CapSearch: Capacity-Based Search in Highly Dynamic Peer-to-Peer Networks

In times of greatly heterogeneous devices, e.g. smart phones vs. desktop PCs, and their users participating in various types of networks new scenarios appear. These scenarios include peer-to-peer networks participants providing their capacities in order to operate this system. In case of weaker participants, which temporarily do not have enough capacities to contribute their share to the network, get overloaded. In a different case, participants may delegate a certain job to another participant, which is capable for this job. In this paper we propose a fast and reliable indexing and search mechanism for participants' capacities. We focus on a churn-resilient structure using a kd-tree with logical nodes in order to be capable of highly dynamic node capacities. Evaluation shows our approach is able to find nearly 60% of all nodes and over 80% of a certain number k of all nodes in big range queries, while keeping the false positive rate close to zero.