Recent Advances in Full-Duplex Relaying · 2015. 4. 29. · • State-of-the-art devices require two separate antenna arrays: one for receiving and the other for transmitting ⊲

Recent Advances in Full-Duplex Relaying

Taneli RiihonenDepartment of Signal Processing and Acoustics

Center of Excellence in Smart Radios and Wireless Research

Aalto University School of Electrical Engineering, Finland

Session B2, April 24, 2013XXXIII Finnish URSI Convention on Radio Science

Presenter: Taneli Riihonen

Taneli Riihonen Recent Advances in Full-Duplex Relaying – 2 / 56

• Master of Science, Helsinki University of Technology (TKK), Finland, 2006

⊲ Received the McKinsey Award for the best graduating student(only one among all 1007 M.Sc. degrees completed at TKK during that year)

⊲ Currently wrapping up D.Sc. thesis at Aalto University

• Productive (co-)author in scientific publications

⊲ 15/38 published journal/conference papers, some under review

• Dedicated (co-)supervisor for younger students

⊲ 8 M.Sc. theses completed, 1 currently in progress⊲ 2 D.Sc. theses in progress (and collaboration with many others as a co-author)

• Diligent and punctual reviewing service for the community

⊲ Regularly since 2008: so far ∼ 200 papers (∼ 1/1 journals/confs.)⊲ Exemplary Reviewer 2012 for IEEE Communications Letters

• Looking for a postdoc position abroad to grow academically and personally

Agenda


• Overview of the presenter’s work on full-duplex relaying in2008–2011 which constitutes ∼ 1/3 of his upcoming dissertation

• Tutorial to essential aspects that need to be considered whenintroducing full-duplex operation into multihop relaying systems

• The basis for seminal research: loopback self-interference!

⊲ Mitigation techniques and evaluation of their performance⊲ The feasibility of full-duplex relaying in the presence of

residual self-interference, i.e., comparison to half duplex⊲ Merging full duplex with MIMO and OFDM techniques

• The results were originally published in multiple conferenceand journal papers [1]–[12] (see the next two slides)

References (published in 2009)


[1] T. Riihonen, S. Werner, and R. Wichman, “Comparison of full-duplex and half-duplexmodes with a fixed amplify-and-forward relay,” in Proc. IEEE Wireless Communications andNetworking Conference, Apr. 2009.

[2] T. Riihonen, S. Werner, R. Wichman, and J. Hämäläinen, “Outage probabilities ininfrastructure-based single-frequency relay links,” in Proc. IEEE Wireless Communicationsand Networking Conference, Apr. 2009.

[3] T. Riihonen, S. Werner, and R. Wichman, “Optimized gain control for single-frequencyrelaying with loop interference,” IEEE Transactions on Wireless Communications, vol. 8,no. 6, pp. 2801–2806, Jun. 2009.

[4] T. Riihonen, S. Werner, R. Wichman, and E. Zacarias B., “On the feasibility of full-duplexrelaying in the presence of loop interference,” in Proc. 10th IEEE Workshop on SignalProcessing Advances in Wireless Communications, Jun. 2009.

[5] T. Riihonen, K. Haneda, S. Werner, and R. Wichman, “SINR analysis of full-duplex OFDMrepeaters,” in Proc. 20th IEEE International Symposium on Personal, Indoor and MobileRadio Communications, Sep. 2009.

[6] T. Riihonen, S. Werner, and R. Wichman, “Spatial loop interference suppression infull-duplex MIMO relays,” in Proc. 43rd Annual Asilomar Conference on Signals, Systems,and Computers, Nov. 2009.

References (published in 2010–2011)


[7] T. Riihonen, S. Werner, and R. Wichman, “Rate-interference trade-off between duplexmodes in decode-and-forward relaying,” in Proc. 21st IEEE International Symposium onPersonal, Indoor and Mobile Radio Communications, Sep. 2010.

[8] T. Riihonen, S. Werner, and R. Wichman, “Residual self-interference in full-duplex MIMOrelays after null-space projection and cancellation,” in Proc. 44th Annual AsilomarConference on Signals, Systems, and Computers, Nov. 2010.

