|
|
|
Project
Design
Benefits and Drawbacks of our Project Improvements, Applications And Related Work
|
If we have a large file, it will probably occupy
more than one chunk. It is also likely that the last chunk will be incomplete,
containing fewer bytes than a normal chunk (for example 334,541 bytes). All the
algorithms above assume that the chunk is complete (i.e, 32 strips of 32 packets
each). How do we apply our algorithms to incomplete chunk? First of all we pad the chunk to contain an integer
number of packets. Second, the incomplete chunk contains fewer strips than a
normal chunk. An incomplete chunk will contain some strips of 32 packets and at
most one strip of less than 32 packets. Suppose there is such strip and it contains 'k'
packets. This strip will be encoded and decoded using special parameters. The
interleaving of the strips will work as usual, until some point at which the
incomplete strip stops and the interleaving continues without it. The ordering
of "good" (not encoded) packets inside strips remains as normal for
strips of the same number of packets (complete and incomplete). Both the encoder and the decoder must carefully
calculate all these indexes and placements.
|
|
Please contact Genady or Nir regarding copyright issues
|
