Evaluating AtomVM for Fault-Tolerant ESP32-Based Systems
summary
Measures the cost of adding Erlang-style fault tolerance by running AtomVM on ESP32 S3 (512 KB RAM) and comparing single-node vs managed failover with a PC-based manager. Finding: AtomVM base requires ≈250 KB memory and enabling distributed Erlang adds ~32 KB, failover to a standby takes about 5 seconds. Under continuous LoRa frames at 300–500 ms the node drops messages once a few extra sensor-reading or prime-checking processes run, and the watchdog finally restarts the device after ~200 added processes.
AI-written summary · about
chapters (7)
- 0:00AtomVM for IoT fault tolerance
- 3:55AtomVM for ESP32
- 8:41Fault-tolerant gateway design
- 12:42Testing Erlang robustness on ESP32
- 16:54Memory and process limits
- 17:37Future research directions
- 20:01Q&A
related talks
24mLow Power Wireless Sensors with AtomVM
54mGive your ESP32s the Gift of Serviceability
18mEMoL: Erlang Messaging and Monitoring over LoRa
38mTo RTOS, or not RTOS: That is the Question
01h 01mBoosting Software Efficiency : A Case Study of 100% Performance Improvement in ...
27mIoT Edge Computing with bare-metal Erlang
41mAzure Sphere: Fitting Linux Security in 4 MiB of RAM
29mFuture-proofed microcontroller designs with FreeRTOS