Software

Information theoretic limits have traditionally been derived though manual efforts, and this task has become quite a challenge as systems become more complex. This software toolbox allows one to investigate such fundamental limits with the help of a computer. The program uses either Cplex or Gurobi as the linear program solver engine, reads in a problem description in terms of entropy relations, and then produces bounds, computes convex hull, and generates proof directly.

• C. Tian, James S. Plank, and Brent Hurst, “An Open-Source Toolbox for Computer-Aided Investigation on the Fundamental Limits of Information Systems“, technical report and user manual, Oct. 2020.
• C. Tian, James S. Plank, Brent Hurst, and Ruida Zhou, “Computational techniques for investigating information theoretic limits of information systems“, MDPI-Information (Invited), Feb. 2021

The software developed for this work can be found here. Some highlights of this package:

• A simple problem description file syntax;
• Three basic functions: compute a bounding plane, compute the convex hull, and produce a proof;
• Taking advantage of symmetry and dependence reductions explicitly for improved efficiency.

We investigated a set of acceleration techniques for erasure coding and their combinations to identify an approach to optimize bitmatrix and perform fast erasure coding. The result was reported in the following paper.

• T. Zhou and C. Tian, “Fast erasure coding for data storage: A comprehensive study of the acceleration techniques,” The 17th USENIX Conference on File and Storage Technologies (FAST ’19), Feb. 2019.

The software developed for this work can be found here. Some highlights of this package:

• Implemented XOR-based vectorization for hardware acceleration;
• Simulated annealing and genetic algorithm to optimize the bitmatrix;
• Higher throughput for most encoding parameters than Jerasure and ISA-L.