An Adaptive Bandwidth-Aware Overlay for the Maximum Throughput in the Unstructured Peer-to-Peer

Heterogeneity among users is the main concern in building the unstructured P2P topology. However, the related algorithms either only pay attention to the weight heterogeneity in hitting the queries to improve the search success rate, or think much of the capability heterogeneity in data transmission to achieve load-balancing or high throughput. Because the search and transmission are two inseparable stages in an application, they perhaps do not work well in the real applications. The peers with high weight are easily overloaded, and the outstanding peers in data transmission may suffer from the deficiency in requests. Therefore, the optimization only in success rate does not simultaneously mean the optimization in throughput, and vice versa. In this paper an optimization solution of overlay network, ABAO (Adaptive Bandwidth-Aware Overlay), is proposed. In ABAO, the factors of weight heterogeneity in hitting requests and the capability heterogeneity in data transmission are both considered. The connections are adaptively adjusted with low cost according to the available bandwidth and the available requests. Accordingly the throughput can be improved to serve more customers. Our simulations show that our method can improve the system throughput as much as 22%.