Software-Defined Radio: Principles and Applications

Software-defined radio (SDR) has transitioned from primarily a research topic to a common tool used by practitioners in many fields of RF. Made possible through hardware integrations, economics, and the growth of processing power. Making SDR devices accessible to the masses and providing a high configuration ceiling for advanced users.

In this talk, we will discuss the evolution of SDR and the key driving forces and methodologies behind both the hardware and software ecosystems. We will provide insights into the philosophies of current and historical SDR architectures. This will include several examples and demos with modern devices like transceivers and RF data converters. Throughout the different discussions, the talk will focus on the usage models and practical workflows that have been developed over the years to leverage the flexibility of SDRs. Touching upon and connecting the complex tooling for FPGAs, embedded processors, and high-performance x86 that are used in applications like communications, radar, and instrumentation. Finally, an outlook upon the next frontier of SDR in the space of Direct-RF and its related challenges will be discussed.

In general, this talk will provide a solid foundation to those new to SDR and provide context for the next generation of solutions for seasoned professionals. The talk will also connect the dots among the sea of software and hardware that has flourished over the last few decades for SDR.

What this presentation is about and why it matters

This talk surveys the practical landscape of modern software-defined radio (SDR): how radios moved from analog boxes and offline analysis to highly integrated transceivers, how the software and hardware pieces fit together, and what the next generation — "direct‑RF" — looks like. For engineers in signal processing or RF systems, the subject matters because SDR blurs traditional boundaries between analog RF front ends, high‑speed data converters (ADCs/DACs), FPGAs, and host software. Understanding those tradeoffs is essential when you design a receiver or transmitter, choose hardware for a project, or move algorithms from simulation into real-time embedded systems.

Who will benefit the most from this presentation

• DSP engineers who are starting to work with real RF hardware and need to map theory to practice.
• RF engineers who want to understand how digital techniques (calibration, digital downconversion, etc.) change front‑end design.
• Students and researchers using SDR platforms (PlutoSDR, USRP, BladeRF, etc.) for prototyping communications or radar systems.
• Embedded/Firmware engineers who must partition work across FPGAs, microcontrollers, and host processors.
• Hobbyists who want to move beyond GUI listeners to building reproducible waveforms and experiments.

What you need to know

This talk assumes a basic grounding in signals and systems; the following concepts will help you get more from the presentation:

• Frequency translation / mixing: mixing shifts signal spectra. For real mixers: $\cos(\omega t)\cos(\omega_0 t)=\tfrac{1}{2}[\cos((\omega+\omega_0)t)+\cos((\omega-\omega_0)t)]$, so a single multiplication creates sum and difference components. Complex mixing via $e^{j\omega_0 t}$ produces a single-sided shift in the complex baseband view.
• I/Q sampling and complex baseband: zero‑IF receivers use separate I (in‑phase) and Q (quadrature) channels to preserve phase. A complex sample represents amplitude and phase and lets digital algorithms demodulate modern waveforms.
• Receiver architectures: know the tradeoffs between superheterodyne (one or more analog IF stages with image filtering), low‑IF/zero‑IF (single conversion into low or baseband with digital correction), and direct‑RF (sample at RF, move all filtering and mixing into digital domain).
• ADC/DAC performance limits: sampling rate, effective number of bits (ENOB), input bandwidth, and dynamic range determine what you can digitize. Direct‑RF pushes ADC sampling into gigasamples/second and increases processing needs.
• Aliasing and Nyquist zones: aliasing is usually undesirable, but direct‑RF systems sometimes exploit higher Nyquist zones intentionally. Understand how sampling rate places signals into folded frequency bins.
• IQ imbalance, DC offsets, LO leakage: nonidealities in mixers and converters produce images and bias terms. Modern transceivers use on‑chip calibration, digital correction, or state machines to mitigate those effects.
• Digital downconversion and decimation: after a wideband ADC you typically use digital mixers, filters, and decimators to extract narrowband signals and reduce data rates sent to the host.
• Software and tooling: high‑level prototyping (GNU Radio, MATLAB, Python) vs. real‑time embedded implementations (FPGA HDL, HLS, generated IP). Know when you need precise timing (FPGA) versus algorithm exploration (host).

Glossary (terms used in this talk)

• FPGA (Field-Programmable Gate Array): A reconfigurable digital device made from programmable logic blocks, interconnect, and I/O that can be programmed to implement custom hardware.
• HLS (High-Level Synthesis): A design flow that converts higher-level behavioral code into synthesizable register-transfer-level hardware.
• Aliasing: Distortion that occurs when frequency components above Nyquist are indistinguishable from lower-frequency components after sampling.
• Software-Defined Radio (SDR): A radio architecture in which functions traditionally performed by analog hardware, such as mixing, filtering, and modulation, are implemented in software or programmable logic.
• Mixer: A nonlinear circuit or algorithm that multiplies signals to shift spectral content, producing sum and difference frequency components.
• Local Oscillator (LO): A stable reference sinusoid used to translate signals in frequency during mixing.
• I/Q (In-Phase / Quadrature): Two orthogonal signal components used to represent complex-valued baseband samples and preserve both amplitude and phase information.
• Zero-IF (Direct Conversion): A receiver architecture that converts a signal directly to baseband, typically using I/Q demodulation.

Toolbox (mentioned in this talk)

• GNU Radio: An open-source software framework for implementing and prototyping software-defined radio signal-processing chains. It provides blocks, runtime infrastructure, and tools for building streaming DSP applications.
• MATLAB: A high-level numerical computing environment widely used for signal processing, data analysis, and algorithm development.
• PlutoSDR: A software-defined radio development platform based on Analog Devices hardware. It is commonly used for experimentation, prototyping, and educational projects.
• USRP: A family of software-defined radio hardware platforms used for prototyping and research. USRP devices provide flexible RF front ends and are widely supported by SDR software ecosystems.
• BladeRF: A family of software-defined radio hardware platforms used for experimentation and prototyping. It integrates RF transceivers with host connectivity for flexible SDR applications.

Final thoughts

This overview is AI-generated from the session transcript. Spot an issue? Let us know.