Recently, people have been trying to undestand the "complexity" of the distances in the Internet: if these distances are indeed "simple" (in a certain mathematically well-defined sense), we have developed very good routing protocols that store very little extra information in the system.

This project will have two conceptual parts: (1) we will try to find out the "dimension" of distances in the Internet, and (2) we will implement the improved routing protocols and empirically evaluate their performance on real(istic) networks.

Students should be proficient in C/C++, and should have basic background in algorithms. Networking background a plus.