Flexible Backhaul Design for Cellular Interference Management · 2015-03-25 · Flexible Backhaul...

Post on 03-Apr-2020

4 views 2 download

transcript

Flexible Backhaul Design for Cellular Interference Management

Venu Veeravalli

Director, Illinois Center for Wireless Systems

Coordinated Science Lab

ECE Department University of Illinois at Urbana-Champaign

Interference in Cellular Networks

Interference Management is critical in dense wireless networks

Veeravalli – ICNC 2/18/15

K-User Interference Channel

Tx1 Rx1

Tx2

Txk

Rx2

Rxk

Veeravalli – ICNC 2/18/15

Information Theory & Interference Management

•  Exact characterization of capacity o  Very hard problem; still mostly

open

•  Approximate characterization of capacity o  Within constant number of bits/sec o  Provides some architectural

insights

•  Degrees of freedom (DoF) o  Pre-log factor of sum-capacity in

high SNR regime o  Number of interference free

sessions per channel use o  Simplest of the three, but can

provide useful insights

Veeravalli – ICNC 2/18/15

Tx1 Rx1

Tx2

Txk

Rx2

Rxk

DoF and PUDoF for K-User IC

•  User orthogonalization o  Every user gets an

interference free channel once every K channel uses

o  DoF = 1 or Per User DoF (PUDoF) = 1/K.

•  Outer Bound on DoF [Host-Madsen, Nosratinia ‘05] o  DoF ≤ K/2 or PUDoF ≤ 1/2

•  Amazingly, this outer bound is

achievable via linear interference suppression!

Interference Alignment [Cadambe & Jafar ‘08]

Veeravalli – ICNC 2/18/15

Tx1 Rx1

Tx2

Txk

Rx2

Rxk

Linear Transmit/Receive Strategies

Interference Channel

End-to-End matrix is Diagonal è No Interference!

# streams = Size of the Diagonal matrix

Channel

⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦

[1,1] [1,2] [1,3]

[2,1] [2,2] [2,3]

[3,1] [3,2] [3,3]

H H H

H H H

H H H

Transmit Beams

⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦

[1]

[2]

[3]

V 0 0

0 V 0

0 0 V

Receive Beams

U[1] 0 00 U[2] 00 0 U[3]

⎢⎢⎢⎢

⎥⎥⎥⎥

H

Veeravalli – ICNC 2/18/15

Interference Alignment with Symbol Extensions (Cadambe & Jafar)

Tx1

Tx2

Tx3

Rx1

Rx2

Rx3

3 Symbol Extensions

4 interference free streams è PUDoF = 4/9

Veeravalli – ICNC 2/18/15

Asymptotic Interference Alignment

# symbol extensions

PUDoF

PUDoF of 0.5 is achieved asymptotically

0 20 40 60 80 1000.44

0.45

0.46

0.47

0.48

0.49

0.5

Veeravalli – ICNC 2/18/15

Asymptotic Interference Alignment

# symbol extensions

PUDoF

0 200 400 600 800 1000 1200 14000.32

0.34

0.36

0.38

0.4

0.42

0.44

0.46

0.48

0.53 User

4 User

Choi, S.W. and Jafar, S.A. and Chung, S.Y. , “On the beamforming design for efficient interference alignment”, IEEE Communication Letters , 2009

Veeravalli – ICNC 2/18/15

Interference Alignment: Summary

+  Achieves optimal PUDoF for fully connected channel

-  Requires global channel state information (CSI)

-  Requires large number of symbol extensions

Veeravalli – ICNC 2/18/15

Tx Cooperation Through the Backhaul: DoF Analysis

Transmitter Cooperation – 2 Users

No Cooperation Per user DoF = 1/2

Time Sharing

Full Cooperation Per user DoF = 1

Zero-Forcing Tx beams

Tx1 W1

Tx2 W2 Rx2

Rx1 Tx1 W1

Tx2 W2 Rx2

Rx1

Backhaul

Veeravalli – ICNC 2/18/15

Transmitter Cooperation – K Users

Per user DoF = 1/2 Interference Alignment

Per user DoF = 1 Zero-Forcing Tx beams

(Broadcast Channel)

•  To achieve PUDoF of 1: o  every message needs to be known at every Tx o  Load on backhaul network increases by a factor of K

•  What happens with partial cooperation?

