PowerPoint Presentation

Cooperative Interference Management in Wireless NetworksI-Hsiang Wangcole Polytechnique Fdrale de Lausanne (EPFL)IE/INC SeminarChinese University of Hong Kong, Hong KongMay 14, 2012Thanks for the nice introduction. Its my great pleasure to be here. Today I would like to share with you some of my research on cooperative interference management in wireless networks, from an information theoretic point of view.0Experience with Wireless?05/14/12Wang, IE/INC Seminar, CUHK1

Skype isso choppy!

My e-mail wont refresh

Why is mytethered connectionso slow?!

I needdirections now!

18XMonthly Mobile Data Traffic

Past Challenges in Wireless

1.Fading

2.Multiplexing(Multiple Access)Past 15 years:MIMOOpportunistic communication

Wideband SystemsCDMA, OFDMASystem Gain: pertains to point-to-point/single-cell performance05/14/12Wang, IE/INC Seminar, CUHK2Example: cellular networkBase Station (BS)MobileWireless communication has been a thriving industry over the past 15 years. We engineers have taken some challenges in the past to make it successful.As an example, lets focus on cellular networks, where mobiles want to communicate with the base station reliably. The first challenge we were faced with, is that the wireless channel is highly unreliable. It is fading over time and frequency. Fading addressed an unique challenge for system engineers to design reliable communication schemes. On the other hand, typically a base station has to serve more than one mobile. Therefore, how to do multiplexing among mobile users economically was another challenge. Over the past 15 years, techniques such as MIMO, where we install multiple antennas onto terminals to exploit spatial diversity, and wideband systems such as CDMA and OFDMA, have been proposed and deployed. As a result, these two challenges have been resolved, and the system performance is boost up so that more users and higher throughout can be supported. However, the system gain from these advances only pertains to point-to-point or single-cell performance.2

A Current Key Challenge

05/14/12Wang, IE/INC Seminar, CUHK31.Fading 2.Multiplexing3.InterferenceSignal not intended to the receiving terminal (intercell)Performance of todays wireless system is majorly limited by interference!As # of mobile & BS Bad news: capacity of two-user interference channel remains open for 35+ years3Narrowband system (GSM): Orthogonalize itPoor frequency reuse; shortage of resource

Wideband system (CDMA, OFDMA): Treat it as noiseDegrades if interferences get strong (cell-boundary users)

Opportunities neglected in traditional paradigm

Cooperation; cooperative interference managementInterference: Major Bottleneck05/14/12Wang, IE/INC Seminar, CUHK44

Opportunities in Cellular Systems05/14/12Wang, IE/INC Seminar, CUHK5

BackhaulDSL, Optical Fiber, Microwave

Distributed MIMOCaveat: cooperation is limitedInformation theory:degree-of-freedom gainpower gainvirtual5Opportunities in Wireless LAN05/14/12Wang, IE/INC Seminar, CUHK6

InterferenceCooperationRadios can overhearIdle or additional devices (femto-cell)Caveat: cooperation is limited6Interference: currently the major bottleneck

Cooperative interference managementOpportunities neglected in traditional paradigmCooperation among terminals helps mitigate interferenceThe rate at which they cooperate, however, is limited

Fundamental information theoretic question: How much capacity gain under limited cooperation?Answered in this talk!Short Recap7Wang, IE/INC Seminar, CUHK05/14/127Overview of Studied Scenarios 05/14/12Wang, IE/INC Seminar, CUHK8

Backhaul

Canonical Setting: Two Transmitters Two Receivers, Orthogonal Coop.

UplinkDownlinkBSBSGeneral Setting: Two Sources Two DestinationsCoop. over Network

WirelessArbitrary # of NodesLens of Information Theory8Rest of this talkFocus on the canonical two-Tx-two-Rx setting

Approximate characterization of capacity region

Gain from limited cooperationQualitative interpretationQuantitative understanding

Optimal scheme in high-SNR regime

Two unicast sessions over layered wireless networks

05/14/12Wang, IE/INC Seminar, CUHK9Gaussian Interference ChannelAll nodes know the whole channelDirect link: Signal-to-Noise Ratio (SNR)Cross link: Interference-to-Noise Ratio (INR)Capacity is open for 35+ yearsCapacity region characterized to within 1 bits/s/Hz [Etkin et.al.07]

10Wang, IE/INC Seminar, CUHK05/14/12Gaussian Interference Channel (GIC)10GIC with Limited CooperationAll nodes know the whole channelCooperation links are noise-free,Orthogonal to each other and the interference channelOf finite capacities and respectively

Out-of-Band Transmitter Cooperation

11Wang, IE/INC Seminar, CUHK05/14/12

Out-of-Band Receiver Cooperation11

Capacity to within a Bounded Gap05/14/12Wang, IE/INC Seminar, CUHK12Rx Cooperation: Capacity region to within 2 bits/s/Hz [W&Tse09]Tx Cooperation: Capacity region to within 6.5 bits/s/Hz [W&Tse10]The first uniform approximation result on the capacity region of GIC with Rx cooperation or Tx cooperationAs SNR goes to infinity, gap is negligible: Capacity at high SNR!