[9] T. Riihonen, A. Balakrishnan, K. Haneda, S. Wyne, S. Werner, and R. Wichman, “Optimaleigenbeamforming for suppressing self-interference in full-duplex MIMO relays,” in Proc.45th Annual Conference on Information Sciences and Systems, Mar. 2011.

[10] T. Riihonen, S. Werner, and R. Wichman, “Hybrid full-duplex/half-duplex relaying withtransmit power adaptation,” IEEE Transactions on Wireless Communications, vol. 10, no. 9,pp. 3074–3085, Sep. 2011.

[11] T. Riihonen, S. Werner, and R. Wichman, “Transmit power optimization for multiantennadecode-and-forward relays with loopback self-interference from full-duplex operation,” inProc. 45th Annual Asilomar Conference on Signals, Systems, and Computers, Nov. 2011.

[12] T. Riihonen, S. Werner, and R. Wichman, “Mitigation of loopback self-interference infull-duplex MIMO relays,” IEEE Transactions on Signal Processing, vol. 59, no. 12, pp.5983–5993, Dec. 2011.

Introduction


Old Terminology


half/full-duplex link

• Recommendation ITU-R V.662-2 (1993), or Wikipedia:

half duplex — “Designating or pertaining to a method of operation inwhich information can be transmitted in either direction, but notsimultaneously, between two points.”

full duplex — “Designating or pertaining to a mode of operation bywhich information can be transmitted in both directionssimultaneously between two points.”

• Ambiguity problems

⊲ What is the level of abstraction, e.g., considered OSI layer?⊲ May the two directions use different transmission media?⊲ What if communication involves more than two points?

... and even ITU itself characterizes the terms as “deprecated”!

New Terminology


half/full-duplex point

• Herein, we shall adopt the following revised definitions:

half duplex — “Designating or pertaining to a mode of operation bywhich information can be transmitted to and from a point in twodirections, but not simultaneously on the same physical channel.”

full duplex — “Designating or pertaining to a mode of operation bywhich information can be transmitted to and from a point in twodirections simultaneously on the same physical channel.”

• Unambiguous and suitable for discussing modern topics

⊲ Focus on the operation mode of any transceiver instead ofbidirectional communication between exactly two points

⊲ Physical-layer perspective creates a link to spectral efficiency

... and it is not only me who already understands the terms like this

Hot Emerging Topic: Full-Duplex Wireless


• Systems where some node(s) operate in the full-duplex mode• Sometimes descriptively referred to as single-frequency

“simultaneous transmit and receive” (STAR)• Progressive physical/link-layer frequency-reuse concept

= up to double spectral efficiency at a system level, if thesignificant technical problem of self-interference is tackled

• Transmission and reception should use the band for the sameamount of time to make the most of full duplex

⊲ (a)symmetry of traffic pattern, i.e.,requested rates in the two simultaneous directions

⊲ (a)symmetry of channel quality, i.e.,achieved rates in the two simultaneous directions

Full-Duplex Radio Transceivers


Full-duplex transceiver

Full-duplex transceiver Full-duplex transceiver

Full-duplex transceiver

• Basic building blocksfor more complex networks

• The benefits go beyondthe physical layer!

⊲ e.g., simultaneousspectrum sensingand transmission

• Will single-array (or -antenna)full-duplex transceiversbe viable some day?

⊲ Our study is not limitedto the dual-array casealthough it is assumed

Full-Duplex Communication Scenarios


Source Destination

Downlink user Uplink user

Relay

Terminal 1 Terminal 2

Access point

1) Multihop relay link• Symmetric traffic• Asymmetric channels• Direct link may be useful

2) Bidirectional communicationlink between two terminals

• Asymmetric traffic (typically)• Symmetric channels (roughly)

3) Simultaneous down- and uplinkfor two half-duplex users

• Asymmetric traffic• Asymmetric channels• Inter-user interference!

