A wireless modular platform for neuro-behavioral recording and closed-loop manipulation in small animals

The Wireless, Interactive, Lightweight Datalogger (WILD) device is an ultra-lightweight multimodal neurologger for local neural data logging, behavioral sensing, onboard processing, and responsive stimulation in freely behaving animals.

64-Channel electrophysiology Closed-loop DSP 9-axis IMU USV audio Camera
WILD wireless modular datalogger device
The WILD device integrates local neural recording, behavioral monitoring, real-time embedded processing, microSD storage, and wireless control for naturalistic systems neuroscience experiments.

Current Public Workflow

Stable public scope: 64-channel local-storage WILD workflow, WILD_console on Windows, validated release images, and post-processing through documented MATLAB and Python scripts.

Not the current public model: full-bandwidth BLE telemetry or arbitrary runtime AI-model upload.

Release Record

Use the latest GitHub release and record the exact release tag, release image filename, hardware revision, WILD_console version, SD card, and battery in every experiment note.

Run a first recording

Hardware Setup -> Data Acquisition -> Data Analysis

Build hardware

PCB -> Mechanical -> Power

Analyze data

Data Format -> MATLAB -> Python -> Spike Sorting

Cite or reproduce a result

Publications -> release metadata -> reproducibility notes -> analysis script version

Embedded control at the headstage

The MCU, IMU, and board-level control electronics sit close to the animal, reducing external cabling while keeping timing-sensitive acquisition local.

Real WILD device with labeled functional regions
MCU and IMU

Key Features

The WILD device is built around local neural logging, responsive stimulation, embedded processing, multimodal sensing, and long-duration experiments.

Local Neural Logging

High-bandwidth electrophysiology is recorded locally to microSD, while BLE supports setup, synchronization, status, and low-bandwidth preview.

Responsive Stimulation

Onboard biomarker detection and behavioral classification can trigger stimulation without requiring continuous wireless data streaming.

TinyML on Edge Devices

Curated models run on the MCU through validated release images with RAM, timing, and closed-loop latency reviewed before deployment.

Multimodal Sensing

Neural signals, IMU, auxiliary inputs, ultrasonic audio, camera, and digital events.

Long-term Recording

Local storage, low power operations, and robust export tools for naturalistic studies.

Device Specification Summary

Representative WILD device operating modes show how channel count, aggregate sample rate, sensing modules, and onboard processing affect power draw.

Power (mW) Channels Aggregate samples/s Mode Weight (g) Video Audio Motion sensor Processing
0.1100Sleep1.48----
24.4284 x 10^4Ephys, IMU1.48--Accelerometer, gyroscope, magnetometerSpectral power, TinyML
194.04642.6 x 10^6Ephys, IMU1.48--Accelerometer, gyroscope, magnetometerSpectral power, TinyML
208.67642.92 x 10^6Ephys, IMU, audio1.48-160 kHzAccelerometer, gyroscope, magnetometerSpectral power, TinyML
245.28645.48 x 10^6Ephys, IMU, audio, video1.48320 x 320 px, 16 Hz160 kHzAccelerometer, gyroscope, magnetometerSpectral power, TinyML

Values summarize measured or configured device operating modes. Report the exact hardware revision, release image, battery, SD card, enabled modules, and whether aggregate samples/s is summed across recorded streams when comparing runtime across experiments.

System Overview

A typical session moves from the animal-mounted WILD device to local storage, synchronization, and offline analysis.

Animal
Device
Sensors
Storage
Synchronization
Analysis
WILD system diagram

Research Highlights

Outdoor recordings

Local storage, low-power modes, and wireless control support recordings in large naturalistic environments where tethering is impractical.

Multi-animal experiments

Wired sync, BLE calibration, and clock correction workflows coordinate multiple WILD devices and external systems.

Ripple detection

DSP filters include ripple-band detection for closed-loop hippocampal experiments.

Theta phase stimulation

Hilbert and filter modes support phase-aware online control designs.

Naturalistic behavior

IMU, video, USV, and neural data support studies of social interaction, outdoor navigation, and other freely expressed behaviors.

Open hardware iteration

The repository contains WILD device hardware files, validated release assets, WILD_console installers, and analysis scripts.

Quick Start

Hardware setup

Connector, battery-polarity, microSD, probe-cabling, sensor-cabling, and release-image checks come before the first recording.

Data acquisition

WILD_console handles BLE connection, synchronization, recording configuration, closed-loop settings, and local logging startup.

Data analysis

Exported recordings are converted into Intan-style files, IMU and camera outputs, and spike-sorting inputs.

Recommended first success target: complete a short bench recording, export the folder, confirm `amplifier.dat`, `analogin.dat`, `time.dat`, `info.rhd`, and `CE_params.bin`, then open the result with the documented MATLAB or Python workflow.

Publications and Citation

Platform manuscript

The WILD platform manuscript is currently under review.

Repository and release versions

Report the hardware revision, release image, WILD_console version, and analysis scripts used for each dataset.

@article{zhao_wild_neurologger,
  title        = {A wireless modular platform for neuro-behavioral recording and closed-loop manipulation in small animals},
  author       = {Zhao, Zifang and Chang, Hongyu and Paudel, Praveen and Park, Jaehyo and Liu, Can and Aurelio, Maria Q. and Oliva, Azahara and Fernandez-Ruiz, Antonio},
  year         = {2026},
  note         = {Under review}
}

Community

Issues and Pull Requests

Use GitHub issues for reproducible bugs or documentation gaps, and pull requests for tested fixes, examples, and compatibility updates.