analog signals, and feeds them through its RF chain. Since it’s the researcher who generates the sequences, there’s complete control over what the radio outputs.
Selecting the Right SDR System
The playback functionality combines the generic nature of SDR systems with the simplicity of use of an arbitrary waveform generator. Working with pre-generated sequences means you don’t need to write any embedded code (assuming that playback functionality is provided out-of-the-box by the vendor). However, you must be careful selecting an SDR system, as some may impose important limitations on the bandwidth or duration of the signals to be played. If the sequences are to be stored on a PC and streamed to the radio, the interface that links the two must sustain a transfer rate equal to or higher than:
number of channels x sampling rate x number of bits per sample x I&Q
To circumvent the potential bottleneck of the PC-to-radio interface, it’s possible on some systems to use the memory located on the radio itself (for example, SDRAM connected to an onboard FPGA). While transfer rates are no longer a limitation, the size of the embedded memory dictates the maximum duration of playback.
SDR for Waveform Recording
Since most SDRs are two-way radios, it’s also possible to perform the reciprocal of arbitrary waveform generation: waveform recording. The main purpose of recording waveforms is to perform offline experiments with real RF signals that were collected at a specific time. One example of such use would be evaluating the performance of various equalizers with a common set of signals taken from a real-world environment.
SDR Can Take the Place of VSG
In addition to being a system that can act as both an arbitrary waveform generator and a waveform recorder at a fraction of the price of a VSG instrument, an SDR also has the ability to process signals in real-time. To that end, it’s equipped with powerful processors to execute the code you’ve programmed it with. You can therefore transition from a static record-and-playback environment to a dynamic protocol stack that runs in real time. This aspect is particularly interesting for research related to hybrid ARQ, cognitive radio, or dynamic spectrum access, for example, where the radio must react to external elements such as missed packets over a two-way link or congestion due to spectral occupation. For all these reasons, an SDR system provides an excellent substitute for a VSG, giving you more flexibility and better functionality.