Full-Duplex Relaying


Source DestinationRelay

• Multihop relay link⊲ Symmetric traffic⊲ Asymmetric channels⊲ Direct link may be useful

Agenda• Tutorial to essential aspects that need to be considered when

introducing full-duplex operation into multihop relaying systems• The basis for seminal research: loopback self-interference!

⊲ Mitigation techniques and evaluation of their performance⊲ The feasibility of full-duplex relaying in the presence of

residual self-interference, i.e., comparison to half duplex⊲ Merging full duplex with MIMO and OFDM techniques

Full-Duplex Relaying


Relaying


Source

Relay

Destination

• The general purpose of a relay node is to forward signalsfrom a source transmitter to a destination receiver

⊲ Other network topologies are also possible,e.g., with multiple hops or parallel relays

⊲ Common protocols: amplify-and-forward (AF),decode-and-forward (DF)

• Full-duplex relays exploit STAR such that source–relay andrelay–destination links share one physical channel

⊲ can be more sophisticated than simple on-channel repeaters

Direct Link


Source

Relay

DestinationPotential direct link

• Two different applications for relays:a) coverage extension where the relay is deployed

because the direct link is weakb) diversity improvement where transmission from both the relay

and the source is strong (on average) at the destination• The former is more potential application for full-duplex relays

⊲ Half-duplex relaying can offer maximum diversity gain⊲ Rate/SNR gain of full-duplex relaying becomes marginal

with a strong direct link: simple switching works well

Inherent Symmetry: Advantage for Full Duplex


Source

Full-duplex relay

Destination Source

Half-duplex relay

Destinationor

• Full duplex can ideally render up to double spectral efficiencywhen compared to conventional half-duplex operation

⊲ Largest gains are achieved when simultaneous transmissionsoccupy the channel for the same amount of time

• Relay links are good candidates for adopting the full-duplex modebecause their traffic pattern is inherently symmetric:

⊲ Equal requested source–relay and relay–destination data ratesto avoid data overflow or underflow in the relay

⊲ Unequal achieved data rates due to channel imbalance

Mitigation of Loopback Self-interference


Mitigation of Loopback Self-interference (Refs)


• The following discussion mainly originates from

[12] T. Riihonen, S. Werner, and R. Wichman, “Mitigation of loopbackself-interference in full-duplex MIMO relays,” IEEE Transactions onSignal Processing, vol. 59, no. 12, pp. 5983–5993, Dec. 2011.

• Related results are available also in conference papers:

[6], [8], [9], [11]• Measurement data on prototype antenna arrays by courtesy of

colleagues from Department of Radio Science and Engineering:

[H+] K. Haneda, E. Kahra, S. Wyne, C. Icheln, and P. Vainikainen,“Measurement of loop-back interference channels foroutdoor-to-indoor full-duplex radio relays,” in Proc. 4th EuropeanConference on Antennas and Propagation, Apr. 2010.

Loopback Self-interference


Full-duplex relay

Loopback self-interference

• Full-duplex operation is possible only after tackling a significanttechnical challenge: unavoidable self-interference

⊲ Huge difference in power levels (interference vs. desired signal)• Full duplex is adopted first for fixed infrastructure nodes and

later (maybe) for small portable, or even handheld, radios⊲ Initially, the concept of full-duplex relaying is different from

cooperative communication among mobile nodes wheretime-division half-duplex operation is the baseline assumption

• Next: self-interference mitigation techniques

Passive Physical Isolation


Full-duplex relay

Loopback self-interference

• State-of-the-art devices require two separate antenna arrays:one for receiving and the other for transmitting

⊲ Mainly antenna design and placement problems:directivity, back-to-back coupling, distance, obstacles

⊲ But using two arrays is useful for relaying in general since thesource and the destination are located at different directions

• In (future?) single-array devices, all physical isolation is providedby a circulator: mainly an electronics design problem

• Next: measured physical isolation with prototype antenna arrays

Experimental Antenna Arraysfor Full-Duplex MIMO Relay ∗


• Design goals:

1. Compact size but high isolation2. 2.6GHz ± 100MHz operation band3. Multiple Rx and Tx antenna elements

• Building and measuring 4× 4 array prototype∗Further details are provided in [H+]:

K. Haneda et al., “Measurement of loop-back interference channels for outdoor-to-indoor full-duplex radio relays,” April 2010.

