Project 3The theme of the project is to study experimentally how the network reli-ability depends on the individual link reliabilities, in the specific situationdescribed below.Network topology: A complete graph on n = 5 nodes. This means, everynode is connected with every other one (parallel edges and self-loops are ex-cluded in this graph). As a result, this graph has m = 10 edges, representingthe links of the network.Components that may fail: The links of the network may fail, the nodesare always up. The reliability of each link is p, the same for every link. Theparameter p will take different values in the experiments.Reliability configuration: The system is considered operational, if thenetwork topology is connected.Specific tasks:1. Create an algorithm to compute the network reliability in the abovedescribed situation, using the method of exhaustive enumeration (seein the Lecture Notes). Note that the high level description given in thenotes is not enough, since you also have to specify how you actuallywant to find the details, such as how to generate the possible states,how to assign an up/down system condition to each, how to convert itinto a reliability value, etc.IMPORTANT: Finding an algorithmic solution for these de-tails is part of the task!Describe how your algorithm works. First briefly explain informally theideas. Then provide pseudo code for the description of the algorithm,with sufficient comments to make them readable and understandableby a human.2. Write a computer program that implements the algorithm You mayuse any programming language under any operating system, this is1entirely of your choice. Make sure, however, that your program is wellstructured to support finding potential errors (debugging), checkingcorrectness or trying out algorithm changes. Explain how your programsupports these goals.3. Run the program for different values of p. Let the parameter p run overthe [0, 1] interval, in steps of 0.02. Show graphically in a diagram howthe obtained network reliability values depend on p.4. Now fix the p parameter at p = 0.9, and do the following experiment.Among the 210= 1024 possible combinations of component states pickk of the combinations randomly, and flip the corresponding systemcondition. That is, if the system was up, change it to down, if itwas down, change it to up. Show in a diagram, how the reliability ofthe system changes due to this alteration. Specifically, show how thechange depends on k, in the range k = 0, 1, 2, 3, . . . , 100. During thisexperiment keep the value of the parameter p fixed at p = 0.9. Toreduce the effect of randomness, run several experiments and averagethem out. for any given value of k.5. Provide a 1-2 paragraph explanation why the obtained diagrams lookthe way they look. In other words, try to argue that they exhibit thebehavior that one could intuitively expect, so the program is likely towork correctly.Note: If something is not specified in this project description, that means itis left to your choice.Submission guidelinesDescribe everything, including verbal description, pseudo code, program, in-put/output data, experimental results, figures, conclusions, in a single doc-ument. Submit it through eLearning. Do not submit executable code, butinclude the source code as an Appendix in the document. Note: your sub-mission will be read as a document and not run as a program (but there aretwo exceptions, see them under Evaluation).2Remark: It might be helpful to think about the whole project that yourtask is not only to solve the technical problem, but you also have to “sell”the results. Try to look at your work from the viewpoint of a potential“customer”: how convincing your presentation would look for a customer?EvaluationThe evaluation will focus on how well each of the specific tasks have beencarried out. Even though the submission will not be run, only read as adocument, there are two exceptions. You will be asked to demonstrate ona computer how your program actually runs, if any of the following casesoccur:1. You do not agree with the grade and want to improve it. In this casethe demonstration should show that your work is actually better thanthe received grade.2. There is suspicion that the work is not original or not individually doneor the results were not produced by your own correctly running pro-gram. In this case a demonstration is required to clarify the situationand to receive any score.Note:The work should be fully individual and original. Any form of cheating is aserious violation of University policies and can lead to serious
View Full Document
Unlocking...