The average number of excess packets needed for successful decoding in
a randomly pre-computed set of checksum packets of constant degree
EG
2006-04-07
For a set of 30 source packets with the method described in 060406-setof-checksums [ ch | us ] one can generate various variants of sets of 90 checksum packets of constant degree. We consider the degrees 3, 5 and 7. For each degree we generate correspondingly 1000 sets and sort each sequence of 1000 sets according the average number of excess packets needed for the successful decoding (assuming that the receiver receives the checksum packets of the given set in a random order). The curves of the average numbers of excess packets are plotted together for each sorted sequence of 1000 sets:
[ xls | d3.xls | d5.xls | d7.xls ]
Related links:
- The C++ program [ cpp | cpp.txt | csv ]
- Classes and functions [ recv.cpp | recv.h | graph.cpp | graph.h | linear.cpp | linear.h | lt.cpp | lt.h ]
- Download all files [zip]
- MS Word version of this page [doc]
- Random linear fountain code versus pre-computed collection of “good” checksum packets of constant degree [ ch | us ]
* * *
© Switzernet (www.switzernet.com)