Channel Measurement Campaignfor Outdoor-to-Indoor Relaying Scenarios ∗


Compact array configuration• Arrays attached side-by-side (2cm)• Small box like a Wi-Fi router• Several positions next to windows

Separate array configuration• Four Tx antenna orientations• LOS: Tx in the same room as Rx• NLOS: Tx in the adjacent corridor

∗Further details are provided in [H+]:K. Haneda et al., “Measurement of loop-back interference channels for outdoor-to-indoor full-duplex radio relays,” April 2010.

Average Physical Isolation


• For compact array configuration,measured isolation is 36.2dB

• For separate array configuration,isolation is directly proportional toantenna separation (2–3dB/m)

1

2

3 4dRR

Rrx Rtx

0 2 4 6 8 10 1250

55

60

65

70

75

80

85

Orientation 1Orientation 2Orientation 3Orientation 4

line-o

f-sigh

t (LO

S)

non-l

ine-of

-sigh

t (NL

OS)

E{P

tx/P

I}[d

B]

dRR [m]

• 20dB isolation from window glass for separate array configuration• Mere physical isolation may not be sufficient which gives

motivation for active mitigation by signal processing

Objective for Active Mitigation


−20 −10 0 10 20 30 40 50 60 70 80

0.01

0.1

0.5

(ǫH,ǫt)

Natural isolationTDC (0.02,0.02)NSP (0.02,0.02)TDC (0,0.02)NSP (0,0.02)TDC (0.02,0)NSP (0.02,0)Half duplex

BE

R

E{PI∣

∣

natural} [dB]

PI∣

∣

man-made

∆PI

The target use case for full-duplex relaying with mitigation

bit-error rate in a DF relay vs. physical isolation

• Transparent minimization of self-interference: the relay protocolcan operate as in the half-duplex mode but at double symbol rate

⊲ Mitigation becomes separated from the protocol designand the schemes are applicable with all kinds of protocols

Active Mitigation


Time-domain cancellation:

R

filter

Spatial-domain suppression:

R filterfilter

• Two main techniques for active self-interference mitigation⊲ Cancellation : time-domain filtering in feedback path⊲ Suppression : spatial-domain filtering in feedforward path

• Both schemes could ideally eliminate all self-interference• Cancellation is a rather straightforward task while

suppression can be implemented in various ways

Imperfect Side Information


R

filternoise

error

• In practice, self-interference cannot be perfectly eliminated⊲ Channel estimation error in filter design⊲ Transmit-side noise due to non-ideal electronics

(the actual transmitted signal is not known)• Sufficient physical isolation and analog pre-cancellation are also

required to cope with limited dynamic range at the receive side

Spatial-Domain Suppression


R filterfilter

noiseerrorerror

• Next: evaluating the main variations of suppression⊲ antenna selection (AS)⊲ beam selection (BS)⊲ null-space projection (NSP)⊲ minimum mean square error (MMSE) filtering

• In some cases, it may be beneficial to combine time-domaincancellation with spatial-domain suppression

Antenna vs. Beam Selection


0 2 4 6 8 10 12 14 160.01

0.1

1

N̂rx × N̂tx: E{∆PI} [dB]

3 × 4: 0.942 × 4: 1.883 × 3: 2.201 × 4: 3.252 × 3: 3.632 × 2: 5.751 × 3: 6.181 × 2: 10.051 × 1: 22.63

F∆

PI(x)

x [dB]

antenna selection (AS)

0 2 4 6 8 10 12 14 160.01

0.1

1

N̂rx × N̂tx: E{∆PI} [dB]

3 × 4: 3.062 × 4: 7.373 × 3: 7.861 × 4: 21.812 × 3: 23.57

F∆

PI(x)

x [dB]

beam selection (BS)

