The simulation encodes a source file into a live stream of randomly degree-selected XOR droplets, drops a configurable fraction of them to mimic packet loss, and animates belief-propagation decoding as degree-one droplets unlock chained recoveries.
Set the source block count and channel loss rate, then start the droplet stream and watch the decoding graph light up as blocks are solved, pausing anytime to inspect which droplets are still unresolved.
Sliders for source block count and packet loss rate, plus play, pause, and step controls for the droplet stream and decoding graph.
Because a fountain code encoder can generate droplets forever, the same encoded stream can simultaneously serve millions of receivers with completely different loss patterns, and each one still only needs to collect roughly the same small number of droplets to fully recover the file.
The simulation encodes a source file into a live stream of randomly degree-selected XOR droplets, drops a configurable fraction of them to mimic packet loss, and animates belief-propagation decoding as degree-one droplets unlock chained recoveries.
The simulation encodes a source file into a live stream of randomly degree-selected XOR droplets, drops a configurable fraction of them to mimic packet loss, and animates belief-propagation decoding as degree-one droplets unlock chained recoveries.
Set the source block count and channel loss rate, then start the droplet stream and watch the decoding graph light up as blocks are solved, pausing anytime to inspect which droplets are still unresolved.
Because a fountain code encoder can generate droplets forever, the same encoded stream can simultaneously serve millions of receivers with completely different loss patterns, and each one still only needs to collect roughly the same small number of droplets to fully recover the file.