PC Rule Grouping

This is my graduation project @NSLab, Tsinghua.

The goal is to provide a rule grouping algorithm for packet classification (PC), to improve the PC performance.

---

The main script is grouping.py under bin\ directory.

Usage:

$ ./bin/grouping.py [-h] [-n MAX_GROUP_NUM] [-m MEMORY_SIZE] [-o OPTIMIZE_RATIO] [-s SAVE_RESULTS] [-v] <ruleset> {bitcuts,efficuts}

positional arguments:
  <ruleset>               the ruleset to load
  {bitgrouping,efficuts}    grouping algorithm selected

optional arguments:
  -h, --help            show this help message and exit
  -n MAX_GROUP_NUM, --max-group-num MAX_GROUP_NUM
                        maximum number of groups
  -m MEMORY_SIZE, --memory-size MEMORY_SIZE
                        expected memory size of data structures
  -o OPTIMIZE_RATIO, --optimize-ratio OPTIMIZE_RATIO
                        the decreasing ratio when optimize the memory size,
                        only works when --memory_size is set
  -s SAVE_RESULTS, --save-results SAVE_RESULTS
                        the file to save the group results
  -v, --verbosity       output the running log of bitcuts algorithm

Note: Using pypy to accelerate.

---

e.g.

$ pypy bin/grouping.py -n 10 -m 39 -o 0.8 -s result/ipc1_1K_grp ruleset/ipc1_1K bitgrouping

This command will use BitGrouping to group the rules from ruleset ruleset/ipc1_1K and save the grouping result into result/ipc1_1K_grp. The maximum grouping number is 10. Optimization is enabled and the memory size is expected to be smaller than 39KB. The decreasing ratio is set to 0.8.