nc-cell: network coding-based content distribution in cellular networks for cloud applications

5
NC-CELL: Network Coding-based Content Distribution in Cellular Networks for Cloud Applications Claudio Fiandrino * , Dzmitry Kliazovich * , Pascal Bouvry * and Albert Y. Zomaya * University of Luxembourg - The University of Sydney Introduction Mobile cloud applications is one of the fastest growing markets: Mobile data traffic will rise up to 15 EB per month by 2018 By 2017 4.4 billion people will use mobile cloud applications $ 45 billion market 90% of all mobile data traffic by 2018 2013 2014 2015 2016 2017 2018 0 50% 100% 18 % 17 % 15 % 14 % 12 % 10 % 82 % 83 % 85 % 86 % 88 % 90 % Non-Cloud Cloud Source: Cisco Visual Networking Index: Global Mobile Data Traf- fic Forecast Update, 2013-2018 Network Coding in Cellular Networks Optimizing information delivery of flows with overlapping or partially overlapping content. Important Facts Geographically co-located users Mobile cloud applications’ content Advertisement Maps Meteo Google Now Network coding to combine information flows The NC-CELL Technique Evolved Packet Core E-UTRAN Cloud Internet P-GW MME S-GW LTE Network UE Buffers Network Coding Key aspects: Monitor and cache in transit traffic Identify coding opportunities Use of network coding to combine packets Coding opportunities eNodeBs can distribute information needed by two or more users with a single coded transmission. Optimal allocation for content distribution t u 1 u 2 u k Users c 1,1 c 2,2 c k,k c k+1,1 c k+2,2 c 2k,k c n-k,1 c n-k+1,2 c n,k c 1,1 c 2,2 c k-1,k-1 c k,k c k+1,1 c k+2,2 c 2k-1,k-1 c 2k,k c n-k,1 c n-k+1,2 c n-1,k-1 c n,k Individual Transmission Encoded Transmission - Results Number of transmissions at eNodeB: measuring throughput improvement 2 4 6 8 10 200 400 600 800 1 000 0 0.2 0.4 0.6 0.8 1 ·10 4 k n Num. Transmissions NC-CELL Enabled NC-CELL Disabled Coding gain: measuring the benefit of network coding 10 100 200 300 400 500 0 1 2 3 4 5 6 7 k =2 k =4 k =6 k =8 k = 10 Num. common chunks n Coding gain η Conclusion NC-CELL provides efficient content distribu- tion for cloud applications in mobile cellular networks. It enables eNodeB nodes to mon- itor and cache in transit traffic and exploits network coding to combine packets. Acknowledgements The authors would like to acknowledge the funding from National Research Fund, Luxem- bourg in the framework of ECO-CLOUD project (C12/IS/3977641).

Upload: claudio-fiandrino

Post on 07-Aug-2015

48 views

Category:

Presentations & Public Speaking


0 download

TRANSCRIPT

Page 1: NC-CELL: Network Coding-based Content Distribution in Cellular Networks for Cloud Applications

NC-CELL: Network Coding-based Content Distributionin Cellular Networks for Cloud ApplicationsClaudio Fiandrino∗, Dzmitry Kliazovich∗, Pascal Bouvry∗ and Albert Y. Zomaya†

∗University of Luxembourg - †The University of Sydney

Introduction

Mobile cloud applications is one of the fastestgrowing markets:

• Mobile data traffic will rise up to 15 EB permonth by 2018

• By 2017 4.4 billion people will use mobilecloud applications

• $ 45 billion market• 90% of all mobile data traffic by 2018

2013 2014 2015 2016 2017 20180

50%

100%18% 17% 15% 14% 12% 10%

82% 83% 85% 86% 88% 90%

Non-CloudCloud

Source: Cisco Visual Networking Index: Global Mobile Data Traf-

fic Forecast Update, 2013-2018

Network Coding in CellularNetworks

Optimizing information delivery of flows withoverlapping or partially overlapping content.

Important Facts

• Geographically co-located users• Mobile cloud applications’ content• Advertisement• Maps• Meteo• Google Now

Network coding to combine information flows

The NC-CELL Technique

Evolved Packet CoreE-UTRAN

CloudInternetP-GW

MME

S-GW

LTE Network

UE

Buffers Network Coding

Key aspects:

• Monitor and cache in transit traffic• Identify coding opportunities• Use of network coding to combine packets

Coding opportunities

eNodeBs can distribute information neededby two or more users with a single codedtransmission.

