Engineers Build Appendable NibbleRun Compression Algorithm
Engineers used an LLM-guided process to design an appendable compression format for temperature and humidity time-series, achieving roughly 53.6x storage reduction. The NibbleRun format combines delta and nibble-coded RLE to exploit frequent zero-deltas and supports O(1) appends and in-memory serving for thousands of devices. Benchmarks compared TSZ, PCO L4 and custom schemes, highlighting trade-offs in speed, appendability, and serialization overhead.
Key Points
- 1Introduces NibbleRun achieving about 53.6x compression for 5‑minute temperature/humidity time series
- 2Exploits frequent zero-deltas with nibble-coded RLE to prioritize common-case compactness
- 3Enables O(1) appendable writes and in-memory serving, reducing rewrite overhead for mobile queries
Scoring Rationale
Practical, well-benchmarked compression design with high applicability + limited novelty outside IoT and single-source validation depth.
Sources
Public references used for this report.
Practice with real Hotels & Lodging data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Hotels & Lodging problems
