Please use this identifier to cite or link to this item:
http://dspace.cityu.edu.hk/handle/2031/7359
Title: | Designing survivable networks |
Authors: | Yuan, Yuan |
Department: | Department of Electronic Engineering |
Issue Date: | 2014 |
Supervisor: | Supervisor: Prof. ZUKERMAN, Moshe; Assessor: Dr. WONG, Eric W M |
Abstract: | Modern wireline networks keep growing in range, size and depth. For large scale networks, even single-link failure may result in severe service disruption towards a great number of users. Therefore, designing the survivable network to provide a steady and reliable service while certain failures occur is no longer a new topic to network specialists. Manual maintenance and intervene become an increasingly demanding and difficult problem due to the widely spread structure and highly dynamic failure situations. The various causes of network failures, such as road construction, extreme weather condition, traffic overload during certain time of a day or year and etc., make them impossible to prevent only by prediction. The urge of designing the survivable network naturally came up as a mean to keep up a reliable network performance under certain failure scenarios. Among all the researches done towards survivable network designing solution, two-link failure scenario is widely investigated. As the manufacturing technology of related hardware for network systems keeps advancing, such as the appearance of the optical fiber and more robust network intermediate devices and etc., although in real-life cases, one-link failure is one of the most frequent failure scenarios from the perspective of probability, there is a relative high possibility that a second link fails during the time that the first failed link is still down and being repaired and restored. The main purpose of this project is to investigate in the ability of a network to survive certain link failures due to unexpected circumstances or traffic overload. The design and planning of the survivable network for a fully meshed N-node network under any random 2-link failure scenarios is investigated and explored. After the study of currently existing design algorithms, three detailed ones, the feasible solution, the optimal solution and the heuristic solution, are carefully examined, verified and compared. The comparison shows that: while the feasible solution will not necessarily produce a protection network with optimal capacity (used interchangeably in this paper with the word cost) assignment, it provides us with a time-efficient solution to large network. The optimal solution can achieve the minimum capacity assigned to the protection network, but its need for the huge amount of computing time makes this algorithm non-practical for real-life cases. The heuristic solution combining both algorithms above approximates the optimal capacity assignment and is relatively time efficient in the meantime, at least for the problem that only involves a moderately sized network. |
Appears in Collections: | Electrical Engineering - Undergraduate Final Year Projects |
Files in This Item:
File | Size | Format | |
---|---|---|---|
fulltext.html | 145 B | HTML | View/Open |
Items in Digital CityU Collections are protected by copyright, with all rights reserved, unless otherwise indicated.