An efficient method for end-to-end available bandwidth measurement

Measuring available bandwidth on the Internet becomes an important issue for improving network path performance for streaming application such as multimedia flows, dynamic server selection and congestion control transports. In this paper, we study an efficient active algorithm. for end-to-end available bandwidth estimation called MAPP (Measuring Available Bandwidth by Random Packets). By sending a lot of probe packets with different size from source to destination, we construct a linear square optimization with nonnegative constrains model and employ new techniques of error control and cancellation for estimating end-to-end available bandwidth. Performance data observed through extensive simulations show that our scheme is relatively accurate, efficient and non-intrusive to the networks.