Recently, the peer-to-peer (P2P) distributed technology has become associ-
ated with many things, ranging from file sharing, software applications to network
technologies. A fundamental problem common to P2P applications is that of an
efficient and scalable lookup service to locate or discover the peers that store (or
share) a particular resource. Some design criteria for such an indexing and discovery
infrastructure are: fast search, secured index and scalable to work in large networks.
       This project addresses the issue of collision in DGRID (DHT-based Grid Re-
source Indexing and Discovery), a Chord-based hierarchical peer-to-peer system
with secured index that is designed for grid systems. Collision occurs when P2P
operations cause more than one node to be created in the first-level overlay network
and thus increases the lookup path length. We propose new collision avoidance algo-
rithms for the join and leave operations. For node failures, our proposed algorithm
avoids collision in the case of a leaf node. Collision avoidance due to supernode
failures remains an open issue. Simulation results confirm that collisions can be
avoided and our algorithms improve the overall lookup performance significantly.