Overview
In November 2017, I started writing my master thesis in the
Computer Engineering Group at ETH Zürich. My
thesis - officially titled “Event-based Geophone Platform with Co-detection” -
is about developing a new, state-of-the-art networked sensor platform focusing
on sensing micro-seismic activities to provide early warnings for natural hazard
mitigation. Beside extensive research and collaboration with the
Department of Geography at University of Zürich,
my work included advanced hardware design from scratch and embedded software
development.
After developing the first prototypes, initial field experiments were conducted
which showed remarkably great results. Consequently, after finalizing every
detail, the device went into production state, and currently 50 pieces were
manufactured. I am extremely proud that a device which I developed and built
from scratch within just 6 months were manufactured in those numbers and being
deployed and used this summer in the Swiss mountains.
In May 2018, the geophone platform was featured on SRF Einstein (Swiss TV show).
The part of the episode, titled “Forschung extrem: Datensammeln im Permafrost”
introduces the geophone platform that I’ve developed alongside with the wireless
sensor network deployed and operated in the Swiss mountains.
Watch the video
Features
Major hardware features
- Ultra-low power system design: as low as 29uA overall current drain from
battery (including DC/DC conversion)
- Robust hardware design for high-alpine environment and noise-resilient circuit
design
- Ultra-low power Cortex-M4 core microcontroller from the STM32L4 family
- Low-power radio module with
BOLT
- Logging to SD card with FAT32 and exFAT file system support
- Omnidirectional geophone
- Ultra-low power, always-on triggering
- Dual-side (positive and negative) triggering
- High dynamic range, high precision triggering with hysteresis
- Independent positive and negative thresholds
- Switchable, dual-stage signal amplification for maximum efficiency and
dynamic range
- Remotely configurable threshold values and amplification stage settings
- 26uA total triggering consumption with single-stage amplification at 3.0V
- 46uA total triggering consumption with dual-stage amplification at 3.0V
- High-precision 140dB SNR 24-bit ADC with integrated PGA
- Battery-operated with battery voltage monitoring
- USB and external power supply support
- Inertial measurement unit with triggering capabilities
- Temperature & humidity sensing
- Metal enclosure with IP67 rating and custom-designed mounting plate
- Custom, 3D-printed geophone clamp
Major software features
- FreeRTOS Real-Time Operating System
- Self-customized RTOS kernel for ultra-low power consumption
- High precision (sub-millisecond) time synchronization mechanism
- Ultra-fast wakeup and sampling-launch: < 3ms
- Logging to SD card with FAT32 and exFAT support
- Configuration parameters stored on SD card
- Remote on-the-fly re-configuration
- Robust, fail-safe software design
- Custom bootloader for firmware updates
- Support for OTA (over-the-air) firmware update
- Self-test and self-monitoring features
- Detailed hardware-testing feature for production: manufacture, flash, connect
to PC via USB and immediately see results of hardware-test to quickly identify
possible manufacturing and/or component issues
- Advanced RTOS tracing support for debugging
- USB device support:
- USB DFU (Device Firmware Upgrade) support: if bootloader needs to be
upgraded
- USB MSC (Mass Storage Class) support: SD card can be mounted directly to PC,
without ejecting it from the device
- USB CDC (Communication Device Class) support: for serial communication and
command-line interface
Report
Please note that the file size of full report is approximately 19MB, thus it may
take some time to download.
Read the report
Publications
Event-triggered natural hazard monitoring with convolutional neural networks on the edge
|
|
Authors: |
Matthias Meyer, Timo Farei-Campagna, Akos Pasztor, Reto Da Forno, Tonio Gsell, Jérome Faillettaz, Andreas Vieli, Samuel Weber, Jan Beutel and Lothar Thiele |
Publication: |
18th ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN), April 16-18, 2019, Pages 73–84 |
DOI: |
10.1145/3302506.3310390 |
How many climb the matterhorn? (Demo abstract)
|
|
Authors: |
Matthias Meyer, Timo Farei-Campagna, Akos Pasztor, Reto Da Forno, Jan Beutel and Lothar Thiele |
Publication: |
18th ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN), April 16-18, 2019, Pages 350–351 |
DOI: |
10.1145/3302506.3312488 |
|
|
Authors: |
Matthias Meyer, Timo Farei-Campagna, Akos Pasztor, Reto Da Forno, Tonio Gsell, Samuel Weber, Jérome Failletaz, Andreas Vieli, Jan Beutel and Lothar Thiele |
Publication: |
EGU General Assembly 2019, Vienna, Austria, April 7-12, 2019 |
DOI: |
10.3929/ethz-b-000340844 |
Pictures