After my thesis-writing exercise, I returned to research.
The current goal is to dig out more about the performance (or performance bounds) of the control mechanisms. My current idea is to study the Network Calculus (as pioneered by Rene Cruz, Chang Cheng-Shang, and Jean-Yves Le Boudec) and use it to derive something useful.
http://www.icg.isy.liu.se/courses/netcal/
Good to start, especially the first two to three lecture notes (first half of the course). However, it is quite hard to follow afterwards.
Two short papers. Nice to read but much of the results are given without proof. You may feel quite unconfortable at the first reading but it is a good summary on the elementary part of netcal.
Cheng-Sheng Chang, “On Deterministic Traffic Regulation and Service Guarantees: A Systematic Approach by Filtering”, IEEE Trans on Information Theory, 44(3) 1097–1110, May 1998.
Good paper, self contained. No background knowlege required and start with assumption. Obviously, Chang has simplified the work of Cruz to make it clear. It worths to read in very detail.
Yong Liu, Chen-Khong Tham, Yuming Jiang, “A Stochastic Network Calculus”, Tech Report ECE-CCN-0301, Department of Electrical & Computer Engineering, National University of Singapore, Nov 2004.
A TR to summarize the stochastic version of network calculus. It gives the stochastic version of delay bound/backlog bound of network.
Rajeev Agrawal, R. L. Cruz, Clayton Okino, Rajendran Rajan, “Performance Bounds for Flow Control Protocols”, April 1998
A easy-to-read paper (and yet another) on net cal.