• Ideal side information; four receive and transmit antennas• AS improves isolation significantly only in the single-stream case

⊲ BS is reduced to null-space projection (NSP) and eliminatesself-interference completely if less than five streams are used

Rank of Loopback Channel


0 2 4 6 8 10 12 14 160.01

0.1

1

Nrx × Ntx (rk{HLI}): E{∆PI} [dB]

AS, 3 × 4 (3): 1.29AS, 3 × 4 (2): 1.62AS, 4 × 4 (4): 2.20AS, 4 × 4 (3): 2.57BS, 3 × 4 (3): 5.04BS, 3 × 4 (2): 8.12BS, 4 × 4 (4): 7.86BS, 4 × 4 (3): 11.87

F∆

PI(x)

x [dB]

ideal side information; three receive and transmit antennas

• Spatial-domain suppression can benefit from low channel rank

⊲ Beam selection (BS) directs the self-interference energyto the weakest eigenmodes which include the null space

• Time-domain cancellation (not shown) would not be affected at all

Imperfect Side Information


0.01 0.1 1 100

5

10

15

20

25

30

35

40

45

N̂rx × N̂tx (rk{HLI})

NSP, 3 × 4 (1)

BS, 3 × 4 (2)BS, 3 × 4 (3)BS, 3 × 4 (4)

NSP, 3 × 3 (1)

NSP, 3 × 3 (2)

BS, 3 × 3 (3)

BS, 3 × 3 (4)

NSP, 2 × 4 (1)

NSP, 2 × 4 (2)

BS, 2 × 4 (3)

BS, 2 × 4 (4)

TDC, 4 × 4 (1–4)

ǫH

E{∆

PI}

[dB

]

channel estimation error

0.01 0.1 1 10

0

5

10

15

20

25

30

35

40

Nrx × Ntx (rk{HLI})

NSP, 3 × 4 (1)

BS, 3 × 4 (2)

BS, 3 × 4 (3)

NSP, 4 × 3 (1)

BS, 4 × 3 (2)

BS, 4 × 3 (3)

NSP, 4 × 4 (2)

BS, 4 × 4 (3)

BS, 4 × 4 (4)

TDC, 3 × 3 (1–3)

ǫt

E{∆

PI}

[dB

]

transmit-side noise

• Additional isolation from BS is limited with ideal side information

⊲ Imperfect side information determines the additional isolationachieved with NSP or time-domain cancellation (TDC)

• NSP can be made immune to transmit-side noise

Cancellation vs. Suppression


0 8 16 24 32 40 48 56 640.01

0.1

1

(rk{HLI}): E{∆PI} [dB]

BS, (3): 5.04BS, (2): 8.15MMSE, (3): 8.77MMSE, (2): 14.76TDC, (1): 25.20TDC, (2): 25.30TDC, (3): 25.34MMSE, (1): 40.20NSP, (1): 40.52

F∆

PI(x)

x [dB]

minimum MSE filtering

0 8 16 24 32 40 48 56 640.01

0.1

1

(rk{HLI}): E{∆PI} [dB]

BS, (4): 7.80BS, (3): 11.56TDC, (1): 25.25TDC, (2): 25.32TDC, (3): 25.35TDC, (4): 25.36NSP, (2): 29.67both, (4): 34.49both, (3): 34.49both, (2): 34.49both, (1): 36.81NSP, (1): 40.51

F∆

PI(x)

x [dB]

combined time/spatial-domain mitigation

• Loopback channel rank defines which scheme is preferable• The combination of TDC and suppression offers better

performance than either alone, except when rank-deficientloopback channel enables the usage of NSP

Transmit Power Adaptation


Transmit Power Adaptation (Refs)



[3] T. Riihonen, S. Werner, and R. Wichman, “Optimized gain controlfor single-frequency relaying with loop interference,” IEEETransactions on Wireless Communications, vol. 8, no. 6,pp. 2801–2806, Jun. 2009.

[10] T. Riihonen, S. Werner, and R. Wichman, “Hybridfull-duplex/half-duplex relaying with transmit power adaptation,”IEEE Transactions on Wireless Communications, vol. 10, no. 9,pp. 3074–3085, Sep. 2011.


