What Is Software-Defined Radio (SDR), and Why Does It Matter?

At AESTECHNO, an electronic design house in Montpellier, France, we are starting a radio prototyping platform built on a Zynq-7000 FPGA and an AD936x RF transceiver from Analog Devices. That choice rests on one simple idea: software-defined radio, or SDR. Rather than freezing a radio function into components, SDR defines it in software. Here is what that means, and why it changes how a wireless product is designed. Updated May 2026.

What is software-defined radio (SDR)?

Software Defined Radio (SDR) is an architecture in which radio signal processing is done in software rather than by dedicated hardware circuits. The hardware is limited to an antenna and a minimal radio-frequency stage, and everything else becomes code.

To grasp the idea, picture a boundary. Every radio has a line between the analog part, the real signal travelling through the air, and the digital part, numbers handled by a processor. A conventional radio places that boundary as late as possible: the signal crosses a long chain of specialised analog components before it is eventually digitised. Software-defined radio does the opposite. It digitises the signal as early as possible, as close to the antenna as it can, then hands all remaining work to software. The concept is described in depth in the reference definition of software-defined radio.

The consequence is radical. The same electronic circuit, without any physical change, can become an FM receiver, a LoRaWAN transmitter, a spectrum analyser or a simple radar. You just load a different program. The hardware becomes a generic platform; the radio personality becomes a file.

Radio frozen in silicon: the problem SDR solves

A conventional hardware radio is a chain of specialised components, each wired for a single radio function. That specialisation delivers excellent performance, but it imposes rigidity: a filter tuned to one precise frequency, a demodulator built for one modulation, a decoder dedicated to one protocol.

In practice, as soon as one parameter changes, frequency band, modulation scheme or protocol version, the hardware must change. That means going back through schematic design, circuit board layout, fabrication, testing and often recertification. For a wireless product, this cycle takes weeks and thousands of euros in non-recurring engineering cost.

The problem worsens in today's context. Wireless standards multiply and evolve fast, a connected product often has to speak several protocols, and the radio bands grow more crowded. Their regulation, handled in France by the ANFR and at European level by ETSI under the RED directive, keeps tightening, with standards such as ETSI EN 300 220 for short-range devices. A frozen architecture ages badly; an architecture defined in software is updated. That is why we fold SDR into our RF board design methodology.

How an SDR works: from antenna to software

An SDR chain is a sequence of four stages: an antenna, a wideband radio-frequency stage, an analog-to-digital converter and a compute resource that runs the processing. The key point is that the analog-to-digital boundary sits as early as possible.

The radio-frequency stage is deliberately simple and wideband: a low-noise amplifier, or Low Noise Amplifier (LNA), and moderate filtering. It does not try to isolate one precise station, only to present a clean slice of spectrum to the converter. The analog-to-digital converter turns that signal into a stream of numbers, the samples. In a modern SDR these samples come as two channels called I and Q, which together describe the amplitude and phase of the signal. That representation is what lets software reconstruct everything.

The stream still has to be processed. It is too fast for an ordinary processor: a converter produces tens of millions of samples per second. That is the job of the Field Programmable Gate Array (FPGA), a reconfigurable logic device able to filter and decimate the stream in parallel, in real time. Once the rate is brought down to a manageable value, a processor takes over for the application logic: the protocol, the network interface, the display. This split between a fast FPGA and a flexible processor is at the core of any SDR platform, and explains why the choice of an FPGA is so structural.

SDR versus classic hardware radio

The choice between SDR and a classic hardware radio is a trade-off between flexibility and unit cost. The hardware radio aims for raw performance on a frozen function; the software-defined radio aims for upgradability and speed of development. The table below summarises the trade-offs.

The reading is clear. For a mature consumer product, built in very high volume with a stable radio function, the dedicated hardware radio remains unbeatable on cost. To explore, validate, cover several standards or hedge against evolving regulation, SDR takes the lead. And there is one case where it is especially valuable: the design phase itself.

What is an SDR actually used for?

Software-defined radio is a versatile tool that serves research, deployed products and, above all, prototyping. Four families of applications come up most often, and the last is the one that drives our own platform.

Multi-protocol connectivity. A single SDR device can speak several wireless standards in turn. A gateway that talks LoRaWAN, Bluetooth and a proprietary link adjusts its radio personality in software, without stacking as many radio modules as protocols. The same principles apply to NB-IoT, LTE-M, Sigfox and emerging satellite IoT links. To place these technologies, our comparison of LPWAN networks gives the full picture, from LoRaWAN to NB-IoT.

Spectrum monitoring. Because an SDR observes a wide slice of band at once, it naturally serves to detect, measure and classify the signals present in the air. It is the basic tool of spectrum analysis, interference diagnosis and radio compliance checking.

Research and teaching. SDR has opened up a field that was long reserved: you can experiment with a modulation or a processing algorithm without etching a circuit. As noted by GNU Radio, the open-source reference ecosystem, software-defined radio has become a standard experimentation ground, including for demanding standards such as IEEE 802.11 (Wi-Fi) and IEEE 802.15.4.

Prototyping and risk reduction. This is the least visible use and, for a design house, the most strategic. Before freezing a product's radio into a circuit board, an SDR makes it possible to validate the concept: does the frequency plan hold, does the chosen modulation clear the link budget, does the protocol behave as expected. You separate two questions too often confused: is the radio concept sound, and is the board well designed.

