PReGO: A Generative Methodology for Satisfying Real-Time Requirements on COTS-Based Systems: Definit
summary
PReGO is a generative methodology that turns a legacy search-and-rescue app on an Apalis TK1 into a schedulable system for a fixed‑wing drone by producing profiling code, measuring task WCETs (using PowerProfiler), then generating a partitioned, no‑migration, fixed‑priority, non‑preemptive schedule checked with Casini’s analysis. A handcrafted worst case image with 500 boats drives WCET extraction; baseline runs use about 3.8 J/frame and miss the 0.5s deadline, taming Linux stabilizes timing but raises energy, and the generated deployment eliminates deadline misses, bounds memory growth, and reduces per‑frame energy.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:25An UAV to Save Lives
- 0:39Requirements
- 1:27System Overview
- 2:13Hardware Specifications
- 2:42Operating Environment
- 3:05PreGo Outline
- 3:54Worst-case scenario
- 5:22First run
- 7:12Taming Linux: results
- 7:51Task identification
- 8:38Re-engineering
- 9:35Code Generation
- 10:31Profiling & Scheduling
- 11:55Final run
- 12:27Future work
- 12:47Take away
related talks
28mGo Make Something Real – The Potential for Go on the Factory Floor
35mProactive Energy-Aware Programming with PEEK
01h 03mOLD VERSION: Building Abstractions at the Hardware-software Boundary
38mAn embedded soft real time system in Go
01h 12mPower, Energy and Analysis
20mEarly performance estimation using virtual prototypes and open-source SW
34mPREEMPT_RT isn't Just for Lasers: The Perfect Match
22mDemo: From Blueprint to Reality: Navigating SoC Tradeoffs, IP, and Ecosystem