Welcome to Adobe GoLive 6

Sliding Window Protocols

Overview

Protocols without any form of acknowledgement have reliability problems, but stop-and-wait and automatic request protocols suffer from very poor utilization. The solution is a type of protocol commonly known as a sliding window protocol. Most reliable protocols are sliding window protocols of one type or another, and they are very similar in structure.

Objectives

Understand the structure of sliding window protocols.
Be able to objectively evaluate the alternative ways of implementing sliding window protocols.
Be able to calculate the utilization of a sliding window protocol.
Understand the properties of byte-based sliding window protocols.

Preparation

Read section 2.5 in the text and the notes.

Inquiry

Why do we need something like a sliding window protocol?
What exactly is the window in an SWP?
What are the components of an SWP and how do they affect the packet format?
How does an SWP handle each of the four reliability issues? Is there any of them that it cannot resolve completely?
What are the issues in choosing a window size?
What are the issues in choosing a sequence number size?
What are the ways that error recovery can be handled in an SWP?
How do byte-based SWP's differ from buffer-based SWP's?
How would you implement an SWP?

Knowledge, Comprehension & Problem Solving

Terms

Acknowledgement number
Byte-based SWP
Flight
Go-back-N
Last ack received
Last frame acknowledge
Last frame sent
Maximum sequence number
Next frame expected
Piggybacking
Selective repeat
Sequence number
Window size
Sequence number wraparound

Questions

Using a finite-state automata description, describe the protocol that you use to get a can of soda pop out of a machine. Is the protocol complete and consistent, so that it always recovers from any error and ends in reasonable state.
Suppose that you have an SWP which uses 3 bit sequence numbers and a window size of 4 on each end. Simulate the behavior of this protocol for 10 packets where the following things happen:
1. Message 0 has no problems
2. Message 1 is lost once.
3. The ACK for message 2 is lost.
4. Message 3 is lost twice.
5. Message 5 is delayed and doesn't arrive until after message 7.
6. The ACK for message 6 is delayed and doesn't arrive until after the ACK for message 8.
Start with a sender and receiver with 4-bit sequence numbers and window sizes of 4.
1. The sender has LFS = 9, LAR = 6 and the timer for frame 7 expires. What does it do?
2. The receiver has NFE = 14, LAS = 12 and a frame with sequence number 15 arrives. What does it do?
3. The receiver has NFE = 14, LAS = 12 and a frame with sequence number 0 arrives. What does it do?
4. The sender has LFS = 9, LAR = 6 and an ACK for frame 8 arrives. What does it do?
5. The sender has LFS = 9, LAR = 6 and an ACK for frame 10 arrives. What does it do?
6. The receiver has NFE = 14, LAS = 12 and a frame with sequence number 13 arrives. What does it do?

Analysis, Synthesis & Evaluation

What do you gain by adding a NAK to an SWP (If a frame arrives with an error in it or out of order, send a NAK so that the sender hears about the problem sooner)? What do you lose?
What are the advantages and disadvantages of the 1-buffer receiver model?
What is the difference between an SWP operating at the Data Link Layer and one operating at the Transport Layer?
Why would you choose Go-Back-N error recovery instead of Selective Repeat and vice versa?
Is it reasonable for the receiver in an SWP to use a window of a single message buffer even though the sender is using a larger window? Explain?
What are the advantages and disadvantages of using byte-based SWP?