Universiti Teknologi Malaysia Institutional Repository

Routing optimization using antella for peer-to-peer in an enterprise network

Puniran, Roime (2010) Routing optimization using antella for peer-to-peer in an enterprise network. Masters thesis, Universiti Teknologi Malaysia, Faculty of Computer Science and Information System.

[img]
Preview
PDF
158kB

Abstract

Enterprise Peer-To-Peer (P2P) is an Internet overlay network where resources and infrastructure are structurally developed. P2P is basically categorized into three different models; Centralized, Unstructured-Decentralized and Structured- Decentralized. In the Enterprise Unstructured P2P model, two types of nodes are used. User's node known as Regular Node (RN) and Enterprise Node (EN) belongs to a network provider. Flooding technique that is usually used in this model during routing works has been identified as the main problem. It causes high traffic and drops the system performance. Thus, it limits the resources finding and decreases the chance a peer being visited. Also, there is an interaction difficulty between RN and EN due to the different characteristics in both nodes. Deriving from the problem highlighted, the main objective of this research is to resolve the routing technique. A new technique has been proposed by combining Ant Algorithm and Travelling Salesman Problem (TSP), named Antella. This new technique promotes two strategies. Probing strategy is used to collect the trail information between peers towards their destinations. Controlled flooding strategy is used to update the path cost when an ant has completed the tour. Meanwhile, TSP has been re-evaluated to encourage peers to find the nearest EN during initial connection by measuring all ENs with their neighbors. In the experiment, two metrics are selected, Hit and Resource Usage. Hit refers to a peer being visited by a query message during crawling, while Resource Usage refers to a number of the same resources found in one particular peer within a timestep. Gnutella was chosen as the benchmark during evaluation. The results showed that Antella algorithm proposed in this research produces consistent hit value in every parameter used. The probability value for the resources found exceed six to seven times greater than Gnutella. Significant achievement shows that this new integration algorithm can be a promising technique to solve routing issues for P2P in an enterprise environment.

Item Type:Thesis (Masters)
Additional Information:Supervisor : Assoc. Prof. Dr. Muhammad Shafie Abdul Latiff; Thesis (Sarjana Sains (Sains Komputer)) - Universiti Teknologi Malaysia, 2010
Uncontrolled Keywords:peer-to-peer architecture, computer networks
Subjects:Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions:Computer Science and Information System (Formerly known)
ID Code:19225
Deposited By: Ms Zalinda Shuratman
Deposited On:02 Feb 2012 03:57
Last Modified:13 Sep 2017 04:38

Repository Staff Only: item control page