K-User Interference Channel No Cooperation

K-User Interference Channel Full Cooperation

Veeravalli – ICNC 2/18/15

Cooperative Transmission with Transmit Set Size Constraint

•  Each message is jointly transmitted using at most M transmitters (max backhaul

load factor = M )

•  Message i transmitted

jointly using transmitters in set

•  Consider all message

assignments satisfying cooperation constraint

Ti, |T

i| ≤ M

Veeravalli – ICNC 2/18/15

Backhaul

Tx1 W1

Tx2 W2 Rx2

Rx1

Tx3 W3 Rx3

Cooperative Transmission: Clustering

Clu

ster

1

Clu

ster

2

Per user DoF = 1/2

Achievable w/o any cooperation!

No Degrees of Freedom Gain!

Tx1 W1

Tx2 W2 Rx2

Rx1

Tx3 W3

Tx4 W4 Rx4

Rx3

Veeravalli – ICNC 2/18/15

Cooperative Transmission: Spiral Message Assignments

Wi is available at transmitters {i,i+1,…,i+M-1}

Tx1 W1

Tx2 W2 Rx2

Rx1

TxM WM

TxM+1 WM+1 RxM+1

RxM

Backhaul

Veeravalli – ICNC 2/18/15

Partial Cooperation: Matrix Interpretation

Channel

⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦

[1,1] [1,2] [1,3]

[2,1] [2,2] [2,3]

[3,1] [3,2] [3,3]

H H H

H H H

H H H

Transmit Beams

⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦

[1]

[2]

[3]

V 0 0

0 V 0

0 0 V

Receive Beams

U[1] 0 00 U[2] 00 0 U[3]

⎢⎢⎢⎢

⎥⎥⎥⎥

H[1] [3]1 2[1] [2]2 1

[2] [3]2 1

⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦

V 0 V

V V 0

0 V V

M: # non-zero blocks in the columns of V

Veeravalli – ICNC 2/18/15

Example: K=3, M=2

Each receiver chooses an interference direction Transmitters oblige the receivers

PUDoF of 2/3 with only 3 symbol extensions

Tx1 W1

Tx2 W2 Rx2

Rx1

Tx3 W3 Rx3

Veeravalli – ICNC 2/18/15

Spiral Message Assignment: Results [Annapureddy, El Gamal, VVV – IT’12]

•  DoF with spiral message assignment satisfies:

•  Proof of Achievability:

o  First M-1 users enjoy interference-free communication

o  Interference occupies half signal space at each other

receiver

Generalization of interference alignment scheme

Veeravalli – ICNC 2/18/15

K +M −1

2≤ DoF(K,M ) ≤dK +M −1

2e

Fully Connected IC with Cooperative Transmission: Summary

•  Transmit cooperation constraint M < K

•  Spiral assignment: backhaul load factor = M

•  Interference alignment can be used to achieve DoF gains

•  Symbol extension requirements less stringent

•  As K is increased with M fixed, PUDoF è 1/2

No asymptotic PUDoF gain!

Veeravalli – ICNC 2/18/15

Backhaul

Tx1 W1

Tx2 W2 Rx2

Rx1

Tx3 W3 Rx3

Tx Cooperation in Locally (Partially) Connected Interference Networks

Interference in Cellular Networks

Locally (partially) connected interference channel!

Veeravalli – ICNC 2/18/15

Locally Connected IC Model

Wyner Model: L =1

Tx i is connected to receivers {i, i+1,…, i+L}

Veeravalli – ICNC 2/18/15

Rx1 Tx1

Rx2 Tx2

Rx3 Tx3

Rx4 Tx4

Tx5 Rx5

L = 2

Rx1 Tx1

Rx2 Tx2

Rx3 Tx3

Rx4 Tx4

Tx5 Rx5

Results for Wyner Model [Lapidoth, Shamai, Wigger ‘07]

