This information is similar to the presentation in section 3.4 of the text and what was presented in class.

Reliable Transport Built on Top of an Unreliable Network (and lower) Layer

Concept	RDT Level	Description	Mechanisms
Reliable	1.0	Underlying channel IS reliable
Bit Errors	2.0.	Bit errors can occur in data transfer in either direction.	ACK/NAK on checksums Automatic Repeat ReQuest (ARQ) protocols Stop and Wait Problem: ACK/NAK themselves corrupt. A. Bigger checksum so can recover. Not practical. B. Resent on bad ACK/NAK but can get duplicates
	2.1	Sequence Numbers	ACKs have sequence of 0/1 so can detect duplicate
	2.2	Replace NAK with ACK of last segment.	Instead of NAK, just ACK prior segment (e.g. 0 instead of 1)
Lose Packets	3.0	Can lose packets either direction - data or ACK - not corrupt, just nothing! Therefore, must timeout and resend which raises possibility of duplicate data if ACK was lost, not original data.	Use ACK0/ACK1 Still Stop and Wait Consider performance of 1000 byte message, 1Gbps link, 8 microsec transmit time. If Coast to Coast, RTT (due to propagation) = 30ms
Pipeline		More than 1 packet outstanding awaiting an ack.	Larger sequence numbers than just 0/1. Two ways: Go Back N Selective Repeat