Node Evaluation in the Chord P2P Systems

Abstract

In this paper we investigate two statistical estimators for the number of nodes in the Chord peer-to-peer system and one estimator for the number of information items stored in the system. The first estimator of the number of nodes is based on the Maximum Likelihood Estimation (MLE) technology and the second is based on the Methods of Moments (MM). We compare its statistical properties, show that the MM estimator has a better properties for our purposes and we discuss how we can implement this estimators in the Chord system.Our solutions can be implemented in Chord in such a way that they do not require any additional transmissions since all necessary information can be gathered during the standard operations of checking consistency of the Chord structure.Our results can be applied not only to the Chord P2P system but can be adopted to other decentralized and structured systems.