[4], [5], [7], [11]

Transmit Power Adaptation


Source

Full-duplex relay

Residual leakage

Destination

power control

• In practice, there will always be residual self-interferenceafter applying all means of mitigation

• Fortunately, transmit power adaptation can still exploitthe channel imbalance caused by residual interference

⊲ In principle, the relay should appropriately lower its owntransmit power if the first hop is the bottleneck of the system

• Win–win solution: energy savings can be achievedwhile performance is also optimized

Example with Amplify-and-Forward Protocol


0 1 2 3 4 5 6 7 8 9 100

5

10

15

20

25

β2 [linear]

γ[li

near

]

Varying β2, fixed |hLI|2

With β2opt, varying |hLI|2

With β2tar, varying |hLI|2

With β2max, varying |hLI|2

|hLI|2

= 0

|hLI|2 ≈ −29.86dB

|hLI|2

= −25dB

|hLI|2

= −20dB

|hLI|2

= −15dB

|hLI|2

= −10dB

end-to-end SINR vs. relay gain

• The end-to-end signal-to-interference and noise ratio (SINR) startsto decrease when increasing relay gain beyond the optimal point

⊲ Relay should use its maximum allowed transmit poweronly in the case of low residual self-interference

Full Duplex vs. Half Duplex


Full Duplex vs. Half Duplex (Refs)



[10] T. Riihonen, S. Werner, and R. Wichman, “Hybridfull-duplex/half-duplex relaying with transmit power adaptation,”IEEE Transactions on Wireless Communications, vol. 10, no. 9,pp. 3074–3085, Sep. 2011.


[1], [4], [7], [11]

• In articles [2] and [3], our results focus on the full-duplex mode,but the analysis itself could be also used for comparison purposes

Fundamental Rate–Interference Trade-off


• Determining the ultimate feasibility of full-duplex relayingin the presence of residual self-interference. In principle,

⊲ half-duplex relay link: ⊲ full-duplex relay link:

Reduced symbol rate due to two allocated channels Residual self-interference even after mitigation

RHD =1

2log2

(

1 +PS

PN

)

RFD = log2

(

1 +PS

PI+PN

)

• Should the system choose to operate with

a) loss of end-to-end symbol rate (half duplex), orb) loss of S(I)NR due to self-interference (full duplex)?

Full- or Half-Duplex (... or Direct Transmission)?


• Rate–interference trade-off: choosing between

⊲ full-duplex (FD) relaying with residual self-interference− Direct link treated as interference at the destination− With and without transmit power adaptation

⊲ half-duplex (HD) relaying− Maximum ratio combining (MRC) for the direct

and relayed transmissions at the destination⊲ direct transmission (DT)

− The same (full) symbol rate as with FD relayingbut low channel SNR on average (coverage extension)

• The comparison yields switching boundaries betweenthe modes according to channel imbalance

Instantaneous Channel State Information


Let us next consider the case of deterministic (static) channels

• This represents, for example, a snapshot of the system withinchannel coherence time in a slow-fading environment

• Instantaneous channel state information (channel SNRs) for

⊲ choosing the proper mode⊲ transmit power adaptation (with FD)⊲ maximum ratio combining (with HD)

• Metric for the comparison: instantaneous transmission rate

⊲ The analysis can be completely conducted in terms ofclosed-form expressions (see the papers)

Instantaneous Switching Boundaries


−15 −10 −5 0 5 10 15 20 250

0.5

1

1.5

2

2.5

ΓLI

ΓSD

γLI = γSD + 15 [dB]

C[b

it/s/

Hz]

pR = 1

pR = p∗

R

Full-duplex

Half-duplex with MRC

Decode-and-forwardAmplify-and-forwardDirect transmission

instantaneous transmission rates

• Full-duplex (FD) relaying is preferred with low self-interference

⊲ Transmit power adaptation extends the range further

• Pure direct transmission (DT) is preferred with a strong direct linkand MRC gives little benefit for half-duplex (HD) relaying

