the importance of being opportunisticnms.lcs.mit.edu/~sachin/talks/copeantalk.pdfhigh performance...

High PerformanceSwitching and RoutingTelecom Center Workshop: Sept 4, 1997.

The Importance of Being Opportunistic

Sachin Katti

Dina Katabi, Wenjun Hu, Hariharan Rahul, and Muriel Medard

Bandwidth is scarce in wireless

Can we send more while consuming less bandwidth?

Current Approach

BobAlice

Relay

A lice’s packet B ob’s packet

B ob’s packet A lice’s packet

• Requires 4 transmissions

• Can we do it in 3 transmissions?

A Network Coding Approach

BobAlice

Relay



3 transmissions instead of 4 Save bandwidth

Network Coding

• Routers mix bits in packets, potentially from different flows

• Theoretically shown to achieve capacity for multicast

• No concrete results for unicast case

How to apply network coding?

• MulticastS R

R

R

• Given Sender & Receivers

• Given Flow Rate & Capacities

State-of-the Art

Min-Cost Flow Optimization

Find the routing, which dictates the encoding


• Multicast


• Given flow rate & capacities

State-of-the Art



• Unicast• Many Unknown Changing Sender & Receivers • Unknown and bursty flow rate

In Practice

?


• Multicast


• Given flow rate & capacities

State-of-the Art



In Practice

Opportunism

• Unicast• Many Unknown Changing Sender & Receivers • Unknown and bursty flow rate

Opportunism (1)Opportunistic Listening:

o Every node listens to all packetso It stores all heard packets for a limited time



D

A

B



o Node sends Reception Reports to tell its neighbors what packets it heard • Reports are annotations to packets • If no packets to send, periodically send reports

Opportunism (2)

Opportunistic Coding:o Each node uses only local informationo Use your favorite routing protocolo To send packet p to neighbor A, XOR p

with packets already known to A• Thus, A can decode

o But how to benefit multiple neighbors from a single transmission?

Efficient Coding

Arrows show next-hop

D

A

B

C

Efficient Coding


D

A

B

C

Bad Coding

C will get RED pkt but A can’t get B LU E pkt

Efficient Coding


D

A

B

C

Better Coding

Both A & C get a packet

Efficient Coding


D

A

B

C

Best Coding

A, B, and C, each gets a packet

To XOR n packets, each next-hop should have the n-1 packets encoded with the packet it wants

But how does a node know what packets a neighbor has?

• Reception Reports

• But reception reports may get lost or arrive too late

• Use Guessing o If I receive a packet I assume all nodes

closer to sender have received it

Putting It Together

D

A

BC

D

A

BC

Putting It Together

D

A

BC

• D on’t reorder packets in a flow Keeps TCP happy• No scheduling No packet is delayed

Putting It Together

Beyond Fixed Routes

BA DS

Route Chosen by Routing Protocol

B heard S ’s transm ission

S transmits No need for A transmission

Opportunistic Routing [BM05]• Piggyback on reception report to learn whether

next-hop has the packet• cancel unnecessary transmissions

Opportunism

• Unicast

• Flows arrive and leave at any time

• No knowledge of rate

• No assumption of smooth traffic

O

GN

A

C

K

M

EF

BDC

K

Performance

Emulation Environment• We use Emstar

o “R eal code” sim ulatoro 802.11 radioso Power Level: 200mWo 11Mbps bit rateo Simulated radio channel

Recall Our Simple Experiment

XOR

BobAlice

Relay

3 transmissions instead of 4 25% throughput increase



• Network coding requires broadcast• But 802.11 broadcast has no backoff more

collisions

0

100

200

300

400

500

1

No CodingXOR

Throughput (KB/s)

270 KB/s

Practical artifacts make network coding perform poorly

330 KB/s

Pseudo Broadcast

Piggyback on 802.11 unicast which has synchronous Acks and backoffo Each XOR-ed packet is sent to the MAC

address of one of the intended receivers

Our Solution:

• Ideally, design a back-off scheme for broadcast channels

• In practice, we want a solution that works with off-the-shelf 802.11 drivers/cards

XOR with Pseudo-Broadcast Improves Throughput

0

100

200

300

400

500

1

No CodingXOR

XOR with Pseudo-Broadcast

Throughput (KB/s)

330 KB/s270 KB/s

495 KB/s

Improvement is more than 25% because 802.11 MAC gives nodes equal bandwidth shares• Without coding, relay needs twice as much bandwidth• With coding, all nodes need equal bandwidth

Larger experiment

• 100 nodes

• 800mx800m

• Senders and receivers are chosen randomly

• Metric: Total Throughput of the Network

Opportunistic Coding vs. Current

No Coding

Our Scheme

Network Throughput (KB/s)

0

200

400

600

800

1000

1200

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58No. of flows

A Unicast Network Coding scheme that works well in realistic situations

Opp. Coding

Opp. Coding (no guessing)

No Coding

Preliminary Implementation Results• Linux Kernel• 802.11 MAC• Click Elements (part of Roofnet)• Only Opportunistic Coding & Pseudo

Broadcast (No opportunistic Listening)

Alice

Relay

Bob

Implementation Results

1

1.2

1.4

1.6

1.8

2

1 2 5.5 11802.11 Bit Rates (Mb/s)

Ratio of throughput with coding to without coding

Network Coding Doubles the Throughput

Conclusion• First implementation of network coding

in a wireless network

• Learned Lessonso Be opportunistic (greed is good!)o Can do a good job with multiple unicast o 5x higher throughput in congested networkso Preliminary implementation results show

throughput doubles

The Wireless EnvironmentMulti-hop wireless networks (e.g., Roofnet)

Opportunistic Coding vs. Current

0

200

400

600

800

1000

1200

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58

No Coding

Our Scheme

Network Throughput (KB/s)

No. of flowsA Unicast Network Coding scheme that works

well in realistic situations

the importance of being opportunisticnms.lcs.mit.edu/~sachin/talks/copeantalk.pdfhigh performance...

Documents