[Solved] NetworkProgramming Homework 2-UDP Reliable File Transfer

In this homework, you need to implement a UDP reliable file transfer protocol on the application level using 3 different timeout mechanisms.

The UDP reliable file transfer protocol includes a sender and a receiver. The sender keeps sending the content of a file to the receiver by using UDP packets. The receiver keeps receiving the content from the sender and outputs the received file content to another file. Important: the maximum size of every UDP datagram (packet) transfered is limited to 1024 Bytes.

To cope with the packet loss and re-ordering in the network, the sender and receiver should detect the packet loss events and re-ordering events (by using the timeout methods below) and deal with them (by re-transmitting lost packets).

Another important thing you should know is that the available link bandwidth between the sender and the receiver will be limited when we test your programs. Make sure that your retransmission mechanism works correctly under any link bandwidth.

Also, you have to take RTT and queuing delay into consideration. We will set RTT to a non-zero value and add variable queuing delays to packets when testing your programs.

Specification

You should implement senders and receivers using 3 different timeout methods, respectively:

• Timeout using SIGALRM
• Timeout using select
• Timeout using setsockopt

You can write as many source files as you need in this homework, and you can freely name your source files. However, after you compile your codes with Makefile, you have to make sure that there will be 6 executables generated, and named as follows:

• sender_sigalrm
• receiver_sigalrm
• sender_select
• receiver_select
• sender_sockopt
• receiver_sockopt

sender UDP clientreceiver UDP server (you can reuse)

Note: The receiver must NOT read a file locally to pretend that it has received the file from the sender.

You can output whatever information you need on the screen (stdout). We will only examine if the file sent and the file received are the same.

You can only use C or C++ in this homework.

SIGALRM

sender_sigalrm will send a file to receiver_sigalrm, and receiver_sigalrm will save the file. Due to the unreliable underlying network conditions, your sender_sigalrm and receiver_sigalrm must use SIGALRM to implement your retransmission mechanism.

The command to start the receiver:

receiver_sigalrm [save filename] [bind port]

The command to start the sender:

sender_sigalrm [send filename] [target address] [connect port]

For example, if the sender is about to send a file named a.txt to the receiver, whose IP address and port are 127.0.0.1 and 5000, and the receiver will save the file as b.txt:

• Start the receiver first:

receiver_sigalrm b.txt 5000

• Then start the sender:

sender_sigalrm a.txt 127.0.0.1 5000

• After finishing the file transfer, both sender_sigalrm and receiver_sigalrm should terminate automatically.
• We will test if a.txt is the same as b.txt.

Note: For the SIGALRM timeout method, you need to use siginterrupt to allow the SIGALRM signal to interrupt a system call.

select

sender_select will send a file to receiver_select, and receiver_select will save the file. Due to the unreliable underlying network conditions, your sender_select and receiver_select must use select to implement your retransmission mechanism.

The command to start the receiver:

receiver_select [save filename] [bind port]

The command to start the sender:

sender_select [send filename] [target address] [connect port]

For example, if the sender is about to send a file named a.txt to the receiver, whose IP address and port are 127.0.0.1 and 5000, and the receiver will save the file as b.txt:

• Start the receiver first:

receiver_select b.txt 5000

• Then start the sender:

sender_select a.txt 127.0.0.1 5000

• After finishing the transmission, both sender_select and receiver_select should terminate automatically.
• We will test if a.txt is the same as b.txt.

setsockopt

sender_sockopt will send a file to receiver_sockopt, and receiver_sockopt will save the file. Due to the unreliable underlying network conditions, your sender_sockopt and receiver_sockopt must use setsockopt to implement your retransmission mechanism.

The command to start the receiver:

receiver_sockopt [save filename] [bind port]

The command to start the sender:

sender_sockopt [send filename] [target address] [connect port]

For example, if the sender is about to send a file named a.txt to the receiver, whose IP address and port are 127.0.0.1 and 5000, and the receiver will save the file as b.txt:

• Start the receiver first:

receiver_sockopt b.txt 5000

• Then start the sender:

sender_sockopt a.txt 127.0.0.1 5000

• After finishing the transmission, you have to both sender_sockopt and receiver_sockopt should terminate automatically.
• We will test if a.txt is the same as b.txt.

