[SOLVED] COMP3331/9331 Assignment 2 Computer Networks and Applications

COMP3331/9331 Computer Networks and Applications Assignment 2 for Session 2, 2017

Version 1.0.1
Due: 11:59pm Monday, 23 October 2017

be informed of major changes/updates through class announcements.

READ ALL SECTIONS OF THIS DOCUMENT BEFORE ATTEMPTING.

Plagiarism is a SERIOUS matter. Please read it carefully.

Page 1 of 11

Learning Objectives: On completing this assignment you will gain sufficient expertise in the following skills:

• Understanding and developing routing protocols
• Performance analysis of some routing protocols over virtual circuit and virtual packet

  networks

  Specification:

  You will implement the following program:

  RoutingPerformance

The RoutingPerformance program will accept the following command line arguments:
o NETWORK_SCHEME: this argument specifies the network type that will be evaluated. You program should implement two different network types, called here as virtual circuit network and virtual packet network. Accordingly, the argument will take one of the following values: CIRCUIT and PACKET corresponding to those

network types, respectively.
o ROUTING_SCHEME: this specifies the routing scheme that will be evaluated. Your

program should implement 3 routing protocols, which are essentially variants of Dijkstras algorithm, as explained later in the specification. This argument will take on one of the following values: SHP, SDP and LLP corresponding to the 3 routing protocols: Shortest Hop Path (SHP), Shortest Delay Path (SDP) and Least Loaded Path (LLP), respectively.

o TOPOLOGY_FILE: this file contains the network topology specification.
o WORKLOAD_FILE:thisfilecontainsthevirtualconnectionrequestsinthenetwork. o PACKET_RATE:thispositiveintegervalueshowsthenumberofpacketspersecond

which will be sent in each virtual connection.

• The network topology that will be used for the evaluation will be specified in the TOPOLOGY_FILE (the second argument). We will be using both a virtual circuit network and a virtual packet network to evaluate the performance of the routing protocols. Note that a routing protocol is required in both networks to determine the path between the source and destination of a virtual circuit during the connection. The main difference between the virtual circuit network and virtual packet network is how end-to- end path is determined. The former is similar to the circuit switching networks and the path is determined only at the connection establishment phase and such a path will be used for all packet transmission through the connection. In other words, one virtual connection in the virtual circuit network follows the same path for transmitting all packets. However, a virtual connection in the virtual packet network uses the routing protocol to determine the path for each packet independently. Thus, such virtual connection needs to invoke the routing protocol N times, where N is the number of packets transmitted through the connection. In other words, one virtual connection in the virtual packet network has N virtual circuits for transmitting N packets.
• A simple example of a network topology specification for a network with 4 routers (routers will be referred to as nodes in the rest of the specification), labelled A, B, C and D is as follows:

  A B 10 19 A C 15 20

Page 2 of 11

B C 20 20 B D 30 70 C D 8 20

This example network topology has 5 links connection the 4 nodes as shown in the figure below (Figure 1 on the next page). Each line in this file defines a point-to-point link. For example, the first line specifies a link from node A to node B, with a one-way propagation delay of 10 milliseconds and a capacity that can accommodate up to 19 simultaneous virtual circuits at any given time. All links are assumed to be bi-directional, with identical propagation delay in both directions (hence, the ordering of the names of the two endpoints of the point-to-point link does not matter, i.e. the first line in the example above could have been replaced with B A 10 19). Further, each connection in either virtual circuit network or virtual packet network is also assumed to be bi-directional and consumes unit resource (i.e. a resource of 1). There will be at most one direct link between any two nodes in the graph. You may assume that the topology will form a connected graph i.e. there will be no isolated nodes. The first task for your program is to read in the topology file and construct a suitable internal representation of the network topology, using an appropriate data structure. You may want to consult standard data structures textbooks for standard representations of undirected graphs, which would be an appropriate way to model the network. You may assume that all node names are single upper-case alphabetic characters (i.e., a maximum of 26 nodes from A to Z), all propagation delays d are positive integers (0 < d < 200) and expressed in milliseconds, and all link capacities C are positive integers (0 < C < 100) and indicate the number of virtual connections that can be supported by a link. A more complex network topology, which you may use for testing your code is available on the assignment webpage (topology.txt).

