Scaling WiFi Performance for Large Audience Environments Arpit Gupta Jeongki Min and Injong Rhee NC State University Interesting Faster WiFi Manageable solution ID: 436001
Download Presentation The PPT/PDF document "WiFox" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
WiFox: Scaling WiFi Performance for Large Audience Environments
Arpit Gupta, Jeongki Min and Injong RheeNC State UniversitySlide2
Interesting ??
Faster WiFi !!
Manageable solution !!
WiFox
:
AP-only S/W solutionSlide3
Large Audience Environments (LAEs)Slide4
Large Audience Environments
Any location with a large WiFi user population WBA projected a growth rate of 350% per year for such WiFi deploymentsVarious successful deployment models such as Boingo-Google,
Mobily-Aruba etc. already exist
Source: WBA(Wireless Broadband Alliance)
Source: Wireless Broadband Alliance (WBA)Slide5
What about User Experience ?
As number of active clients increases,
user experience diminishes significantlySlide6
Problem AnatomySlide7
Well Known Factors Contention and collision
Increases with growing competitionRate DiversitySlower STA slows down all other STAsVarious fairness realizations like WFQ, TBR etc.Random Losses and TCP performanceTCP treats packet losses as congestion signalsUsage of TCP ECN and proxy servers isolate wired and wireless networksTraffic AsymmetrySlide8
Traffic Asymmetry
Downlink Traffic dominates for 90% of data tracesSlide9
Traffic Asymmetry
Majority of data traffic is Web basedDownlink Traffic dominates compared to uplink
Majority of data packets are for HTTP based web activitiesSlide10
Traffic AsymmetryIn scope of our problem it is:
Downlink/Uplink AsymmetryChannel Access AsymmetryImplications:Packets spend more time at AP’s TxQ
Frequent packet dropsSlide11
Wireless Channel
Uplink Traffic
Downlink Traffic
Channel access for uplink traffic is moreSlide12
Performance Bottleneck
TxQ saturates as associated clients increase
Associates UsersSlide13
Goodput Performance
Traffic Asymmetry causes TxQ saturation resulting in poor goodput performanceSlide14
Possible SolutionsSlide15
Wireless Channel
Uplink Traffic
Downlink Traffic
Equal Channel Access for Uplink/DownlinkSlide16
Wireless Channel
Uplink Traffic
Downlink Traffic
Statically Assign Higher Priority to Downlink Traffic Slide17
Wireless Channel
Uplink Traffic
Downlink Traffic
Dynamically
Assign Higher Priority to Downlink Traffic Slide18
Our SolutionSlide19
Priority Control
Packet A
ACK
DIFS
DIFS H
Channel Access
N Slots
N Slots
Busy medium
Busy medium
Wins Contention
STA C
STA B
STA A
Smaller IFS
Class
C
W
min
CWmax
AIFS
TXOP Limit
AP
1
5
1
64
STAs
5
10
N/A
N/ASlide20
Linear Scaling Priority Model
Priority LevelLinear relationship between Goodput
and Priority LevelSlide21
Adaptive Prioritization
100
ms
D
H
D
H
D
D
D
H
D
D
Time
High Priority
Default Priority
Decision Points
Priority Level 3Slide22
EvaluationSlide23
Test Bed
2600 Sqft Area with multiple APs, 45 STAsNetgear 802.11 b/g wireless cards with Atheros chipsets and MADWIFI driversLatency emulation using DummyNet
Modified SURGE
for web traffic generation
Requests inter-arrival closely follows the ones observed for SIGCOMM traces
Uplink UDP traffic using
Iperf
to emulate Background TrafficSlide24
Performance
Downlink: N/W Goodput
WiFox
W/O
WiFox
Significant improvement in Network’s Downlink
GoodputSlide25
Performance
Experiment involves sending 25 requests and observe response for 4 minutes durationRequest Serving rate is 4 times better than NPCSlide26
Robustness
Performance in presence of Multiple Aps ??WiFox
w/o
WiFoxSlide27
Robustness
w/o WiFox
WiFox
Unfair Distribution
Fairness Realization
??Slide28
Performance: TxQ Dynamics
WiFoxw/o WiFoxSlide29
ConclusionWiFox Delivers
:Deployment Ready SolutionEnhanced user experience with 400-700% Downlink Goodput improvements40-60% faster response timeOpen Problems:Characterizing asymmetry problem for 802.11nSupport for real time applications like chats etc.
QoSSlide30
30
Merci !!Slide31
Multi AP Scenario
D
D
D
D
D
D
D
D
D
D
D
D
D
AP 1 ( Priority Level 4)
AP 2 ( Priority Level 3)
D
D
D
D
D
D
D
D
D
D
D
D
AP 1 ( Priority Level 3)
AP 2 ( Priority Level 5)
time
100
msSlide32
Performance: InsightEnables AP to switch to HIGH priority state under heavy load
Avoids TxQ saturationSignificant reduction in ReTx rate compared to stock WiFi implementationSlide33
Robustness: Uplink TrafficScenarios where few users indulge in heavy uplink activities like video uploading, cloud synchronization etc.Slide34
Observations
Source: Rodrig et al.