Optimal allocation for content distribution

t

u1

u2

⋮uk

Use

rs

c 1,1

c 2,2

⋱c k

,k

c k+1,1

c k+2,2

⋱c 2

k,k

⋮ ⋮c n−k,1

c n−k+

1,2

⋱c n

,k

c 1,1⊕c 2

,2 ⋮c k−1,k−1⊕

c k,k

c k+1,1

⊕c k+2,2 ⋮

c 2k−1

,k−1⊕

c 2k,

k

c n−k,1

⊕c n−k+

1,2

⋮c n−1,k−1⊕

c n,k

Individual Transmission Encoded Transmission-

Results

Number of transmissions at eNodeB:measuring throughput improvement

24

68

10 200 400 600 800 1 000

0

0.2

0.4

0.6

0.8

1

·104

kn

Num

.Tra

nsm

issi

ons

NC-CELL Enabled

NC-CELL Disabled

Coding gain:measuring the benefit of network coding

10 100 200 300 400 5000

1

2

3

4

5

6

7

k = 2

k = 4

k = 6

k = 8

k = 10

Num. common chunks n

Cod

ing

gainη

Conclusion

NC-CELL provides efficient content distribu-tion for cloud applications in mobile cellularnetworks. It enables eNodeB nodes to mon-itor and cache in transit traffic and exploitsnetwork coding to combine packets.

Acknowledgements

The authors would like to acknowledge thefunding from National Research Fund, Luxem-bourg in the framework of ECO-CLOUD project(C12/IS/3977641).

Page 2: NC-CELL: Network Coding-based Content Distribution in Cellular Networks for Cloud Applications

NC-CELL: Network Coding-based Content Distributionin Cellular Networks for Cloud ApplicationsClaudio Fiandrino∗, Dzmitry Kliazovich∗, Pascal Bouvry∗ and Albert Y. Zomaya†

∗University of Luxembourg - †The University of Sydney

Encoding and DecodingProcedure

Encoding

• At eNodeB• Combining only content not packetheaders

• Delivery to all interested users throughPDSCH (primary and secondary users)

Packet received

Encoded packet?

Can be decodedusing packet

stored in buffer?Store in buffer

Performdecoding

Send packet toupper layer

Can any of thepreviously stored

coded packetsbe decoded?

End procedure

No

No

Yes

Yes

Yes

No

Decoding

• At users’ side• Immediate if one of the two contentpackets is already available

NC-CELL Operation

• Two users: UE1 and UE2• Need to retrieve content A and content B from cloud application• eNodeB exploits coding opportunity

UE1 UE2 eNodeB Cloud Application

Request A Packet request

Send content APacket AUE1

Cache and forward AUE1Packet AUE1

Process and store AUE1

Request BPacket request

Send content BUE2Packet BUE2

Cache and forward BUE2Packet BUE2

Process and store BUE2

Request B Packet request

Send content BUE1Packet BUE1

Check if B is in buffer

Coding (A ⊕ B)UE1,UE2Packet (A ⊕ B)UE1,UE2

Decode A using BUE2Decode B using AUE1

Protocol Stack

Encoding

• After GTP header removal• Before PDCP performs IP headercompression

Decoding

• Primary users receive and decodeimmediately

• Secondary users do not discard frame

PHY

MAC

RLC

PDCP

IP

NC

APP

PHY

MAC

RLC

PDCP

NC

User Equipment eNodeB

LTE-Uu

Contact Information

• Claudio Fiandrino• Email: [email protected]• Phone: +352 46 6644 5531

Page 3: NC-CELL: Network Coding-based Content Distribution in Cellular Networks for Cloud Applications

NC-CELL: Network Coding-based Content Distributionin Cellular Networks for Cloud ApplicationsClaudio Fiandrino∗, Dzmitry Kliazovich∗, Pascal Bouvry∗ and Albert Y. Zomaya†

∗University of Luxembourg - †The University of Sydney

Encoding and DecodingProcedure

Encoding

• At eNodeB• Combining only content not packetheaders

• Delivery to all interested users throughPDSCH (primary and secondary users)

Packet received

Encoded packet?

Can be decodedusing packet

stored in buffer?Store in buffer

Performdecoding

Send packet toupper layer

Can any of thepreviously stored

coded packetsbe decoded?

