Dynamic adaptation burst assembling method

Abstract

A dynamic adaptive burst assembling method is to utilize an one-step auto-regression linear adaptive random process estimation model and adopt an orthonormalization minimum mean square deviation predict method to estimate network flow property on line timely an dynamically adjust burst assembling time by estimating the packet length. With the predict time, it combines this time flow weight and one step p stage estimation model and its maximum burst assembling time is dynamically determined by estimating its packet length and the latest load to deliver the control packet before finishing the burst assemble.

Description

technical field [0001] The invention relates to an optical network-based assembly method, in particular to a dynamic self-adaptive burst assembly method based on the real-time characteristics of optical network edge access services. Used in the field of optical communication technology. Background technique [0002] Optical Burst Switching (OBS) network has become one of the research hotspots in the field of optical communication. An OBS network is composed of several edge routers and core routers. The edge router is responsible for completing the interface with the traditional network and assembling traditional electronic packets into bursts. It is the basic transmission and switching unit in the OBS domain. OBS is a one-way resource reservation scheme. The edge router creates a control packet for each burst and sends it out at an offset time before the burst. The core router electronically processes it to configure the optical switch matrix. OBS can realize transparent ...

AI Technical Summary

Problems solved by technology

[0004] Existing burst assembly methods (X.Yu, Y.Chen, and C.Qiao, A Study of traffic statistics of assembled burst traffic in optical burst switched networks , Proceedings, Optical Networking and Communication Conference (OptiComm) 2002, Boston, MA, July-Aug 2002) continue to use the traditional Poisson arrival interval distribution traffic pattern under circuit switching, and still use the traditional time-based And the centralized mechanism of packet length, the offset time and the centralized threshold are only in the network traffic simulation research in the corresponding large time range of microseconds and milliseconds. The superposition of the two brings additional assembly delay, which is This traffic concentration mode with a fixed time range and packet length has a good ability to assemble voice services, but it lacks adaptability to the suddenness, self-similarity, and long-range correlation of IP service Ethernet traffic in networks where data traffic dominates. It cannot be dynamically adjusted according to the real-time characteristics of traffic to automatically adapt to network changes, and due to the limitation of its assembly granularity and the irrelevance of real-time traffic, small burst assembly introduces large overhead and reduces network throughput, while large burst increases assembly delay

These defects of the existing burst assembly method cannot enable the network aggregation layer to give full play to the advantages of electronic intelligence to dynamically and adaptively change the service characteristics of the access layer, which increases the pressure on the core optical network, resulting in deterioration of network throughput performance, and also reduces Optical Network Flexibility

Images