OPTIMAL RESPONSE ARCHITECTURE FOR NETWORK OVERLAY PROTOCOL

Abstract

Genetic networks are often designed to work with a single routing path, such as the shortest path, which is known as sub-optimal performance. On the other hand, the optimal performance policies suggested above (i.e. back pressure) require that each device in the network take dynamic routing decisions. In this paper, we examine the overlay structure of dynamic routing, so you only need a subset of devices (overlay nodes) to make dynamic routing decisions. We define the basic set of nodes that must form traffic to maximize network performance in many products. We apply the optimal node approach algorithm to many graphs and the results show that a small fraction of the overlapping nodes is sufficient for maximum performance. Finally, we propose a policy based on guiding thresholds and policy, which dynamically controls the traffic jams in overlapping nodes. The BP-T policy has been clarified to increase the performance of the situation when the underlying tracks do not overlap. In all simulated simulation scenarios, OBP not only achieves total productivity but also reduces delay compared to optimal performance guidance.