Static and self-scalable filter range selection algorithms for peer-to-peer networks

• Main Author: Kweh, Yeah Lun
• Format: Thesis
• Language: English; English
• Published: 2011
• Online Access: http://psasir.upm.edu.my/id/eprint/27381/1/FSKTM%202011%2019R.pdf; http://psasir.upm.edu.my/id/eprint/27381/
• Institution: Universiti Putra Malaysia

In this research, problems on the selection of keys in peer-to-peer networks are investigated. The key selection is about finding the target key with kth rank in a global file with n keys that are distributed evenly among p nodes with each node holding n/p keys in the peer-to-peer network. In the literature, there are many selection algorithms proposed for different network topologies. For peer-to-peer networks, Loo (2005) selection algorithm has been selected as the benchmark as it is an established algorithm that claimed to be the best proposed for this network. The research works were implemented by simulation in which it was used to identify the selection problem, implementation of the proposed algorithms and the measurement of the results. Two multiple selection algorithm, which are known as “static filter range selection algorithm” and “self-scalable selection algorithm” are proposed. These algorithms are based on the statistical knowledge about the uniform distribution nature of the data and arranged in certain order in the file. The selection algorithms can perform multiple selections concurrently to find multiple target keys with different predefined target ranks. The static filter range selection algorithm uses a fixed filter range approach to locate the target key, in which the filter range is preset at the beginning of the searching process. The range will be adjusted and becomes narrower while ensuring the target keys are still within it as the process iterates until the keys have been found. The selfscalable selection algorithm uses dynamic range where the filter range is not preset and is determined by the algorithm itself based on the distribution and the values of the data in the global and local file. After that, the ranges are made smaller and smaller until the target keys are found. Four parameters have been applied in this research to measure the performance of the algorithm. These are number of rounds needed, number of messages needed, success rate and execution time. The static filter range selection algorithm and the self-scalable selection algorithm are able to reduce the number of rounds and the number of messages needed, increase the success rate but longer execution time compared to the Loo (2005) selection algorithm. The self-scalable selection algorithm is also able to reduce the number of rounds and the number of messages needed, increase success rate and shorten the execution time compared to static filter range selection algorithm with filter range 15000, 20000 and 25000.