Rx2 Tx2

Rx3 Tx3

Rx4 Tx4

Tx1 Rx1

Rx5 Tx5

Rx6 Tx6

W2

W3

W4

W1

W5

W6

W2

W4

W1

W5

W1

W4

Backhaul load factor =1

PUDoF (L=1,M=2) = 2/3 > 1/2

Veeravalli – ICNC 2/18/15

Results for Wyner Model [Lapidoth, Shamai, Wigger ‘07]

•  Spiral transmit sets

•  PUDoF (L=1,M) = M/(M+1) Backhaul load factor = M/2

•  Local cooperation can achieve PUDoF gains for locally connected channels

•  Achievable scheme relies on only: o  Zero-forcing transmit beamforming

o  Local CSI

o  Fractional reuse

•  No interference alignment/symbol extensions

Is spiral message assignment optimal?

Veeravalli – ICNC 2/18/15

Example with M=1

Interference-aware message assignment + Fractional reuse

PUDoF(L =1,M =1) = 1

2 PUDoF(L =1,M =1) = 23

Veeravalli – ICNC 2/18/15

Rx1 Tx1

Rx2 Tx2

Rx3 Tx3

W1

W2

W3

Rx1 Tx1

Rx2 Tx2

Rx3 Tx3

W1

W2

W3

Example: Wyner Interference Model

Rx2 Tx2

Rx3 Tx3

Rx4 Tx4

Tx1 Rx1

Rx5 X5

W2

W4

W1

W2

W4

W1

W5

W1

W5

W3

W5

Veeravalli – ICNC 2/18/15

Backhaul load factor =6/5 PUDoF (L=1,M=2) = 4/5 > 2/3

Locally Connected IC with Cooperative Transmission [El Gamal, Annapureddy, VVV, IT ‘14]

•  Result: Under cooperation constraint of M

•  Corollary:

• With interference avoidance constraint:

2M

2M + L≤ PUDoF(L,M ) ≤ 2M + L −1

2M + L

PUDoF(L =1,M ) = 2M

2M +1

PUDoF(L,M ) = 2M

2M + L

Veeravalli – ICNC 2/18/15

DoF Upper Bound: Useful Message Assignments

Veeravalli – ICNC 2/18/15

Rx1 Tx1

Rx2 Tx2

Rx3 Tx3

Rx4 Tx4

W3

Assigning W3 to Tx1 not useful

Cooperative Transmission for IC: Summary

•  Local Cooperation o  no PUDoF gain for fully connected channel

o  is optimal for locally connected channel

•  Interference aware message assignments allow for higher throughput

•  Fractional reuse and zero-forcing transmit beam-forming are sufficient to achieve PUDoF gains, without need for symbol extensions and interference alignment

Veeravalli – ICNC 2/18/15

Tx Cooperation with Backhaul Load Constraint

Backhaul Load Constraint

• More natural cooperation constraint that takes into account overall backhaul load:

Veeravalli – ICNC 2/18/15

|i∈[K ]∑ T

i|

K≤ B

•  Solution under transmit set size constraint can be used to provide solutions under backhaul load constraint

Wyner’s Model with Backhaul Load Constraint

Result:

Veeravalli – ICNC 2/18/15

PUDoF(B) = 4B −1

4B

Rx1 Tx1

Rx2 Tx2

Rx3 Tx3

Recall: PUDoF(M ) = 2M

2M +1

Coding Scheme for B=1

Veeravalli – ICNC 2/18/15

Rx2 Tx2

Rx3 Tx3

Rx4 Tx4

Tx1 Rx1

Rx5 X5

W2

W4

W1

W3

W5

Rx1 Tx1

Rx2 Tx2

Rx3 Tx3

W1

W2

W3

B = 2

3 PUDoF = 2

3

B = 6

5 PUDoF = 4

5

3K

8users

5K

8users

PUDoF (B =1) = 3

4

Application to L-connected network

Veeravalli – ICNC 2/18/15

Result: Using only zero-forcing transmit beamforming and fractional reuse:

Tx i is connected to receivers {i, i+1,…, i+L}

PUDoF(L,B =1) ≥ 1

2,∀L ≤ 6

without need for interference alignment and symbol extensions

Rx1 Tx1

Rx2 Tx2

Rx3 Tx3

Rx4 Tx4

Tx5 Rx5

L = 2

Interference in Cellular Networks

Locally (partially) connected interference channel!