Performance

You are free to decide whether or not to enroll in this competition to earn at most 20 bonus.

If you decide to enroll in the competition, you have to sign up the registration form, and you can only choose one retransmission mechanism that you implement (SIGALRM, select, or setsockopt) to participate in the competition.

There are two performance metrics to measure your programs performance under unreliable underlying network conditions (packet loss, limited link bandwidth, non-zero RTT, and variable queuing delay):

1. File Transfer Time (sec) (R2“>R2R2) The smaller, the better.

Suppose there are r1“>r1r1 and M“>MM students, your final ranking would be f“>ff, you will receive 30“>3030 students enrolling in the competition. Further suppose that in R1 you are ranked 3“>33. Therefore, your total ranking would be 2,3,3,7,10,13,14,17,18,18,...,N“>2,3,3,7,10,13,14,17,18,18,,N2,3,3,7,10,13,14,17,18,18,,N. In the sequence, 20×(1−930)=14“>20(1930)=1420(1930)=14 bonus.

Note that if the program you choose cannot be compiled or transfer the file correctly, you will be eliminated from the competition and get 0 bonus even though you have registered for the competition. (However, you will not get any penalty if that happens.)

Network Environment Settings

We will test if your program can work correctly under specific bi-directional packet loss rates, a non-zero RTT, and a fixed link bandwidth. We will randomly drop your packets and check whether the results are still correct.

To test your programs, you need to simulate the network scenarios (packet loss rate and packet re-ordering), and check if the file received is no different from the input.

Use the tool tc to simulate packet losses. tc is a linux built-in command line program to control the network traffic in the kernel. With tc, you can set network state parameters on a network interface card.

You should NOT use tc on the workstation (i.e. inplinux, bsd, and linux). Instead, you should do that on your own PC.

Because it requires administrator permission (i.e., sudo) to use tc, you can only use tc on your local pc to test if your senders and receivers work correctly when packet losses and re-ordering occur.

Of course, you still have to test your programs on the workstations to check if they can be built successfully with your Makefile, and check if they can work correctly in the ideal case (no packet loss, no link bandwidth limitation, and RTT is 0).

Refer to man tc for more information.

Add a rule to set the packet loss rate, queuing delay, link bandwidth, and re-ordering

In your linux shell, use the following command to set the packet loss rate, queuing delay, link bandwidth, and re-ordering:

sudo tc qdisc add dev <Device> root netem loss random <Packet Loss Rate> delay <Delay Time> <Variation> distribution <Distribution> rate <Link Bandwidth> reorder <Percent>

For example, you can add a rule on egress to set the packet loss rate to 5%, set the queuing delay to between 7 ms~13 ms (i.e., 10ms 3ms) with normal distribution, set the link bandwidth to 200 kbps, and set packet re-ordering probability to 25% on device lo.

sudo tc qdisc add dev lo root netem loss 5% delay 10ms 3ms distribution normal rate 200kbit reorder 25%

Delete all rules on a device

In your linux shell, use the following command:

sudo tc qdisc del dev <Device> root

For example, you can delete all rule settings on egress on the device lo.

sudo tc qdisc del dev lo root

Note: If there is an error message like RTNETLINK answers: File exists when you add a rule, try to delete all rules first.

Display all rules on a device

In your linux shell, use the following command:

sudo tc qdisc show dev <Device>

For example, you can display all rule settings on the device lo.

sudo tc qdisc show dev lo

Generate Test Files Randomly

You can use dd to generate a test file of a specific file size with random contents. dd is a linux built-in command line program used to convert and copy a file.

Refer to man dd for further information.

Type the following command in your linux shell to generate one file named <Ouput File Name> with the file size <File Size>:

dd if=/dev/urandom of=<Output File Name> bs=<File Size> count=1

For example, the following command generates a file named a_big_file with file size = 5MB:

dd if=/dev/urandom of=a_big_file bs=5MB count=1