Backlog balancing flow control in high-speed data networks

The asynchronous transfer mode (ATM) has been accepted by the CCITT as the multiplexing and switching technique for the future B-ISDN. Flow control algorithms in high-speed networks are resources sharing policies implemented in a distributed manner. The open-loop control is too conservative, and may result in poor utilization of resources. On the other hand, the credit-based close-loop control would be too expensive to implement in practice. A simple rate-based flow control algorithm is proposed. The aim of flow control is to maximize the network utilization for achieving high throughput and tolerable delay for each virtual circuit (VC). In a resource sharing environment, this objective may also create network congestion when a cluster of aggressive VCs is contending for the same resource at a particular node. The basic idea of our flow control algorithm is to adjust the service rate of each node along the VC according to backlog discrepancies between neighboring nodes. The hand shaking procedure between any two consecutive nodes is carried out by a link-by-link feedback protocol. Each node will update its service rate periodically with a linear projection model on the flow dynamics and binary feedback information from downstream. Since the rate updating algorithm is designed to reduce the backlog discrepancy, it is expected that a nearly balanced backlog distribution throughout the VC can be achieved over the long run time. Furthermore, the problem of network congestion can be alleviated or totally avoided if all VCs in the network agree to comply with the some set of rules.