Direct Transmission vs. Relaying


−10 −5 0 5 10 15 20−10

−5

0

5

10

15

20

γSR [dB]

γR

D[d

B]

γSD = −6dB

γSD = −1dB

γSD = 3dB

γSD = 6dB

DT

FD/HD

Decode-and-forwardAmplify-and-forward

switching boundaries

• FD relaying is suitable for the scenario of coverage extension

⊲ When the direct link exists in fortunate fading states,the relay is not momentarily needed at all

• Simple switching yields also good diversity improvement

Full-Duplex vs. Half-Duplex Relaying


−10 −5 0 5 10 15 20−10

−5

0

5

10

15

20

γSR [dB]

γR

D[d

B]

γLI = 3dB γLI = 6dB γLI = 9dBHD

FDDecode-and-forward Amplify-and-forward

without transmit power adaptation

−10 −5 0 5 10 15 20−10

−5

0

5

10

15

20

γSR [dB]

γR

D[d

B]

γLI = 6dB

γLI = 9dB

γLI = 12dB

γLI= 9

dB

γLI=

12dB

γLI=

15dB

HD

FD

Decode-and-forward Amplify-and-forward

p∗R = 1 for AF

with transmit power adaptation

• Instead of adhering to any mode at early design stage, it isadvantageous to implement hybrid full-duplex/half-duplex relaying,i.e., opportunistic switching between the modes, because therate–interference trade-off favors them alternately during operation

Statistical Channel State Information


Let us then consider the case of fading channels

• Fixed infrastructure relay node for coverage extension

⊲ Static link between the base station and the relay⊲ Rayleigh-fading link between the relay and a mobile user

• Statistical channel state information (average channel SNRs) for

⊲ choosing the proper mode⊲ transmit power adaptation (with FD)

• Metric for the comparison: average transmission rate

⊲ The actual rate expressions can be calculated in a closed formbut switching boundaries and transmit power adaptation neednumerical look-up tables (see the papers)

Statistical Switching Boundaries


0 5 10 15 20 250.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

Decode-and-forward

Amplify-and-forward

γ̄RD [dB]

C̄[b

it/s/

Hz]

FD, pR = p∗

RFD, pR = p̄

∗

RHybrid FD/HDFD, pR = 1HD

FD, pR = p∗

RFD, pR = p̄

∗

RHybrid FD/HDFD, pR = 1HD

average transmission rates

• Statistical mode switching and transmit power adaptation yieldrather good performance with much lower signaling overhead

⊲ Hybrid FD/HD relaying (instantaneous switching) givesthe largest gains near statistical switching boundaries

Full-Duplex vs. Half-Duplex Relaying


0 5 10 15 20 25 300

5

10

15

20

25

30

γ̄SR [dB]

γ̄R

D[d

B]

C̄FD = C̄HD 25%25% 50%50%

75%

AF in DLDF in DLAF in ULDF in UL

without transmit power adaptation

0 5 10 15 20 25 300

5

10

15

20

25

30

γ̄SR [dB]

γ̄R

D[d

B]

C̄FD = C̄HD

35%

60%

75%

AF in DLDF in DLAF in ULDF in UL

with transmit power adaptation

• Illustrating downlink (DL) vs. uplink (UL) transmission

⊲ self-interference in a mobile channel vs. in a fixed channel

• Rate is significantly improved by choosing the proper modewhich is typically FD when using transmit power adaptation

Conclusion


Conclusion


• Wireless full duplex: A progressive frequency-reuse concept!• Herein: overview of recent work on full-duplex relaying• Essential aspects that need to be considered when introducing

full-duplex operation into multihop relaying systems

⊲ Loopback self-interference⊲ Mitigation techniques and evaluation of their performance

− physical isolation− time-domain cancellation− spatial-domain suppression− transmit power adaptation

⊲ Rate–interference tradeoff: the feasibility of full-duplexrelaying in the presence of residual self-interference

• ... and how is all this related to OFDM mentioned in the beginning?