Joint work with David Tse12Nature of the Gain from Coop05/14/12Wang, IE/INC Seminar, CUHK13

Linear RegionCooperation is efficientSaturation RegionCooperation is inefficientdegree-of-freedom gainpower gainReceiver CooperationSymmetric Case

Focus on the Linear RegionWirelessWirelessBackhaul13Coop. Efficiency in Int. Mitigation05/14/12Wang, IE/INC Seminar, CUHK14

degree-of-freedom gainpower gainSlope is either 1 or , depending on channel strengthCorollary (DoF Gain) Depending on the channel strength, either

One additional coop bit buys one more bit over-the-air, or

Two additional coop bits buy one more bit over-the air14High-SNR Approximate Capacity05/14/12Wang, IE/INC Seminar, CUHK15

Capacity per user

With cooperation

Without cooperation [Etkin et.al.07]

High-SNR Normalized Capacity

The same picture for Tx cooperation!

The same definition for Tx cooperation!Normalized Capacity (by the interference-free capacity)Strength of InterferenceNormalized Backhaul Capacity15

Linear Deterministic Model05/14/12Wang, IE/INC Seminar, CUHK16

[Avestimehr et.al.07]

Captures the interaction of signals in wireless networksApproximate!

Unit Tx powerUnit noise power(Roughly speaking), # of bits that is above the noise level

16

One Cooperation bit buys one bit05/14/12Wang, IE/INC Seminar, CUHK17

Slope = 1

Tx1Tx2Rx1Rx2

Two cooperation bits buy two more bits

commonprivate

17

Two Cooperation bits buy one bit05/14/12Wang, IE/INC Seminar, CUHK18

Slope = 1/2

Tx1Tx2Rx1Rx2Two cooperation bits buy one more bit

18Near Optimal Coding SchemeSuperposition codingCommon-private split facilitates partial interference cancellationPrivate interference is at or below noise level at the unintended receiver05/14/12Wang, IE/INC Seminar, CUHK19

Blue: commonRed: privateQuantize-Map-ForwardQuantize at private+noise signal levelJointly decode message and quantization codeword

19

Uplink-Downlink Reciprocity05/14/12Wang, IE/INC Seminar, CUHK20

Primary Downlink ScenarioDual Uplink ScenarioChannel matrix HermitianSwap two cooperation linksCapacity regions are within a bounded gap20ReflectionsJust two special cases!Techniques in the proofs are tailored for specific problemsSingle-flow problem:Solved in the linear deterministic scenario, for arbitrary network topology [Avestimehr et.al.07]Max Flow = Min CutIs there a common principle/approach to solve a richer set of problems?05/14/12Wang, IE/INC Seminar, CUHK21IC with Rx Coop[W & Tse09]IC with Tx Coop[W & Tse10]Multiple Information Flows over Networks21

Multiple-Unicast Wireless NetworkK=1, single unicast [Avestimehr et al.07]Max-Flow = Min-Cut Random linear coding achieves min-cut

Insights from network coding in wired networksExtends to single multicast05/14/12Wang, IE/INC Seminar, CUHK22

WirelessArbitrary # of Nodes

22

Two Unicast SessionsTwo Unicast Wired Networks (directed)Capacity unknown!MinCut(si; di) = 1: Capacity characterized [Wang & Shroff IT10]Cut-set bound is not tightRouting or random linear network coding no longer sufficeOnly a bounded # of edges has to take special operations

05/14/12Wang, IE/INC Seminar, CUHK23

WirelessArbitrary # of NodesWired (integer edge capacity)23Two-Unicast Wired NetworksThe region must be one of the two:

Necessary and sufficient conditions are given

05/14/12Wang, IE/INC Seminar, CUHK2424An Analog in Wireless Two-UnicastLayered linear deterministic network MinCut(si; di) = 1, i = 1,2

Time sharing inner boundTrivial outer bound

Capacity?05/14/12Wang, IE/INC Seminar, CUHK25

ExampleLayer 0Layer 1Layer 2Baseline25Main Result

05/14/12Wang, IE/INC Seminar, CUHK26Layered linear deterministic network MinCut(si; di) = 1, i = 1,2Characterize the two-unicast capacity regionMust be one of the following five

Joint work with S. Kamath and D. Tse26Key Idea in the ResultSome nodes are special!

Achievability all nodes do random linear coding,Except 4 of these nodes

Outer Bound suffices to check their propertiesNo need to check others

Systematic approach to identify them05/14/12Wang, IE/INC Seminar, CUHK2727ConclusionCooperative Interference ManagementCapacity characterized approximatelyLinear vs. Saturation RegionCooperation Efficiency in Linear Region1 Coop bit buys 1 bit over-the-air or 2 Coop bits buy 1 bit over-the-airInsights to cellular system design with limited backhaul

General Two-unicast Wireless NetworksLayered linear deterministic network, individual min-cut constrained to be 1: Capacity characterizedGeneral case: open05/14/12Wang, IE/INC Seminar, CUHK2828Thank You!More details can be found athttp://sites.google.com/site/ihsiangw/Email: i-hsiang.wang@epfl.ch