Queue Management with Flow Number Estimation and Queue Length Feedback Control

碩士 === 國立成功大學 === 電機工程學系碩博士班 === 91 === Many active queue management schemes such as Random Early Detection (RED) [3] [4], BLUE [2], Random Exponential Marking (REM) [5], Stabilized RED (SRED) [7], and Dynamic RED (DRED) [8] were proposed for IP routers in recent years. We evaluate the effects of pa...

Full description

Bibliographic Details
Main Authors: Yong-Shun Su, 蘇永順
Other Authors: Jung-Shian Li
Format: Others
Language:en_US
Published: 2003
Online Access:http://ndltd.ncl.edu.tw/handle/93435115753975787095
Description
Summary:碩士 === 國立成功大學 === 電機工程學系碩博士班 === 91 === Many active queue management schemes such as Random Early Detection (RED) [3] [4], BLUE [2], Random Exponential Marking (REM) [5], Stabilized RED (SRED) [7], and Dynamic RED (DRED) [8] were proposed for IP routers in recent years. We evaluate the effects of parameters of queue management on the performance using 2k factorial designs. Among these systematic and environmental parameters, number of flows and Pmax are two dominate factors affecting the performance of RED queue management. Then, we propose a novel active queue management scheme, NRED, employing active flow-number estimation and queue length feedback control, which are motivated from Bloom filter and control theory, respectively. Fluctuation of queue length under NRED is smaller than any existing queue management schemes. Especially, NRED can stabilize the queue length very fast even when a lot of flows join in the competition suddenly. NRED is very suitable to be deployed in today’s networks where hot spot frequently occurs. Furthermore, delay jitter regulated by NRED is very low.