Please refer to the PSTA article. This repository contains coder prototypes in Perl5 and test data used in the article.
To run all tests found in the article, you should also download test data from EMBED project data site http://embed-dataset.org. Download the following three files with "Ground truth plug load consumption" data:
% wc -c *.zip
26825002 Apt1_GT_Plug.zip
54172382 Apt2_GT_Plug.zip
28561226 Apt3_GT_Plug.zip
109558610 total
% md5sum *.zip
eb96d8b967727e71aa5ae03979a81940 Apt1_GT_Plug.zip
78f8ec041ba45afa07c430fbad140b3b Apt2_GT_Plug.zip
6fb3663e4422a201ec704915326b9c20 Apt3_GT_Plug.zip
Say make to run all tests. This is known to work in Debian 10
Buster on AMD64, but hopefully any GNU/Linux on any CPU architecture
will do. After some hours (yes: EMBED data are bulky while coder
prototypes are not written for efficiency) you should obtain results
in grandtab-1.tex and grandtab-2.tex. If you have
pdflatex installed, make will also build
grandtab.pdf containing formatted table of results.
make outputs command lines for each test as it runs, e.g.
cat 'embed/Apt1_GT_Plug/Iron.csv' | embed/from-csv.pl | ./rlgr.pl -18 -k10 -L256
If you execute the command line outside of make you will see
data points being compressed, the codewords generated and some
auxiliary output specific for each coder.