An asynchronous wireless network for capturing event-driven data from large populations of autonomous sensors.
Jihun Lee1, Ah-Hyoung Lee1, Vincent Leung2
1School of Engineering, Brown University, Providence, RI USA.
Nature Electronics
|May 13, 2024
Summary
This study introduces a novel wireless radiofrequency network for scalable biomedical sensor data collection. The system efficiently captures event-driven data from numerous autonomous microsensors, enabling advanced neuromorphic computing analysis.
Area of Science:
- Biomedical Engineering
- Wireless Sensor Networks
- Neuromorphic Computing
Background:
- Scalable networks for wearable/implantable biomedical sensors are challenging.
- Existing methods struggle with large populations of distributed microsensors.
Purpose of the Study:
- To develop a scalable wireless network for sparse, event-driven data from distributed autonomous microsensors.
- To demonstrate the integration of spike-based wireless communication with neuromorphic computing for biomedical data analysis.
Main Methods:
- A spectrally efficient, low-error-rate asynchronous networking concept using code-division multiple-access.
- Experimental demonstration with dozens of submillimeter silicon microchips.
- Large-scale in silico simulations.
- Spiking neural network (SNN) machine learning model for decoding neural data.
Main Results:
- Successful demonstration of wireless network performance with multiple microsensors.
- In silico simulations validated network scalability.
- Accurate prediction of primate hand movement using SNN to decode neural data from 8,000 neurons.
Conclusions:
- The developed wireless network approach is effective for large-scale, distributed biomedical sensing.
- Neuromorphic computing, specifically SNNs, can effectively process spike-based wireless data for complex tasks.
- This technology holds promise for advanced wearable and implantable biomedical applications.
