Data collection from embedded devices: challenges and solutions
summary
Practical roadmap for embedded data collection: clear models (internal memory sync, RAM chunking, continuous transfer), concrete mitigations for limited storage, bandwidth and power, and a deployable pattern of a multiplexer server/client plus a separate firmware-side data-collection service. Advice includes x2–x5 scaling factors, prioritizing critical frames, compression where feasible, local-network or synthetic-scene anonymization, and pushing conversion to the receiver; company-grade in-house infrastructure is used and reported successful for beta collection and CI-based regression checks.
AI-written summary · about
chapters (14)
- 0:00Intro
- 1:35What is data collection?
- 5:03Why do we need DC?
- 8:39Why DC on embedded devices is a challenge?
- 11:36Ensuring Security & Privacy
- 14:56Scaling and support of new devices
- 17:04Power source
- 19:12Data multiplexing
- 21:20Bandwidth
- 24:11Architecture & design
- 26:08DC conditions
- 27:56Relevancy of collected data
- 29:17Data conversion
- 30:40What should be done before DC
related talks
08mSenseCollect: We Need Efficient Ways to Collect On-body Sensor-based Human Activity Data!
32mI need data (for research)!
35mData science and Internet of Things on the edge
27mCollect All the Data; Protect All the Things
16mEdge Computing and next-gen of IoT Sensors
31mFinding the Needle in the Haystack: Predicting Storage Device Failures in...
22mExtending European Strength with Embedded AI
17mFrom Sensor to Value: A journey