ESP-SDR: the ESP32 radio as a tool for investigating the wireless environment

In short

ESP-SDR lets you explore the integrated radio in compatible ESP32 devices as a signal receiver for software analysis. For hobbyists and wireless product developers, our view is that it opens up an opportunity for learning and bench investigation. For those maintaining firmware in the field or purchasing industrial equipment, this news does not require replacing devices or updating installations.

What it is

SDR stands for software-defined radio: part of the signal processing is handled by software, rather than relying exclusively on fixed-function circuits. Firmware is the program that runs on an embedded device. ESP-SDR is open-source firmware that uses the ESP32's own radio for this purpose, without adding an external receiver, according to the overview from CNX Software.

The novel feature is access to I/Q samples before the usual Wi-Fi processing. I and Q are two signal components, called in-phase and quadrature, that allow its amplitude and phase to be represented. Hackaday describes the use of an undocumented mode and reports that the available bands depend on the microcontroller. Therefore, the reference to 2.4 and 5 GHz does not mean that any ESP32 can now receive signals in both bands.

To understand the difference, consider the receive path. The antenna picks up electromagnetic energy, the radio circuits select and convert the signal, and digital processing attempts to recover the transmitted information. A Wi-Fi receiver normally delivers frames, the protocol's communication units. I/Q samples allow you to examine an earlier stage, even when a valid communication cannot be interpreted.

What it is for

Spectrum analysis shows how received energy is distributed across frequencies. A waterfall display adds time: each new strip in the image records another observation, and color represents intensity. This helps distinguish persistent activity from intermittent transmissions and formulate hypotheses about the radio environment.

Consider a sensor that sends temperature readings over Wi-Fi and experiences occasional delays. MQTT, a messaging protocol based on publish and subscribe, carries its readings to an intermediary server called a broker. A delay can arise in the application, the network connection, or the radio link. Observing the spectrum adds evidence about that last part, while the sensor and broker logs show how communication behaves.

One useful experiment is to compare periods of normal operation with periods of failure, keeping the antenna position and capture conditions unchanged. Coincidence between radio activity and delays guides the investigation, but establishing causality requires controlled tests. This distinction is standard diagnostic practice: observing one event close to another does not prove that one caused the other.

Where it is used

Spectrum observation is used in wireless equipment development, in assessing coexistence between devices, and in investigating coverage problems. In industrial installations, metal structures, machinery, and changes in position affect propagation. A closed control cabinet and the same cabinet with its door open can produce different reception conditions, even without any change to the sensor's program.

In an IIoT architecture, the Industrial Internet of Things, sensors and controllers carry process data to supervisory applications. SCADA is the system that combines supervision, data acquisition, and operational commands. There are several stages between the sensor and the screen; analyzing the radio helps investigate the wireless segment, while network and application logs help locate losses in the other stages.

It is also important to distinguish between physical media. RS-485 is a differential electrical interface used for wired communication. Modbus is a data exchange protocol, often used over RS-485 or Ethernet networks. Termination, biasing, or electrical reference faults on an RS-485 bus call for measurements on the circuit. A radio spectrum plot does not directly reveal these problems.

As a proposed bench setup for exploring ESP-SDR, it makes sense to use a dedicated observation board alongside the equipment being investigated. This is an application suggested by our analysis, not a report of an industrial deployment of the project. Separating the observer from the device under test allows you to compare the product's behavior without replacing its operating firmware.

Relevance for industry

Editorial view: ESP-SDR's appeal for automation lies in making investigation of the radio layer more accessible. Teams accustomed to looking only at error messages can gain another perspective on intermittent failures. Its initial value lies in producing better hypotheses and organizing tests, not in justifying an immediate infrastructure change.

Industrial diagnostics need to connect the observation to its effect on the process. It is useful to record when a reading failed to arrive, how long the device remained disconnected, and whether it restarted. These events can then be compared with observations of the wireless environment. Synchronized clocks and timestamped logs are essential for that comparison to be meaningful.

Our guidance is to experiment first on a bench setup representative of the installation. Use the same type of enclosure, power supply, and antenna position as the product whenever possible. Changing several factors at once makes it harder to identify the cause of an improvement. For a maintenance decision, the result needs to be reproducible and consistent with the other evidence from the system.

Impact on embedded systems

Editorial view: the most interesting contribution for developers is bringing embedded programming and signal processing closer together. ESP-SDR encourages looking beyond whether a connection succeeds or fails. For hardware designers, it also reinforces that the antenna, positioning, and environment are part of the product's behavior, along with the firmware.

Working with samples brings different challenges from sending sensor readings. The program needs to receive data, store it temporarily, and process or transfer it before the available memory runs out. A buffer is that temporary storage area. When data production exceeds its removal for long enough, data loss occurs, regardless of whether the processor executes each instruction correctly.

The fast Fourier transform, known as FFT, is a method for calculating the distribution of frequency components in a set of samples. Processing locally and transmitting a spectrum summary reduces the output volume. On the other hand, retaining the original samples allows other analyses later. The choice depends on the question the experiment needs to answer.

When organizing development, our recommendation is to keep the experiment separate from the production project and record the board, code revision, and parameters used. PlatformIO, a tool that manages dependencies, building, and flashing for embedded projects, illustrates the importance of reproducible environments. The same principle applies regardless of the build system adopted: someone else needs to be able to repeat the test and understand exactly what was run.

Limits and what is still missing

The main operational limitation described by CNX Software is transferring samples to the computer. The visualization presented uses short captures and may miss signals between them. An apparently quiet display therefore does not guarantee an absence of transmissions throughout the entire interval. This point matters particularly when investigating brief, sporadic events.

The supplied articles do not establish calibrated accuracy, guaranteed performance under industrial conditions, or a commitment to manufacturer support for this access mode. Nor do they demonstrate that it can coexist with a product's Wi-Fi application without changing its behavior. Among the candidate sources on this subject, there is no primary source for ESP-SDR; primary sources for other projects do not confirm its characteristics.

An experimental receiver is also not automatically equivalent to a measuring instrument. Gain, antenna response, and saturation influence the observed intensity. Saturation occurs when the signal exceeds the receiver's capacity to represent it and distorts the result. Relative comparisons can help on the bench, but absolute measurements require appropriate characterization and a suitable reference.

Our conclusion is to treat ESP-SDR as an opportunity for technical experimentation and follow the compatibility documentation and developments in capture capabilities. Before incorporating it into a diagnostic routine, it is worth verifying that it observes the type of event you are looking for and that the results are repeatable.

What to do now:

- Identify the exact ESP32 model and check its compatibility with the project. - Set aside a bench board and preserve the firmware of the equipment in operation. - Define a test question, keeping the antenna and capture conditions constant. - Compare observations with connection and message delivery logs. - Confirm critical conclusions with instrumentation suited to the problem.