Publications

Academic Publications

2024

[1] Binary Folding Compression for Efficient Software Distribution
J. Li, Q. Li, H. Wan, C. J. Xue
Proceedings of the 39th ACM/SIGAPP Symposium on Applied Computing (SAC), 2024

• Developed novel binary compression technique reducing software size by 15-30% through pattern folding
• Designed pattern recognition algorithms to identify recurring patterns across multiple executables
• Implemented shared extracted pattern code mechanism allowing binaries to reuse compressed components

2023

[2] Shared Dictionary Compression for Efficient Mobile Software Distribution
J. Li, H. Wan, C. J. Xue
IEEE 29th International Conference on Real-Time Computing Systems and Applications (RTCSA), 2023

• Created cross-file redundancy detection for iOS apps using LLVM bitcode distribution format
• Built shared dictionary compression framework exceeding traditional per-file compression limits
• Addressed critical App Store and Google Play size restrictions for mobile applications

2020

[3] Acclaim: Adaptive Memory Reclaim to Improve User Experience in Android Systems
Y. Liang, J. Li, R. Ausavarungnirun, R. Pan, L. Shi, T.-W. Kuo, C. J. Xue
USENIX Annual Technical Conference (ATC), 2020

• Designed Foreground Aware Eviction (FAE) to relocate memory pages from background to foreground apps
• Implemented Lightweight Prediction-based Reclaim (LWP) using ML to optimize reclaim patterns
• Achieved 40% reduction in app launch latency and 60% decrease in page re-faults

[4] Differentiating Cache Files for Fine-grain Management to Improve Mobile Performance and Lifetime
Y. Liang, J. Li, X. Chen, R. Ausavarungnirun, R. Pan, T.-W. Kuo, C. J. Xue
12th USENIX Workshop on Hot Topics in Storage and File Systems (HotStorage), 2020
Extended to FAST 2022 as CacheSifter

• Analyzed heterogeneous cache file behaviors impacting flash storage lifetime in Android
• Developed online classification algorithm for hot/warm/cold cache files with optimized eviction
• Extended to CacheSifter, reducing flash writebacks by 70% while maintaining cache hit rates