fat virtual access points srikanth kandula kate lin, tural badirkhanli and dina katabi

Post on 26-Mar-2015

228 Views

Category:

Documents

1 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Fat Virtual Access Points

Srikanth Kandula

Kate Lin, Tural Badirkhanli and Dina Katabi

Connect to the AP with the highest RSSI

State-of-the-art (802.11):

Wireless Link ~ 54Mbps(Theoretical Max)

AP Uplink ~ 2Mbps(DSL/Cable Modems)

Can Aggregate Bandwidth from nearby APs!Can Aggregate Bandwidth from nearby APs!

Problem 1: In Homes, Hotspots…

2+2+2+…2+2+2+…

Uplink Bottleneck

Load Imbalance

Unnecessary congestion; nearby APs are idle

Spread load

Individual changes help globally

Problem 2:

Divide Total Bandwidth Among Users Divide Total Bandwidth Among Users

20Mbps

15Mbps

At Work …

4Mbps4Mbps 7Mbps7Mbps

Join closest AP Join a Virtual AP, that is the sum of nearby APs

State-of-the-art: Abstraction:

1. User can aggregate bandwidth from all APs

2. Compete for total balance load across APs

1. User can aggregate bandwidth from all APs

2. Compete for total balance load across APs

2 Problems, 1 Solution

Realize a “fat virtual AP” withonly client-side changes

Basic Operation

2Mbps

20Mbps

But, what about receive?

2Mbps

Basic Operation

Drop

Power-Save Q

Pretend in power-save, so AP buffers when disconnected (similar to Chandra et. al. VirtualWiFi)

Divide Time and Data Across APs to get “Fat Virtual AP” Divide Time and Data Across APs to get “Fat Virtual AP”

2Mbps

20Mbps

2Mbps

Realizing a Fat Virtual AP is Hard

• Sustain TCP flows through each AP– Cannot lose packets yet Switch quickly

• Which APs to connect to and for how long?– Some APs are more valuable than others

• How to divide traffic across the APs?

FatVAP, an 802.11 driver design1. divides time across APs to maximize throughput

2. is transparent to APs and remote ends

FatVAP, an 802.11 driver design1. divides time across APs to maximize throughput

2. is transparent to APs and remote ends

FatVAP Overview

1. Scan for available APs2. Compute a schedule to divide time across APs3. Switch APs as per schedule4. Spread traffic by pinning flows to APs

Channel 1Channel 6 Channel 6

Channel 36

How much time to spend at an AP?

w

e

Useful fraction of time w

e≤

Achievable Bandwidths– end-to-end e, wireless w

Subsumes Wireless Link Quality, Contention at AP, APs uplink capacity

Subsumes Wireless Link Quality, Contention at AP, APs uplink capacity

How to Divide Time Across APs?

AP Bandwidth (Mbps) AP1 AP2 AP3

End-to-end Achievable 5 4 3

Wireless Achievable 5 8 8

Usable Fraction 100% 50% 38%

Optimal = 7 Mbps

Pick APs Greedily, on End-to-end rate

5 Mbps, 100% busy

! more bang for the buck if wireless b/w is large

Pick APs Greedily, on End-to-end ratePick APs Greedily, on Wireless rate

AP Bandwidth (Mbps) AP1 AP2 AP3 AP4 AP5 AP6

End-to-end Available 1 1 1 1 1 4.5

Wireless Available 5 5 5 5 5 4.5

Usable Fraction 20% 20% 20% 20% 20% 100%

5 Mbps, 100% busy

! cost to switch is ≈ 5 ms! can’t linger too long (100ms period)

No Greedy Solution!No Greedy Solution!

How to Divide Time Across APs?

Only 75% usable

Say, fi is fraction of time at APi

⎡ ⎤( )∑

≤+

≤≤

DsfDf

w

efts

wf

ii

i

ii

iifi

0..

max

Usefulness Constraint

Value (Bandwidth)

Cost (Time)

Let s be switching time and D be the period

(pseudo)-polynomial solution(pseudo)-polynomial solution

Like Bin Packing, maximize value with bounded cost!

But, How to Estimate Bandwidths?Wireless Achievable

Client TX Queue

Time from head of tx queue to end of transmission (ack)

Naively– send-rate of probe burst, APs report load

AP Buffers

Idea: Use synchronous acks

w

But, How to Estimate Bandwidths?Wireless Achievable End-to-end

Client TX Queue

Time from head of tx queue to end of transmission (ack)

Naively, send-rate of probe burst or APs report load

AP Buffers

Idea: Use synchronous acks

e

tCount bytes rcvd in a window

But, How to Estimate Bandwidths?

