From: Janet Davis (jlnd_at_cs.washington.edu)
Date: Thu Jan 22 2004 - 16:51:23 PST
Here are a few hints on homework 1:
Here's a hint for getting started on problem 3. Page 91 says, "a pair of
single bit errors, one of which increments a word, one of which decrements
another word by the same amount, will go undetected." So, not all two-bit
errors are detected. What does this look like at the bit level? How can
you characterize the probability of one bit error occuring in the right
place to cancel another out?
I was asked whether I want to see numeric probabilities for problem 3. I
do want to see numbers, but they can be pretty approximate. To make your
computation easier, you can make simplifying assumptions such as
- Any bit in the data is equally likely to be a 1 or a 0.
- If there are two single-bit errors, they occur in different words.
- A burst error flips all of the bits in a single word.
Be sure to state what your assumptions are.
For problem 5(a), you may assume that no packets are lost, and that the
sender always sends a new packet after receiving an acknowledgment.
Also, you should look at the steady-state throughput, in the middle of the
connection rather than at the beginning or end.
For problem 5(b), you may assume that the TCP receiver consumes data as
quickly as it arrives. However, the sender doesn't know that the receive
buffer isn't filling up; it only knows the advertised window on the last
acknowledgement it received and the amount of unacknowledged data. You may
find the description of flow control in section 5.2.4 helpful.
Cheers,
Janet
-- Janet Davis jlnd_at_cs.washington.edu http://www.cs.washington.edu/homes/jlnd/ _______________________________________________ Cse461 mailing list Cse461_at_cs.washington.edu http://mailman.cs.washington.edu/mailman/listinfo/cse461
This archive was generated by hypermail 2.1.6 : Thu Jan 22 2004 - 16:51:26 PST