Please use this identifier to cite or link to this item: http://localhost:8081/xmlui/handle/123456789/9711
Authors: Rajput, Upendra
Issue Date: 2001
Abstract: In this thesis we present load-balancing methods to improve network utilization when static routing algorithms are employed. We have implemented two existing load-balancing algorithms and also proposed and implemented a new load-balancing algorithm. Routing schemes considered in this thesis are suitable for networks where the source node maintains call control. ATM Networks are good examples of such networks. The focus is on static routing algorithms, i.e., each source destination pair (or traffic relation) is assigned a fixed set of routes, and the network performs the Call Admission Control (CAC) operations for each traffic relation within the pre-assigned paths. The algorithms that are implemented in this thesis have the same aim, but they use different heuristics to achieve the aim. The first algorithm tries to allocate the shortest _ paths first, to reduce the waste of bandwidth due to routing over long paths. The second algorithm allocates paths for every source-destination pair, trying to assign the primary path first, then alternate ones. The third one, which we have proposed, uses the good points of both the above algorithms. It initially assigns primary path to each source destination. But if primary path cannot be assigned it assigns shortest path as alternate path. All the three algorithms are simulated and their performance is compared. Performance metrics include the network call blocking probability and load distribution between the links. It is shown that the proposed algorithm performs better than the existing algorithms. The comparison is done by simulating the algorithms in both simplex and complex topologies. Results are compared with those obtained by routing without load balancing on both a symmetric and an asymmetric traffic scenarios. The simulation is carried out using ATM Network Call Level Simulator (ANCLES) on Redhat Linux 6.2. The algorithms are implemented in C and then added to the ANCLES.
Other Identifiers: M.Tech
Research Supervisor/ Guide: Mishra, Manoj
metadata.dc.type: M.Tech Dessertation
Appears in Collections:MASTERS' DISSERTATIONS (E & C)

Files in This Item:
File Description SizeFormat 
ECDG10439.pdf2.09 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.