Wireless Achievable End-to-end

Client TX Queue

Time from head of tx queue to end of transmission (ack)

Naively, send-rate of probe burst or APs report load

AP Buffers

Idea: Use synchronous acks

e

tCount bytes rcvd in a windowMay not receive data alwaysIdea: only count back-to-back large packets!

How to Spread Traffic Across APs?

How to Spread Traffic Across APs?

Put flows through all APs• virtualize 802.11 state• an IP for each interface• toggle APs (and channels)

By default, kernel sends all traffic to one AP

T-Mobile 192.168.3.0/24

MIT 128.30.79.0/24

Hardware (Wireless Card)802.11 State

AP1 AP2

AP1 State AP2 State

Two Interfaces

Toggler

How to Spread Traffic Across APs?

Put flows through all APs• virtualize 802.11 state• an IP for each interface• toggle APs (and channels)

By default, kernel sends all traffic to one AP

• Spread flows to APs• Fast header re-writing

Hardware (Wireless Card)

AP1 AP2

AP1 State AP2 State

Two Interfaces

Toggler

Spreader

Distribute load w/o changing APs and applicationsDistribute load w/o changing APs and applications

Switching Quickly Without Drops

AP1 AP2A Driver For Each Interface?– warm-up cost on switch

– one instance + soft-switch

One Driver

Switching Quickly Without Drops

A Driver For Each Interface?– warm-up cost on switch

– one instance + soft-switch

802.11 Control Packets– Isolate Transitions

AP1 AP2

Hardware (Wireless Card)

AP1 State AP2 State

Re-send AUTHSend AUTH

Switching Quickly Without Drops

A Driver For Each Interface?– warm-up cost on switch

– one instance + soft-switch

802.11 Control Packets– Isolate Transitions

Pkts stuck in driver at switch– Private Queues

AP1 AP2

1. Delay Switch till pkts drain

2. Drop Packets (madwifi)

Switching Quickly Without Drops

A Driver For Each Interface?– warm-up cost on switch

– one instance + soft-switch

802.11 Control Packets– Isolate Transitions

Pkts stuck in driver at switch– Private Queues

AP1 AP2

Attach/Detach Queue = Pointer Swap

Enables high-rate TCPs through multiple APs Enables high-rate TCPs through multiple APs

FatVAP Realizes a Fat Virtual AP

• Which APs to connect to and for how long?– Estimate Bandwidths, Solve Optimization

• How to divide traffic across the APs?– Virtualize, Pin Flows to APs, rewrite headers

• Switch quickly but without losing packets– In-driver, Private Queues, Isolation

And, with only client-side changesAnd, with only client-side changes

Related Work

VirtualWiFi (Microsoft Research)AP Selection (Intel Research, U Michigan)SyncScan (UCSD)MadWifi (open-source)

o Divide Time across APs to maximize throughputo Sustain TCP flows through multiple APso Transparently spread traffic across APs

Results

Experimental Setup

Compare FatVAP driver with unmodified MadWifi• Scenarios

– Testbed built from Cisco, NetGear and MadWifi APs– Residential deployments– Commercial hotspots

• Traffic– Long-lived TCP flows– BitTorrent (Azureus client, Planetlab peers)– Mimic Web Browsing (modified WebStone)

Can FatVAP Aggregate Bandwidth?

6Mbps

5.8

11.4

17

20.6 20.5

5.8 5.8 5.8 5.8 5.8

0

5

10

15

20

25

1 2 3 4 5

FatVAP Unmodified MadWifi

Thr

ough

put

(Mb/

s)

Number of APs Aggregates end-to-end up to the wireless bottleneck Aggregates end-to-end up to the wireless bottleneck

~22 Mbps

Can FatVAP Balance Load?

12Mbps2Mbps

C1 C2 C3 C4 C5

Can FatVAP Balance Load?

12Mbps2Mbps

Thr

ough

put

(Mb/

s)5

4

3

2

1

0

FatVAP

2.93.5

3.1 2.8 2.7

.9.9

4.4

3.33.8

Unmodified MadWifi

Simplifies Network Deployment!Simplifies Network Deployment!

C1 C2 C3 C4 C5

C1 C2 C3 C4 C5 C1 C2 C3 C4 C5

Can FatVAP Adapt to Changes?

Thr

ough

put

(Mb/

s)

5Mbps 15Mbps

Re-adjusts time@AP as necessary Re-adjusts time@AP as necessary

Contributions

A new model for managed 802.11 LANs– Aggregate uplink, Balance load

First to realize a fat virtual AP– Divide time and traffic across APs

Transparent to APs, applications, servers

top related