End procedure

No

No

Yes

Yes

Yes

No

Decoding

• At users’ side• Immediate if one of the two contentpackets is already available

NC-CELL Operation

• Two users: UE1 and UE2• Need to retrieve content A and content B from cloud application• eNodeB exploits coding opportunity

UE1 UE2 eNodeB Cloud Application

Request A Packet request

Send content APacket AUE1

Cache and forward AUE1Packet AUE1

Process and store AUE1

Protocol Stack

Encoding

• After GTP header removal• Before PDCP performs IP headercompression

Decoding

• Primary users receive and decodeimmediately

• Secondary users do not discard frame

PHY

MAC

RLC

PDCP

IP

NC

APP

PHY

MAC

RLC

PDCP

NC

User Equipment eNodeB

LTE-Uu

Contact Information

• Claudio Fiandrino• Email: [email protected]• Phone: +352 46 6644 5531

Page 4: NC-CELL: Network Coding-based Content Distribution in Cellular Networks for Cloud Applications

NC-CELL: Network Coding-based Content Distributionin Cellular Networks for Cloud ApplicationsClaudio Fiandrino∗, Dzmitry Kliazovich∗, Pascal Bouvry∗ and Albert Y. Zomaya†

∗University of Luxembourg - †The University of Sydney

Encoding and DecodingProcedure

Encoding

• At eNodeB• Combining only content not packetheaders

• Delivery to all interested users throughPDSCH (primary and secondary users)

Packet received

Encoded packet?

Can be decodedusing packet

stored in buffer?Store in buffer

Performdecoding

Send packet toupper layer

Can any of thepreviously stored

coded packetsbe decoded?

End procedure

No

No

Yes

Yes

Yes

No

Decoding

• At users’ side• Immediate if one of the two contentpackets is already available

NC-CELL Operation

• Two users: UE1 and UE2• Need to retrieve content A and content B from cloud application• eNodeB exploits coding opportunity

UE1 UE2 eNodeB Cloud Application

Request A Packet request

Send content APacket AUE1

Cache and forward AUE1Packet AUE1

Process and store AUE1

Request BPacket request

Send content BUE2Packet BUE2

Cache and forward BUE2Packet BUE2

Process and store BUE2

Protocol Stack

Encoding

• After GTP header removal• Before PDCP performs IP headercompression

Decoding

• Primary users receive and decodeimmediately

• Secondary users do not discard frame

PHY

MAC

RLC

PDCP

IP

NC

APP

PHY

MAC

RLC

PDCP

NC

User Equipment eNodeB

LTE-Uu

Contact Information

• Claudio Fiandrino• Email: [email protected]• Phone: +352 46 6644 5531

Page 5: NC-CELL: Network Coding-based Content Distribution in Cellular Networks for Cloud Applications

NC-CELL: Network Coding-based Content Distributionin Cellular Networks for Cloud ApplicationsClaudio Fiandrino∗, Dzmitry Kliazovich∗, Pascal Bouvry∗ and Albert Y. Zomaya†

∗University of Luxembourg - †The University of Sydney

Encoding and DecodingProcedure

Encoding

• At eNodeB• Combining only content not packetheaders

• Delivery to all interested users throughPDSCH (primary and secondary users)

Packet received

Encoded packet?

Can be decodedusing packet

stored in buffer?Store in buffer

Performdecoding

Send packet toupper layer

Can any of thepreviously stored

coded packetsbe decoded?

End procedure

No

No

Yes

Yes

Yes

No

Decoding

• At users’ side• Immediate if one of the two contentpackets is already available

NC-CELL Operation

• Two users: UE1 and UE2• Need to retrieve content A and content B from cloud application• eNodeB exploits coding opportunity

UE1 UE2 eNodeB Cloud Application

Request A Packet request

Send content APacket AUE1

Cache and forward AUE1Packet AUE1

Process and store AUE1

Request BPacket request

Send content BUE2Packet BUE2

Cache and forward BUE2Packet BUE2

Process and store BUE2

Request B Packet request

Send content BUE1Packet BUE1

Check if B is in buffer

Coding (A ⊕ B)UE1,UE2Packet (A ⊕ B)UE1,UE2

Decode A using BUE2Decode B using AUE1

Protocol Stack

Encoding

• After GTP header removal• Before PDCP performs IP headercompression

Decoding

• Primary users receive and decodeimmediately

• Secondary users do not discard frame

PHY

MAC

RLC

PDCP

IP

NC

APP

PHY

MAC

RLC

PDCP

NC

User Equipment eNodeB

LTE-Uu

Contact Information

• Claudio Fiandrino• Email: [email protected]• Phone: +352 46 6644 5531