Predictive Congestion Control Scheme for ATM Network using Multi-Source Virtual Dynamic Routing

Vol.2 No.3

Year: 2008

Issue: January-March

Title: Predictive Congestion Control Scheme for ATM Network using Multi-Source Virtual Dynamic Routing   

Author Name: Subramani A, Krishnan A   

Synopsis:    

The ATM Forum has chosen a rate-based scheme as an approach to congestion control for Available Bit Rate (ABR) services. So far, many rate-based schemes, such as Forward Explicit Congestion Notification (FECN), Backward Explicit Congestion Notification (BECN), and Proportional Rate Control Algorithm (PRCA), have been proposed. In FECN and BECN, which use negative feedback rate control, the over-all network congestion collapse may occur, if all notification cells in backward direction experience extreme congestion. The MSVDRS use positive feedback to solve the drawbacks of the FECN and BECN, but unfair distribution of available bandwidth among Virtual Connections (VCs) may occur. To resolve the problems of the existing  rate-based schemes, in this paper, we propose a new adaptive rate based congestion control scheme called the Multi-Source Virtual Dynamic Routing. The Scheme Mathematical analysis is given to demonstrate the stability of the closed-loop system so that a desired quality of service (QoS) can be guaranteed. The QoS is defined in terms of Cell Loss Ratio (CLR), transmission delay and buffer utilization. We derive design rules mathematically for selecting the parameters of the Multi Source Virtual Dynamic Routing (MSVDR) algorithm such that the desired performance is guaranteed during congestion and potential tradeoffs are shown. The performance of the proposed congestion control scheme is also evaluated in the presence of feed-back delays for robustness considerations. Finally, comparison studies are also included to show the effectiveness of the proposed method over conventional rate-based and thresh holding techniques during simulated congestion. We show that the MSVDR scheme can achieve good fairness performance among connection classes existing in the network.