EXTENDED SKIP GRAPHS FOR EFFICIENT KEY SEARCH IN PEER-TO-PEER ENVIRONMENT
In this paper, we propose three techniques to improve the cost/performance of the skip graph that was recently proposed by Aspnes and Shah. The skip graph, which is a distributed data structure that could efficiently support find, insert, and delete operations of a key drawn from a totally ordered set, consists of N nodes each of which is connected with exactly log 2 N nodes determined by a set of random binary vectors called membership vectors. In the following, we will extend the construction of the skip graph in the following two directions: 1) proposal of a subgraph of the skip graph which realizes a graceful degradation of the routing performance when the number of neighbors reduces from log 2 N, and 2) proposal of a supergraph of the skip graph which realizes a significant performance improvement when the number of neighbors increases from log 2 N. The performance of those extended graphs will be evaluated analytically.