Figure 1: Topology for the above example

The network is initially empty, i.e., there are no virtual circuits established. The virtual circuit requests that arrive in the network will be specified in the virtual connection requests in WORKLOAD_FILE (third argument), which is ordered by time. The next step for your program is to read in the arriving virtual connection request workload (from the file), one request at a time, in timestamp order, and attempt to establish the virtual circuits in the network according to the routing algorithm in use (routing algorithms are explained later). The virtual connection workload for the network is specified in a simple four- column format as follows:

Page 3 of 11

0.123456 A D 12.527453 7.249811 B C 48.129653 8.975344 B D 6.124743 10.915432 A C 106.724339 15.817634 B C 37.634569

Each line of this file describes one virtual connection request. The first column specifies the time (in seconds) at which the connection is established. You may assume that time starts at 0 seconds and that time values will be represented up to 6 decimal digits (i.e. microseconds). The second column specifies the originator (source node) for the request, and the third column specifies the recipient (destination node) for the request. The final column specifies the time duration for which this virtual connection remains active for sending the packets. As an example, the first line in the above file contains a connection request that originated at time 0.123456 seconds for a virtual circuit to be established from node A to node D. This request will be active for duration of 12.527453 seconds. A more comprehensive virtual connection workload, which you may use for testing your program, is available on the assignment webpage in the workload.txt file. (Note that, this workload is consistent with the topology specified in topology.txt)

• We assume that all virtual connections have the same value for packet rate (fourth argument), which determines the number of packets per second in a connection. For example, if the packet rate value is 2 packets per second, the first virtual connection in the above example will have 25 packets (rounded value of 2 * 12.527453), the first packet starts at exactly the connection establishment time 0.123456 and finishes at 0.123456+0.5, the second one starts at time 0.123456+0.5 and finishes at 0.123456+1, , the last packet starts at time 0.123456+12.5 and finishes at 0.123456+12.527453. Clearly, for the case of virtual circuit network, you will create only one virtual circuit using the routing protocol for sending all packets. In other words, you dont need to consider the packet rate as each virtual connection will be assigned to only one virtual circuit with the same specification presented for the connection (including above four features). However, in the case of the virtual packet network, you need to extract the starting time and duration of each packet according to the packet rate value as you must invoke the routing protocol to find an appropriate path for each packet. For example, ..
• For each virtual circuit request in the above workload, your program must use the specified routing algorithm to determine if the circuit can be established. To be more specific, your program must select the best route depending on the routing protocol in use (routing protocols are explained in the next bullet point) from the source to the destination of the circuit and then determine if there is sufficient capacity along each link of this end-to-end path to accommodate the circuit. Recall that each virtual circuit uses unit capacity on each link. You may assume that the routing decision for each request can be made in zero processing time. A virtual circuit that is successfully established must be counted as such, and the network resources associated with that circuit marked as busy for the duration of the circuit. For this purpose, assume that each circuit consumes exactly one unit (i.e., one circuit) of link capacity on each link. Of course, recall that virtual circuits are bidirectional. When a circuit terminates, the resources (i.e. unit capacity) along the individual links of the end-to-end path are released, and become available for subsequent circuits that are established in the network. A circuit request that is not routed successfully is said to be blocked. A blocked request means that there is no physical path currently available from the specified source to the specified destination (according to the routing protocol in use). Such requests (packets) must be immediately discarded from further

Page 4 of 11

consideration by your program. Your program must count and report the number of blocked requests (packets). A list of performance metrics that your program must measure and report is specified later in the specification.

