Control theory optimization of MECN in satellite networks

Congestion in the Internet results in wasted bandwidth and also stands in the way of guaranteeing QoS. The effect of congestion is multiplied many fold in satellite networks, where the resources are very expensive. Thus congestion control has a special significance in the performance of satellite networks. In today´s Internet, congestion control is implemented mostly using some of the de facto standard, RED. We have proposed previously a new scheme called the multi-level explicit congestion notification (MECN), which works with the framework of ECN, but enhances the performance using multiple level congestion feedback from the network. But like RED, the tuning of MECN could be a problem. Tuning of parameters is necessary for achieving high throughput with corresponding low delays. Also it is desirable to keep the oscillations in the queue low to reduce jitter, which is the major concern in real-time applications such as voice or video over IP. In this paper, we use a linearized fluid flow model of TCP-MECN to study the performance and stability of the queue in the router. We use classical control tools such as delay margin and tracking error to provide guidelines for optimizing satellite IP networks, which use MECN. We apply our results for optimizing the performance of satellite networks. We use ns simulator to validate the results of our analysis.