This file describes a possible project based on the simulation of
go-back-n in these pages.
Compute;
SerD = Serialization Delay of Data Frame.
SerA = Serialization Delay of ACK Frame.
Prop = Propagation Delay.
W = Window (in frames).
It makes sense that SerA < SerD. Choose your numbers so that this holds.
(A)
Find a mathematical expression that gives the ``long range'' ``goodput''
of the system under the assumption that the drop_probability and the
damage_probability are zero, and the time-out interval is large enough
so no time-outs occur.
Hint:
Consider the two cases
(i) W*SerD < SerD + 2*Prop + SerA and
(ii) W*SerD > SerD + 2*Prop + SerA .
Verify that the expression is correct by running a few simulations.
What is the best value of W and time_out given Prop, SerD, SerA,
if there is no random loss or damage?
(B)
Consider two cases:
(j) Prop < SerD (say SerD ~ 2*Prop)
(jj) Prop >> SerD (say Prop ~ 10*SerD).
In those two cases, how do you think the material in A changes if
there is a small but positive probability of loss or damage?
Develop some intuition and check it with the simulation.
(I do not expect an exact analysis!).
Make a few plots of W vs goodput, for fixed (sensible) values of
SerD, Prop, SerA, time_out, p_drop , p_damage . Discuss what you
see.
---
This is not the only possible project using the simulation!
Any project of somewhat comparable size, that uses the simulation
to check or create some insight in the behavior of protocols
will do.