Future Work


Joint Signal and Interference Processing


Joint processing

Joint processing

• Herein: “transparent” self-interference mitigation schemes⊲ Any existing relaying protocol could be used⊲ But the joint design of mitigation and a specific protocol

would probably bring performance gains• Herein: simple switching between direct transmission and relaying

⊲ Direct link is regarded as interference when using the relay⊲ The destination could apply signal processing techniques

to separate and constructively combine the superimposedsignals from the source and the relay

Extensions to Other Full-Duplex Scenarios


Source Destination

Downlink user Uplink user

Relay

Terminal 1 Terminal 2

Access point

Full-duplex communication1) Multihop relay link2) Bidirectional communication3) Simultaneous down- and uplink

Other potential uses for STAR• medium access control• cognitive radios

Generic full-duplex radios• improved isolation and mitigation

Full-duplex transceivers

Limited Receiver Dynamic Range


A/D D/AR

filter

filter

• Severe risk of saturating analog-to-digital (A/D) converters⊲ quantization noise due to limited resolution⊲ clipping noise which is pronounced with OFDM

• Digital cancellation is useless if dynamic range is not sufficient• It is difficult and expensive to adapt the response of an analog

filter to match the time- and frequency-selective MIMO channel

Example on Quantization Noise ( 4-bit A/D)


Signal of interestInterference signalSum signal

• ∼1-bit resolutionfor the signal of interest

before A/D

after A/D

after digitalcancellation

andscaling


Example on Clipping Noise ( 4-bit A/D)


Signal of interestInterference signalSum signal

• ∼2-bit clipped resolutionfor the signal of interest

before A/D

after A/D

after digitalcancellation

andscaling


Mitigation in Analog Domain


A/D D/AR filter

filter

noise

• Self-interference should be minimized before A/D conversion⊲ Physical isolation is an antenna design problem⊲ Analog cancellation is an electronics design problem

• Transmit-side beamforming can eliminate the interference“on-the-air” before it even reaches the receiver front-end

⊲ A digital signal processing problem!


Presenter: Taneli RiihonenAgendaReferences (published in 2009)References (published in 2010–2011)IntroductionOld TerminologyNew TerminologyHot Emerging Topic: Full-Duplex WirelessFull-Duplex Radio TransceiversFull-Duplex Communication ScenariosFull-Duplex Relaying

Full-Duplex RelayingRelayingDirect LinkInherent Symmetry: Advantage for Full Duplex

Mitigation of Loopback Self-interferenceMitigation of Loopback Self-interference (Refs)Loopback Self-interferencePassive Physical IsolationExperimental Antenna Arraysfor Full-Duplex MIMO Relay*Channel Measurement Campaignfor Outdoor-to-Indoor Relaying Scenarios*Average Physical IsolationObjective for Active MitigationActive MitigationImperfect Side InformationSpatial-Domain SuppressionAntenna vs. Beam SelectionRank of Loopback ChannelImperfect Side InformationCancellation vs. Suppression

Transmit Power AdaptationTransmit Power Adaptation (Refs)Transmit Power AdaptationExample with Amplify-and-Forward Protocol

Full Duplex vs. Half DuplexFull Duplex vs. Half Duplex (Refs)Fundamental Rate–Interference Trade-offFull- or Half-Duplex (... or Direct Transmission)?Instantaneous Channel State InformationInstantaneous Switching BoundariesDirect Transmission vs. RelayingFull-Duplex vs. Half-Duplex RelayingStatistical Channel State InformationStatistical Switching BoundariesFull-Duplex vs. Half-Duplex Relaying

ConclusionConclusion

Future WorkJoint Signal and Interference ProcessingExtensions to Other Full-Duplex ScenariosLimited Receiver Dynamic RangeExample on Quantization Noise (4-bit A/D)Example on Clipping Noise (4-bit A/D)Mitigation in Analog Domain

Documents

Recent Advances in Full-Duplex Relaying · 2015. 4. 29. · • State-of-the-art devices require two separate antenna arrays: one for receiving and the other for transmitting ⊲