Robust random early detection: Difference between revisions

Revision as of 04:50, 11 June 2013

Latest revision as of 13:38, 7 January 2024

(15 intermediate revisions by 13 users not shown)

Line 1:

'''Robust random early detection''' ('''RRED''') is a queueing discipline for a [[network scheduler]]. The existing [[random early detection]] (RED) algorithm and its variants are found vulnerable to emerging attacks, especially the [[Denial-of-service attack#Low-rate Denial-of-Service attacks|Low-rate Denial-of-Service attack]]s (LDoS). Experiments have confirmed that the existing RED-like algorithms are notably vulnerable under LDoS attacks due to the oscillating TCP queue size caused by the attacks.<ref name=RRED>{{citeCite journal doi| last1 = Zhang | first1 = C. | last2 = Yin | first2 = J. | last3 = Cai | first3 = Z. | last4 = Chen | first4 = W. | title = RRED: Robust RED algorithm to counter low-rate denial-of-service attacks | doi = 10.1109/LCOMM.2010.05.091407 | journal = [[IEEE Communications Letters]] | volume = 14 | issue = 5 | pages = 489–491 | date=May 2010 | s2cid = 1121461 | url = https://sites.google.com/site/cwzhangres/home/files/RREDRobustREDAlgorithmtoCounterLow-rateDenial-of-ServiceAttacks.pdf?attredirects=0}}</ref>

AThe Robust RED (RRED) algorithm was proposed to improve the TCP throughput against LDoS attacks. The basic idea behind the RRED is to detect and filter out attack packets before a normal RED algorithm is applied to incoming flows. RRED algorithm can significantly improve the performance of TCP under Low-rate [[denial-of-service attack]]sattacks.<ref name="RRED"/>

==The Designdesign of Robust RED (RRED)==

A detection and filter block is added in front of a regular RED block on a router. The basic idea behind the RRED is to detect and filter out LDoS attack packets from incoming explosionsflows before they feed to the RED to a very hungry monkey algorithm. How to distinguish an attacking packet from normal TCP packets is critical in the RRED design.

Within a benign TCP flow, the sender will delay sending new packets if loss is detected (e.g., a packet is dropped). Consequently, a packet is suspected to be an attacking packet if it is sent within a short-range after a packet is dropped. This is the basic idea of the detection algorithm of Robust RED (RRED).<ref name="RRED"/>

==The Algorithm of the Robust RED (RRED)==

'''algorithm''' RRED-ENQUE(pkt)

<nowiki>

01 f ← RRED-ENQUEFLOWHASH(pkt)

02 Tmax←MAX Tmax ← MAX(Flow[f].T1, T2)

01 f←RRED-FLOWHASH(pkt)

03 '''if''' pkt.arrivaltime is within [Tmax, Tmax+T*] '''then'''

02 Tmax←MAX(Flow[f].T1, T2)

04 reduce local indicator by 1 for each bin corresponding to f

03 if pkt.arrivaltime is within [Tmax, Tmax+T*] then

05 '''else'''

04 reduce local indicator by 1 for each bin corresponding to f

06 increase local indicator by 1 for each bin of f

05 else

07 Flow[f].I←maximumI ← maximum of local indicators from bins of f

06 increase local indicator by 1 for each bin of f

08 '''if''' Flow[f].I >=≥ 0 '''then '''

07 Flow[f].I←maximum of local indicators from bins of f

09 RED-ENQUE(pkt) // pass pkt to the RED block

08 if Flow[f].I >=0 then

10 '''if''' RED drops pkt '''then'''

09 RED-ENQUE(pkt) //pass pkt to the RED block

11 T2 ← pkt.arrivaltime

10 if RED drops pkt then

12 '''else'''

11 T2←pkt.arrivaltime

13 Flow[f].T1←pktT1 ← pkt.arrivaltime

12 else

14 drop(pkt)

13 Flow[f].T1←pkt.arrivaltime

15 '''return'''

14 drop(pkt)

15 return

</nowiki>

*''f.T1'' is the arrival time of the last packet from flow ''f'' that is dropped by the detection and filter block.

*''T2'' is the arrival time of the last packet from any flow that is dropped by the [[random early detection]] (RED) block.

*''Tmax = max(f.T1, T2)''.

*''T*'' is a short time period, which is empirically choosechosen to be 10ms10&nbsp;ms in a default RRED algorithm.<ref name="RRED"/>

==The Simulationsimulation code of the Robust RED (RRED)==

The simulation code of the RRED algorithm is published as an [[active queue management]] and [[denial-of-service attack]] (AQM&DoS) simulation platform. The [http://sites.google.com/site/cwzhangres/home/posts/aqmdossimulationplatform AQM&DoS Simulation Platform] is able to simulate a variety of DoS attacks (Distributed DoS, Spoofing DoS, Low-rate DoS, etc.) and active queue management (AQM) algorithms ([[random early detection|RED]], RRED, SFB, etc.). It automatically calculatecalculates and recordrecords the average throughput of normal TCP flows before and after DoS attacks to facilitate the analysis of the impact of DoS attacks on normal TCP flows and AQM algorithms.

==References==

{{reflist}}

Line 40 ⟶ 39:

*[http://sites.google.com/site/cwzhangres/home/posts/aqmdossimulationplatform AQM&DoS Simulation Platform]

*[http://sites.google.com/site/cwzhangres/home/posts/recentpublicationsinlow-ratedosattacks Recent Publications in Low-rate Denial-of-Service (LDoS) attacks ]

*[http://sites.google.com/site/cwzhangres/home/posts/recentpublicationsinrandomearlydetectionredschemes Recent Publications in Random Early Detection (RED) schemes] {{Webarchive|url=https://web.archive.org/web/20160917055738/http://sites.google.com/site/cwzhangres/home/posts/recentpublicationsinrandomearlydetectionredschemes |date=2016-09-17 }}

*[http://sites.google.com/site/cwzhangres/home/posts/recentpublicationsinactivequeuemanagementaqmschemes Recent Publications in Active Queue Management (AQM) schemes]