The requirements for synchronization and acknowledgments seem to imply =
that
this protocol is intended to be suitable for use over unreliable transp=
orts.
1. Must be able to detect data corruption.
2. Must be able to recover from data loss.
One way to address these requirements would be to borrow some ideas fro=
m TCP:
specifically, checksums and sequence numbers. The SDP Packet Header c=
ould
include a checksum for the packet that a recipient could use to verify =
data
integrity. Packets that failed checksum would be dropped and not
acknowledged. Acknowledgments could return the sequence number of the =
last
contiguously received good packet. Then, if synch were lost or packets=
were
dropped, the sender would back up and resend starting with the packet f=
ollowing
the last acknowledged.
-Carl
=