|
| Budget: |
$ 20-100 |
| Status: |
Closed
for Bidding
|
| Project
Creator: |
|
| |
Rating: (No Feedback Yet)
|
| Required
Skills: |
,
|
| Attached
Files: |
(None) |
 |
|
|
|
|
|
|
|
|
|
|
Description
I just need the code for simulator, I will do the testing for the following:
There are 10 hosts connected to the LAN. Flows of packets from all hosts that go through the gateway are queued in one incoming buffer. All hosts generate packets at the same rate. Packet lengths are exponentially distributed with a mean of 1000 octets. The outgoing link capacity is 10 Mbps. Build a simulator for each of the following cases: 1. Assume that packets from each host arrive according to a Poisson process independently of other hosts. Find the minimum gateway buffer size (in octets) to ensure a packet loss probability of not more than 0.001 at the gateway for any utilisation that does not exceed 90%. 2. For the gateway buffer size found in question 1, plot the packet loss probability for packet arrival rates of ©p=60, 80, 100, 120 packets/second per host. What conclusions can you draw? Explain. 3. Now, assume that some of the hosts exhibit a Poisson batch arrival process, instead of a simple Poisson packet arrival. In a Poisson batch arrival process, the arrival events are distributed as in a Poisson process, but each arrival carries a batch of several packets at once (instead of just one packet). The packets in the batch join the queue if the queue is not full; it is possible that only some of the packets in a batch will join the queue, and the rest of the batch will be lost. Assume that the Poisson arrival rate is ©b=10 batches/second per host, and the batch size is distributed geometrically with a mean of ©p/©b, where ©p takes the same values as before (60, 80, 100, 120 packets/second). The packet length distribution is the same as before. For each of the following cases, using the buffer size found in question 1, plot the packet loss probability for the “batch arrival” hosts and the “single-packet arrival” hosts, for the different values of ©p: a. if one host has batch arrivals, and all other 9 hosts have packet arrivals; b. if two hosts have batch arrivals, and all other 8 hosts have packet arrivals; c. if all 10 hosts have batch arrivals. What conclusions can you draw? Explain. 4. This question focuses on case 3a above (one host with batch arrivals). We would now like to provide a fair access for all hosts, and apply the fair buffer allocation (FBA) scheme for this purpose (see lecture on TCP over ATM, or Stallings pp. 347-350). Use the same buffer size obtained in question 1, with the parameter R at 75% of the buffer capacity and Z=0.95. Find and plot the mean packet delay in the gateway as a function of gateway occupancy. Draw a separate graph for the “batch arrivals” host and for the other hosts. In the same manner, find and plot the packet loss probability. What conclusions can you draw about the effectiveness of FBA? Explain.
Reminder
You may not start working in this and any project before
your bid is accepted. Any user who violates this policy
may have their account permanently suspended.
|