A Dynamic Hybrid Congestion Control Algorithm for Structured P2P Networks

In structured P2P networks, the heterogeneity of node capacity and the skewness in query pattern could result in load imbalance and even congestion. VS-based migration and indegree adjustment are two typical solutions to address this problem. The former one is effective but incurs a high maintenance overhead, while the latter one is light-weighted but subject to its interior restriction. To combine the merits of these two existing approaches, this paper presents a hybrid congestion control algorithm. The basic idea behind our algorithm is that we first dynamically identify a group of nodes susceptible to congestion and then reallocate loads among them by indegree adjustment, and finally resort to VS-like proxy routing mechanism if the load exceeds the capacity of these nodes. The results of our simulation experiments show that our algorithm could effectively reduce the level of congestion with a low overhead.