Abstract:
The paper proposes an indexing scheme known as Floating Index Caching. In Floating Index caching, queries carry a summarized index of contents of the peers that are on its way. Each node snoops on the queries passing through it. Queries are sent in the network using ether flooding or random-walk methods. When a query arrives at a node, indexes can be further added to the query or existing indexes replaced with new indexes. In this paper, we classify propagation schemes as depth-wise and breadth-wise, combined with flooding and random-walk methods. Through experimental evaluations, we found that Floating Index Caching with random walkers has the best returns as far as index dissemination is considered. Breadth-wise spreading of indexes through random-walk utilized 54% of the query traffic effectively to spread the indexes. This method has spread indexes of 77% of the nodes onto more than 50% of the network.