Our SDR platform: Zynq-7000 and AD936x

Our prototyping platform is a board that pairs an AD936x RF agile transceiver from Analog Devices with a Zynq-7000 FPGA from AMD. The first handles the radio, the second handles the compute; together they form the industry's reference SDR architecture.

According to Analog Devices, the AD936x is an RF agile transceiver that integrates, on a single chip, the radio-frequency stage, the frequency synthesisers and the analog-to-digital and digital-to-analog converters. It tunes from 70 MHz to 6 GHz, handles channel bandwidths up to several tens of MHz, and provides two transmit and two receive channels. In one part number, it turns a radio signal into a digital sample stream, and back.

According to AMD, the Zynq-7000 brings two worlds onto one chip: a programmable logic part, the Programmable Logic (PL), that is the FPGA, and a processor part, the Processing System (PS), made of two Arm Cortex-A9 cores. That duality is exactly what an SDR needs. The FPGA absorbs the fast stream from the AD936x and applies real-time processing; the processor runs Linux and the application layer. This AD936x plus Zynq-7000 pairing fits so well that it forms the architecture of the Analog Devices FMComms evaluation boards and of the PlutoSDR module. That proven base is what we reuse for our platform.

Field report: why a design house builds its own SDR

Radio risk reduction is a practice that consists of validating a wireless concept before committing to a printed circuit board. It may seem counter-intuitive, but a design house has good reason to build its own radio.

On a recent project to design a wireless product, in our lab in Montpellier we measured the radio-frequency stage and the digital link before any board commitment. Our measurement methodology and test procedure stay consistent on every RF project: the radio-frequency stage is measured with a Keysight VNA, the fast digital link between converter and FPGA is measured on a Tektronix TekExpress bench, then pre-compliance radiated emissions are checked against the ETSI EN 301 489 templates. Contrary to the common assumption that a radio is validated once the board is built, we observed that the genuinely risky decisions, the frequency plan, the modulation choice, the link budget, are made well before layout. We found the same pattern across project after project. Unlike a routing error, which is corrected cheaply, a concept error caught late is expensive. Despite the time it takes to build such a platform, in our practice an SDR lets us play those decisions in software, on generic hardware, before any circuit-board commitment. The field report from our 65 projects delivered since 2022 confirms it, and we recommend separating concept validation from board design to cut the number of hardware iterations and the cost of a wireless product.

Bottom line

Software-defined radio is an architecture that moves radio processing from hardware into reconfigurable software. The hardware becomes a generic platform, the radio function becomes code. Here are the five points to remember.

• Principle. SDR digitises the signal as close to the antenna as possible and hands filtering, demodulation and protocol to software.
• Benefit. One piece of hardware changes band and protocol with no physical change, by loading a different program.
• Architecture. A wideband RF stage, a converter, an FPGA for throughput and a processor for flexibility.
• Our platform. An AD936x tunable from 70 MHz to 6 GHz paired with a Zynq-7000 FPGA, the industry's reference SDR architecture.
• Key use. For a design house, SDR mainly serves to validate a radio concept before freezing a circuit board.

FAQ: software-defined radio and SDR

The questions below are the most common ones from product designers discovering software-defined radio.

What is the difference between an SDR and a normal radio?

A normal radio does signal processing with dedicated components, tuned once and for all at design time. A software-defined radio digitises the signal very early and hands that processing to software. The first is frozen, the second is reconfigurable: changing frequency or protocol is done by loading a different program, without touching the hardware.

Do you need to be an RF expert to use an SDR?

For observation and experimentation, no: open ecosystems such as GNU Radio make SDR accessible. To design a reliable product, however, the RF stage, interference management and regulatory compliance remain engineering subjects. That is precisely the role of a design house.

Why combine an FPGA and a processor in an SDR?

The sample stream coming out of the converter is too fast for an ordinary processor, which would handle it too slowly. The FPGA performs, in parallel and in real time, the filtering and decimation that bring the rate down to a manageable value. The processor then takes over for application logic and protocol, where software flexibility matters most.

What is an RF agile transceiver such as the AD936x?

It is a chip that integrates, on its own, the radio-frequency stage, the frequency synthesisers and the analog-to-digital and digital-to-analog converters. The AD936x from Analog Devices tunes from 70 MHz to 6 GHz. It turns a radio signal straight into a digital stream, which makes it the natural hardware core of an SDR platform.

Does SDR replace the classic hardware radio?

No, the two coexist. For a mature consumer product, built in very high volume with a stable function, the dedicated hardware radio stays the cheapest. SDR wins as soon as flexibility is needed: prototyping, support for several protocols, upgradability against changing standards.

Does an SDR help design a wireless product faster?

Yes, through risk reduction. By validating the radio concept on an SDR platform before designing the circuit board, you catch concept errors when they cost only a software file, not a new board. That cuts the number of hardware iterations and the overall lead time.

Related articles

• RF board design: the routing and shielding constraints specific to radio-frequency circuits.
• FPGA board design: why and how to integrate an FPGA, the compute core of an SDR.
• High-speed PCB design: mastering the fast digital link between the converter and the FPGA.
• LPWAN network comparison: the landscape of long-range wireless protocols.
• Electromagnetic compatibility: the pre-compliance every emitting product needs.