• Your program should implement the following three routing protocols to determine the least cost path between the source and destination. The first argument in the argument list (as explained earlier) will determine which protocol should be used for a particular run of your program. The algorithms are essentially variants of Dijkstras algorithm (i.e. link- state routing) with each scheme using a different cost metric.
  1. (1) Shortest Hop Path (SHP): This algorithm tries to find the shortest path currently available from the source to the destination, where the length of a path refers to the number of hops (i.e. links) traversed. In essence this is Dijkstras algorithm with the cost of each link set to 1. Note that, this algorithm ignores the delay and load associated with each link.
  2. (2) Shortest Delay Path (SDP): This algorithm tries to find the shortest path currently available from the source to the destination, where the length of the path refers to the cumulative propagation delay for traversing the chosen links in the path. In other words, this is Dijktras algorithm with the cost of each link set to the propagation delay. Recall that the network topology file specifies the delay along each link in the network. Note that, this algorithm ignores the number of hops and the load associated with each link.
  3. (3) Least Loaded Path (LLP): This algorithm tries to find the least loaded path currently available from the source to the destination, where the load of a path is defined to be the maximum load on any link in the path. The load on a link is defined as the ratio of its current number of active virtual circuits to the capacity, C, of that link for carrying virtual circuits. Note that, this algorithm ignores the number of hops and the delay associated with each link. There are two main differences between LLP and the other two algorithms (SHP and SDP). Firstly, the path cost in LLP is not an additive function, as is the case with the other two algorithms (in SHP and SDP the cost of the path is simply the sum of the cost along each individual link that constitutes the path). Secondly, link costs (i.e. the link load) change with time, whereas in both SHP and SDP the link costs are static over the entire lifetime.
• For all the above algorithms, whenever ties occur, they should be broken arbitrarily. In other words, if a particular routing algorithm determines that there are two or more paths between the source and destination with exactly similar costs then the algorithm should choose one of these paths randomly.
• Your program MUST NOT look for alternate paths between the source and destination of a virtual circuit request in case the path selected by the routing protocol results in a blocked request. Simply count this as a blocked request (packet) and move on to the next one.
• Your program will need to keep track of time. You may assume that time starts at 0. Recall that the workload file specifies the time at which each virtual connection request originates in the network and the duration of each virtual circuit (provided it is established). Clearly, in the case of virtual packet network, you must extract the starting time and duration for each packet and then make a virtual circuit request per each packet. Note that when the connection request is divided into the packets in the virtual packet network, your program

Page 5 of 11

must be careful about the order of the starting time of the packets of different connection requests. As discussed earlier the routing decision for each circuit request can be processed in zero time. Remember to free up the resources dedicated for a circuit once the duration of the circuit elapses.

Your program should maintain the following statistics:
o The total number of virtual connection requests.
o The total number of packets.
o The number (and percentage) of successfully routed packets. o The number (and percentage) of blocked packets.

o The average number of hops (i.e. links) consumed per successfully routed circuit. o The average source-to-destination cumulative propagation delay per successfully

routed circuit.
Note that in the case of virtual circuit network, your program make one circuit for all packets of a requested connection, thus the program should cascade the results of the virtual circuit over all packets in the corresponding connection. For example, in this case, if a connection has been blocked, it means that all packets of the connection are blocked.

• The following illustrates an example initiation of your program:
  java RoutingPerformance CIRCUIT SHP topology.txt workload.txt (for JAVA) RoutingPerformance CIRCUIT SHP topology.txt workload.txt (for C)
• Once your program has read through the entire workload file and finished processing all virtual circuit requests, it should output all of the above statistics to the terminal and then exit. A sample output from one run of the program is as follows:

Page 6 of 11

You output format MUST be exactly same as the above sample output. You need only to change the above 7 performance metrics values according to your program results. Also, the maximum number of decimal places for real performance values must be set to 2. Moreover, you MUST write the above output format in the standard output.

We will wait for reasonable amount of time (3 4 minutes) for your output to appear (with the provided workload.txt file). If you observe that your program is taking longer than 4 minutes to execute the workload file then please indicate this in your report so that we know that your program is not in an infinite loop.

Additional Notes

• You may choose to work either individually or in a group of two. If you form a group, please send an email with the names/student-id of your group members to the class account ([email protected]) by Sunday, 24th September 2017, 11:59 pm. You must use the email title Programming Assignment 2: Group. Confirmations for groups will be sent out by 12:00 PM (noon) of 25th of September 2017. WE CANNOT ACCEPT EMAILS FOR GROUPS AFTER THIS DATE.
• We neither are able to help you choose a group member nor would be accepting more than 2 people in a group. We may make facilities available on OpeanLearning for this. Please check assignment 2 page and class announcement for this.

