Advanced Moblle CompuLlng

ro[ecL Synopsls
Cn
ConLrolllng 1rafflc CongesLlon
ln a neLwork


Croup Members
1Apoorva khera 08303932
2Monlsha Mallk 08303974
38eema Aswanl 08303986
4CheLna 8handarl08304000


Title: Controlling Traffic Congestion in a network.
Simulator: NS2 Simulator In Linux

TRAFFIC CONGESTION:

Congestion in a packet switched network is a state in which perIormance degrades
due to saturation oI network resources such as communication links, processor
Cycles ,memory buIIers.
In other words,network congestion occurs when a link or node is carrying so much
data that its quality oI service deteriotes.


When congestion occurs, Iollowing eIIects are seen:
1. elay due to the queue at one end oI the connection being too big, so it
takes time Ior your packet to be transmitted.
2. Packet loss when new packets are simply dropped, Iorcing connection
resets (and oIten causing more congestion).
3. Lower quality oI service, protocols like TCP will do a cutback on the
transmission rate, so your throughput will be lowered.
4. Blocking, certain networks have protocol priorities, so your UP
packets may be dropped in Iavor oI allowing TCP traIIic through.
1rafflc CongesLlon Can be conLrolled by elLher avoldlng lL or prevenLlng lL when a
congesLlon acLually occurs Lhus causlng packeL loss and low LhroughLpuL

Some of Lhe CongesLlon Avoldance AlgorlLhms are
1 lalr Cueulng
2 Schedullng AlgorlLhm
3 8andom Larly ueLecLlon AlgorlLhm8Lu

Some CongesLlon ConLrol AlgorlLhms are
1 1C and Llnux luggable CongesLlon ConLrol AlgorlLhms
2 1he Leaky 8uckeL(used prlmarlly ln A1M)
3 Choke ackeLs



1he CongesLlon ConLrol algorlLhms baslcally lnfer congesLlon from packeL loss
because
Sender can lnfer congesLlon from packeL loss Lhrough mlsslng
acknowledgmenLs
8aLe or percenLage of losL packeLs can be used Lo gauge degree of
congesLlon

AlLhough many neLwork congesLlon conLrol algorlLhms and research papers have
been proposed Lhe problem of neLwork congesLlon conLrol remalns a crlLlcal
lssue and a hlgh prlorlLy especlally glven Lhe prospecLlve of Lhe conLlnually
growlng speed and slze of fuLure neLworks

1hls moLlvaLes us Lo lmplemenL an algorlLhm whlch would conLrol congesLlon
ln a neLwork Lopology Laklng lnLo accounL Lhe varlous parameLers bandwldLh
packeL loss Lransmlsslon raLe eLc on uu1C proLocol
ln Lhls pro[ecL we alm Lo reduce neLwork congesLlon by comparlng Lhe dlfferenL
algorlLhms already exlsLlng and Lhereby lmplemenLlng an algorlLhm modeled and
belng used ln a neLwork slmulaLed ln nS2 slmulaLor ln Llnux

All Lhe Lhree algorlLhms1APCL8LnC Anu SACk also uses Lhe followlng
processes wlLh modlflcaLlons Lo avold congesLlon

8eferences
1 AnLlCongesLlon luzzy algorlLhm for Lrafflc conLrol of a Class of Lrafflc
neLworks
hLLp//wwwcompuLerorg/porLal/web/csdl/dol/101109/C8C2007138
2 AcLlve and programmable neLworkslLLL xplore
3 AlgorlLhm for 1C CongesLlon ConLrol 8y nlLln karLlk SLanford unlverslLy
LL 384y Sprlng 2003 rof nlck Mckeown rof 8ala[l rabhakar
4 1he 8esearch of AdapLlve neLwork CongesLlon ConLrol AlgorlLhm 8ased
on ACM
hLLp//wwwcompuLerorg/porLal/web/csdl/dol/101109/lll1A2009367