Veeravalli – ICNC 2/18/15

Interference Graph for Single Tier

Veeravalli – ICNC 2/18/15

Tx,Rx pair

Interference Graph without Intrasector Interference

Veeravalli – ICNC 2/18/15

Veeravalli – ICNC 2/18/15

Partition into Noninterfering Tx-Rx pairs

Rx2 Tx2

Rx3 Tx3

Rx4 Tx4

Tx1 Rx1

Rx5 Tx5

Rx6 Tx6

W2

W3

W4

W1

W5

W6

Veeravalli – ICNC 2/18/15

2

3 4 5

6

1

M=6

B = 6 × 6

9= 4; PUDoF = 6

9= 2

3

Rx2 Tx2

Rx3 Tx3

Rx4

Tx1

Rx5

Tx6 Rx6

Tx5

W3

Tx4

W5

W2

W4

W6

Rx1 W1

Veeravalli – ICNC 2/18/15

1

2

3 4 5

6

M=2

B = 6

9= 2

3; PUDoF = 4

9

PUDoF without Intrasector Interference

PUDoF = 7/15 with backhaul load factor B = 1

Veeravalli – ICNC 2/18/15

Cooperation through Backhaul

•  Similar gains in DoF for other cellular interference models, with only zero-forcing and fractional reuse •  Gains improve with asymmetric cooperation and

interference aware message assignment •  Gains in DoF can also be obtained for uplink with

decoded messages being exchanged through backhaul [V. Ntranos, M. Maddah-Ali, G. Caire ‘14]

o  Requires multiple antennas at both mobiles and basestations

o  For same backhaul load factor, gain is smaller than on downlink with Tx cooperation

Veeravalli – ICNC 2/18/15

Summary

•  Infrastructure enhancements in backhaul can be exploited through cooperative transmission to lead to significant rate gains o Minimal or no increase in backhaul load o  Fractional reuse and zero-forcing transmit beam-

forming are sufficient to achieve rate gains o No need for symbol extensions and interference

alignment •  Open Questions:

o  Partial/unknown CSI o Network dynamics and robustness to link erasures o  Joint design with message passing schemes for

uplink

Veeravalli – ICNC 2/18/15

References

•  V. Cadambe and S. A. Jafar, “Interference Alignment and Degrees of Freedom of the K-User Interference Channel,” IEEE Trans. Inf. Theory, vol. 54, no. 8, pp. 3425 –3441, Aug. 2008.

•  V. S. Annapureddy, A. El Gamal, and V. V. Veervalli, “Degrees of Freedom of Interference Channels with CoMP Transmission and Reception,” IEEE Trans. Inf. Theory, vol. 58, no. 9, pp. 5740-5760, Sep. 2012.

•  A. Wyner, “Shannon-Theoretic Approach to a Gaussian Cellular Multiple-Access Channel,” IEEE Trans. Info Theory, vol. 40, no. 5, pp.1713 –1727, Nov. 1994.

•  A. Lapidoth, S. Shamai (Shitz) and M. A. Wigger, “A linear interference network with local Side-Information,” in Proc. IEEE International Symposium on Information Theory (ISIT), Nice, Jun. 2007. Also in IEEE Trans. on Information Theory 2014.

•  A. ElGamal, V.S. Annapureddy, and V.V. Veeravalli. “Interference Channels with CoMP: Degrees of Freedom, Message Assignment, and Fractional Reuse.” IEEE Transactions on Information Theory, 60(6): 3483-3498, June 2014.

•  A. El Gamal, V. V. Veeravalli, ”Flexible backhaul design and degrees of freedom for linear interference networks,” in Proc. IEEE International Symposium on Information Theory (ISIT), pp.2694-2698, Hawaii, June-July 2014.

•  V. Ntranos, M. A. Maddah-Ali, and G. Caire, “Cellular interference alignment,” CoRR, vol. abs/1402.3119, 2014. [Online]. Available: http://arxiv.org/abs/1402.3119

Veeravalli – ICNC 2/18/15