Abstract:

Due to the specificity of the application environment and the heterogeneity of the nodes in the system, most DHT networks have the problem of load imbalance. This paper focuses on the KAD network of eMule which is a popular DHT system on the internet. Through measurement study, we find that a large fraction of references to popular key words are lost due to peer saturation. It greatly impacts normal resources publishing and retrieving. This paper proposes an adaptive load balancing mechanism,which is using multiple target ID to make more peers responsible to store popular references.It solves problems related to saturated peer and makes an efficient usage of peer resources. Simulation result shows that our modification is effective.

Key words: peer-to-peer network, DHT, KAD, load balancing, resource sharing