Page 7 of 11

Marking criteria will remain the same for groups. Unless advised otherwise, contribution from each member will be considered equal. If there is any problem with group members not cooperating, this must be reported to the LIC immediately as dealing with group problems post submission will be hard to manage. Please keep a log of your

contributions/meetings.

• While you are encouraged to adopt good programming practices, your coding style is NOT subject to marking.
• The programs will be tested on CSE Linux machines. So please make sure that your program runs correctly on these machines. This is especially important if you plan to develop and test the programs on your personal computers (which may possibly use a different OS or version). We are unable to accommodate the request to have you demonstrate it on your laptop/desktop machines. Please plan ahead for any porting issues.
• Tips on how to get started:

o Focus first on making sure that your program can read in and model the network topology. The 4-node example discussed earlier is a good one to start with.

o NextfocusongettingONEroutingalgorithmworking.TheShortestHopPath(SHP) is the easiest one, since it entails implementing Dijkstras algorithm using a link cost of 1 for each hop traversed. Spend a lot of time testing your Dijkstras algorithm to make sure that it is working properly. This will definitely pay off in the long run.

o The Shortest Delay Path (SDP) is a simple variant of SHP and will involve a simple modification (i.e. changing the link cost) to your implementation of SHP. The Least Loaded Path (LLP) will require reasonable modifications but is still relatively easy to implement.

o Test your routing algorithm(s) with a SMALL network topology and a SMALL virtual circuit workload file. Links with very low capacity (e.g. 1) are useful in initial testing. Make sure that the routing is working, and that your program is able to compute the required statistics.

o Extend your program for the virtual packet network case and repeat all tests for all three routing algorithms.

o Once you are fairly confident that your program is working correctly for a small topology, and then move on to the provided sample topology. It may be useful to initially test out the workload file incrementally (e.g.: first 10 requests, first 50 requests, first 100 requests and so on).

Page 8 of 11

Language and Platform: You are free to use either C or C++ or JAVA or Python to implement this assignment (you can choose only one of these languages for the entire assignment, not a combination of them). Your assignment will be tested on the Linux Platform. Make sure you develop your code under Linux (check the version on CSE machines).

Plagiarism:Youaretowriteallofthecodeforthisassignmentyourself.Allsourcecodes are subject to strict checks for plagiarism, via highly sophisticated plagiarism detection software. These checks may include comparison with available code from Internet sites and assignments from previous semesters. In addition, each submission will be checked against all other submissions of the current semester. Please note that we take this matter quite seriously. The LIC will decide on appropriate penalty for detected cases of plagiarism. The most likely penalty would be to reduce the assignment mark to ZERO and reported to school plagiarism register. We are aware that a lot of learning takes place in student conversations, and dont wish to discourage those. However, it is important, for both those helping others and those being helped, not to provide/accept any programming language code in writing, as this is apt to be used exactly as is, and lead to plagiarism penalties for both the supplier and the copier of the codes. Write something on a piece of paper, by all means, but tear it up/take it away when the discussion is over.

Submission: includes 3 main components:

1. Your program file
2. A 3-page report

3. A maximum of 5 min screencast of your program as specified in page 9 of this

document.

1: You really need only one file: RoutingPerformance.c (or RoutingPerformance.java). If you are using C and if you are going to use any other files besides these two such as header files or other .c files then you will have to submit a Makefile along with your code. This is because we need to know how to resolve the dependencies among all the files that you have provided. If you are using Java and have multiple files, a makefile will not be necessary if your program doesnt need any CLASPATH settings (javac *.java usually work here).

