Journal of Network and Computer Applications
A stable rate-based algorithm for active queue management
Computer Communications
FQL-RED: an adaptive scalable schema for active queue management
International Journal of Network Management
| Hi-index | 0.00 |
This paper studies the active queue management(AQM) in high-speed routers. One of the original andpopularly implemented AQM algorithms is random earlydetection (RED). But it is hard to configure theparameters involved in RED. Although there are somenew TCP-modeling based approaches to overcome this,they really show some instability under such dynamicalenvironments with diverse connections (adaptive or non-adaptive,short or long), link rate, RTT, and buffer size.The major reason is that these approaches are primarilybased on TCP model, which has some mismatches underdynamical environment. Since Fuzzy-logic control doesnot need precise control model, it may be suitable forsuch environment. This paper designs an adaptive fuzzy-basedcontrol algorithm (AFRED), which computes thepacket drop (or marking) probability according to pre-configuredfuzzy logic using the instant queue length asinput variable. The important feature of AFRED is that italso designs an adaptive mechanism to dynamically re-adjustthe fuzzy rule so as to make AFRED itselfextensively stable for many dynamic environments.Simulation results show that AFRED can effectively andstably control the queue length to the expected value orso. Compared with RED and PI algorithm, AFREDobtains higher goodput and stable queue length thanRED and PI, even if introducing UDP flows.