Structured peer-to-peer networks : hierarchical architecture and performance evaluation
1University of Oulu, Faculty of Technology, Department of Electrical and Information Engineering
2University of Oulu, Infotech Oulu
|Online Access:||PDF Full Text (PDF, 0.8 MB)|
|Persistent link:|| http://urn.fi/urn:isbn:9789514262487
|Publish Date:|| 2010-08-16
|Thesis type:||Doctoral Dissertation
|Defence Note:||Academic dissertation to be presented with the assent of the Faculty of Technology of the University of Oulu for public defence in OP-sali (Auditorium L10), Linnanmaa, on 26 August 2010, at 12 noon
Professor Noel Crespi
Associate Professor Christer Åhlund
Peer-to-Peer (P2P) networking changes the way of people utilizing Internet, for example, sharing and consuming digital content, from the ground up. It continues to show its power and strength when it is combined with other emerging technologies, such as Web Services. This thesis contributes to the research and development of P2P networks from four aspects.
Firstly, a P2P and Web Services converged multiple-tier system architecture is proposed. The architecture proposed enables providing Web Services in the context of heterogeneous access networks in an efficient way by utilizing P2P paradigm. A lightweight middleware architecture is introduced to fit the diversified mobile terminals. A theoretical analysis is given to provide a comparative study with the conventional centralized architecture.
Secondly, a General Truncated Pyramid Peer-to-Peer (GTPP) architecture is presented to analyze the performance of hierarchical architecture compared with flat architecture. The motivation behind the GTPP architecture is to see whether an added tier can bring with it added value and functionality. A detailed mathematical analysis is provided which takes into consideration various performance metrics, including the lookup hopcount, lookup latency, maintenance traffic from a single peer point of view, and maintenance traffic from the whole system point of view. Furthermore, simulation results with respect to the lookup hopcount are also provided. Through mathematical analysis and simulation results, an optimal value regarding the number of tiers of the GTPP architecture is found, showing that 2~3 tiers are appropriate for most of situations. A specialized model is also proposed to improve the performance of hierarchical architecture.
Thirdly, the performance evaluation of a communication-oriented Kademlia-based P2P system is provided in detail. NetHawk EAST-based simulation models and a prototype are both utilized to evaluate the performance. Simulation results from NetHawk EAST-based simulation models demonstrate the optimal design choices regarding the resource lookup parallelism degree and resource replication degree, and show the unnecessary existence of the messages used to detect the liveness of peers in a DHT overlay. Measurements from the prototype show the feasibility of mobile nodes acting as fully fledged overlay nodes from three different perspectives, namely CPU processing load, network traffic load, and battery consumption. The optimal size of packets which consumes battery in the most efficient way is also found through battery consumption measurements.
Fourthly, the effects of different churn models on the performance of structured P2P networks are analyzed. Specifically, three typical churn models are analyzed to provide a comparative result. The simulation results show that the difference among the effects of different churn models on the performance of structured P2P networks is quantitative rather than qualitative. This provides some guidance for the selection of different churn models for the contemporary researchers.
Acta Universitatis Ouluensis. C, Technica
© University of Oulu, 2010. This publication is copyrighted. You may download, display and print it for your own personal use. Commercial use is prohibited.