Can Your ESP32 Become a Radio Lab?

Key takeaways

• Receiving Wi-Fi packets and accessing underlying radio samples are different capabilities.
• An ESP32’s usefulness for SDR experiments depends on data access, frequency coverage, and bandwidth.
• LLMs can help interpret firmware and suggest experiments, but their explanations need verification.
• A cheap board could become a useful instrument for a specific experiment within its hardware limits.

That spare ESP32 board in your drawer already receives radio waves. The interesting question is how much of that activity you can inspect—and whether an LLM can help you find a way in.

The radio is there. Access is the question.

A Wi-Fi chip has a radio receiver because it needs to turn signals arriving at an antenna into usable data. Applications usually see the results: network traffic, connection status, and other information the chip exposes.

Earlier stages of that process are usually out of reach of ordinary application code. That is where things get interesting for radio experiments.

Software-defined radio, or SDR, puts substantial parts of radio signal processing into software. Given suitable signal data, a program can examine frequency components or decode information carried by a transmission.

Using a Wi-Fi chip this way therefore hinges on a different kind of access. You need to establish whether software can retrieve signal data from an earlier processing stage, before it becomes decoded network traffic.

The chip handling a signal internally does not guarantee that developers can extract it. Internal memory access may be necessary. So may a way to move that data out fast enough to use it.

Packets, CSI, and I/Q tell you different things

“Radio data” can mean several things, and the distinction determines which experiments are possible.

Wi-Fi packets contain data the chip has already interpreted according to Wi-Fi’s rules. Capturing them lets you examine decoded traffic.

I/Q samples represent a signal through two numerical components that together capture amplitude and phase information. They give software access to an earlier stage of signal processing, allowing a different range of analysis.

Then there is channel state information, or CSI. It estimates characteristics of the path a wireless signal has traveled. That can help investigate how the surrounding environment affects a signal, but CSI is distinct from the raw received waveform.

These distinctions matter when looking at ESPARGOS or ESP-SDR. Start by identifying exactly which data you can retrieve. Then check the supported frequency range and the bandwidth you can observe at once.

A feature can be valuable without exposing every stage of the receiver. Its value depends on the question you want the data to answer.

Let AI suggest an explanation. Make the hardware prove it.

Reverse engineering often involves staring at unfamiliar code and asking what a sequence of addresses and operations actually does.

Suppose firmware repeatedly writes values to a particular hardware address. That might relate to receiver configuration or memory transfers. Understanding the surrounding code helps narrow the possibilities.

An LLM can assist by explaining disassembled code, spotting patterns across functions, and suggesting experiments. It may reduce the amount of code a person needs to untangle manually before finding a promising starting point.

But a convincing explanation still needs evidence. Calling an unidentified routine “configure receive mode” does not establish its purpose. The model may have attached a plausible story to ambiguous instructions.

Treat the explanation as a testable hypothesis. Change one setting and measure what happens to the output. Feed in a known signal and check whether the result matches the prediction.

The useful contribution is a shorter path to an experiment. The device’s measured behavior determines whether the explanation survives.

Firmware cannot negotiate with the antenna

Software access has physical limits. Antennas, filters, and amplifiers are designed for particular frequency ranges.

A receiver built for 2.4 GHz does not become an FM broadcast receiver near 100 MHz because someone finds an undocumented setting. Hardware also shapes measurement precision and the ability to distinguish a weak signal beside a strong one.

The practical move is to define the experiment narrowly. Observing changes within a particular band and teaching wireless communication principles may require different capabilities. Combining multiple receive channels adds another challenge: keeping their timing and phase aligned.

This is where inexpensive hardware could prove useful. If a board you already own can expose the signals your experiment needs reliably, it could lower the cost of getting started.

A cheap board becomes a useful instrument when it produces data you understand and trust. For the ESP32 in your drawer, that is the capability worth looking for.