2: You should submit a report, report.pdf (no more than 3 pages) including the following 3 items:

o An explanation on the data structure(s) that you have used for the internal representation of the network topology.

o Atabulatedsummaryofthecomparisonoftheperformancemetricsforthe3routing protocols over the virtual circuit network. This comparison MUST be carried out using the provided topology and workload files (topology.txt and workload.txt). Your table must contain 8 columns (one each for the 8 performance metrics) and 6 rows (one each for the 3 routing protocols for two networks).

o An analysis of the results, i.e., where possible you MUST provide reasons for the performance results that you observe. In particular comment on the reason behind the differences in the following metrics for the 3 protocols: percentage of blocked requests, average number of hops and the average propagation delay.

o Performance evaluation of the virtual packet network with respect to the packet rate and different routing protocols. To this end, you need to collect the values of three

Page 9 of 11

performance metrics percentage of successfully routed packets, average number of hops per circuit, average cumulative propagation delay per circuit by varying the packet rates in the range of 1 5 and changing the routing protocols over the topology.txt and workload.txt files. Then, you will create three plots corresponding to each performance metric. In each plot, the x-axis is the different values of packet rates and the y-axis is the performance values. Also, in each plot you draw three line charts corresponding to each routing protocol. Finally, try to describe the plots and the reason of the results by one paragraph for each plot.

3: A maximum of 5 min video must be included with your submission. This video is your program DEMO and must start from Test 1 and finish by Test 4 in Marking Guide lines specified in the following section. This is mandatory and its like DEMO for assignment 1. We cannot mark your work without this.

Marking Guidelines:
The marking will consist of the following sequence of tests:

Test 1: Correct Compilation Correct compilation of all files: 2 mark

Test 2: Shortest Hop Path (SHP) Test

We will first evaluate the SHP routing protocol for both networks. We will first verify the performance for the provided topology and workload files and then test with a different topology and associated workload: 12 marks

Test 3: Shortest Delay Path (SDP) Test
We will next evaluate the SDP routing protocol for both networks. The same topology and

workload files as in the SHP test will be used for this evaluation: 14 marks Test 4: Least Loaded Path (LLP) Test

Next we will evaluate the LLP routing protocol for both networks. The same topology and workload files as in the SHP test will be used for this evaluation: 18 marks

Test 5: Report

Your report is worth 14 marks. Your report will be evaluated as follows:
Description of the data structure used to create the internal representation of the network

topology: 2 marks
Tabulated comparison of the performance metrics of the 3 routing protocols for the

provided topology and workload files: 2 marks
Explanation of the performance results observed: 2 marks
Plots for performance evaluation of the virtual packet network: 4 marks Explanation of the plots: 4 marks

Page 10 of 11

Link to Mandatory Screencast Demo (Test 5)

Test 5: Mandatory Screencast Demo

The screencast demo must go through Test 1 until Test 4 (inclusive). You may describe your program during this or just run the tests within a maximum of 5 minutes. You must show your code during the screencast and you must be logged in to your own CSE account while doing this. We will compare the submitted program with the DEMO and any cheating will result in ZERO mark.

The screencast must be uploaded on YouTube before the deadline and a link must be included for this in your report. Videos changed after submission date, will be ignored.

Final Mark: The final mark out of 50 is then scaled to 15.

The screencast is mandatory and its like DEMO for Assignment 1. Your work cannot

be marked without it

Late Submission Penalty: Late penalty will be applied as follows: o 1 day after deadline: 10% reduction
o 2 days after deadline: 20% reduction
o 3 days after deadline: 30% reduction

o 4 days after deadline: 40% reduction o 5 or more days late: NOT accepted

NOTE: The above penalty is applied to your final total. For example, if you submit your assignment 1 day late and your score on the assignment is 60, then your final mark will be 60 6 (10% penalty) = 54.

Tutors Marking Procedure:

and your mark will be severely penalized

The following shows the sequence of events that will be involved in marking your assignment:

1. 1) We will start from your Screencast DEMO.
2. 2) We will then test your program with the network topology and workload files that are

   provided on the assignment page on OpenLearning (topology.txt and workload.txt). All 3 routing protocols and two networks will be tested.

3. 3) We will repeat the tests for a different topology and associated workload file. Note that, the format of the topology and workload files will be consistent with those provided on OpenLearning.
4. 4) Finally, your report will be marked.

IMPORTANT NOTE: We will not read through your code to evaluate your coding style, etc. For assignments that fail to execute all of the above tests, we will be unable to award you a substantial mark.

with no exceptions

whatsoever.

Page 11 of 11