On the performance of optimization based scheduling algorithms in wireless Ad-Hoc networks

Abstract

Optimization techniques have been applied to networking problems since the seminal work of Kelly. They have been used not only in wired networks but also wireless scheduling problems. However, those algorithms are based on many assumptions such as centralized scheduling, global information and so on. Due to those reasons, they cannot be applied to real networks without modifications. In this thesis, we would like to understand performance implications of those assumptions. Specially, we considered the factors such as delay in the queue information and the difference between centralized and distributed scheduler. We performed simulation study to investigate their impacts using four different versions of the optimization based scheme. Our conclusions from the simulation study are as follows: First, the scheduling algorithm plays a very important role in determining the performance. Especially, the performance gap between the centralized scheduler and distributed one is wide. Second, queue information and control overhead have a little influence on the performance.