In-band Full-Duplex in Hand-held Applications: Analysis of canceller tuning requirements
1 like458 views
This document presents an analysis of the tuning requirements for an in-band full-duplex communication system in hand-held applications, focusing on self-interference cancellation. It describes various technologies such as electrical balance duplexers and the dynamic behavior of antennas to maintain isolation during user interactions. The findings indicate that achieving sufficient self-interference cancellation is challenging, especially with user movements, highlighting the potential impracticality of in-band full-duplex systems in hand-held scenarios.
![Communication Systems & Networks Group
University of Bristol
Background: Spectral Issues 3
24.3 Exabytes per month of mobile data traffic by 2019 [1]
10 Times
Mobile Data Traffic ↑
Mobile Connections in 5G… 2020 ↑
V2V
Useable Frequency Allocations
Spectrum Shortage
Explosive Requirements
?
United States’ Radio Spectrum Frequency Allocations [3]
China Mobile 2014 [2]: 39.4 million net additional customers.
This figure should goes up, 720,000 4G BS put in use in 2014.
Smart
City
[1] Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2014–2019.
[2] “China Mobile’s annual report 2014”. [Online]: www.chinamobileltd.com/en/ir/reports/ar2014.pdf
[3] [Online] http://www.ntia.doc.gov/files/ntia/publications/spectrum_wall_chart_aug2011.pdf](https://image.slidesharecdn.com/leolaughlinsenseworkshop27-06-18-180628073447/85/In-band-Full-Duplex-in-Hand-held-Applications-Analysis-of-canceller-tuning-requirements-3-320.jpg)
In-band Full-Duplex in Hand-held Applications: Analysis of canceller tuning requirements
- 1. In-band Full-duplex in Hand-held Applications: Analysis of canceller tuning requirements Leo Laughlin, Chunqing Zhang, Mark Beach and Kevin Morris Communication and Networks Group, University of Bristol, Bristol. UK http://www.bristol.ac.uk/engineering/research/csn/ IEEE WIAD: SENSE Workshop. KCL, London. 27th June 2018
- 2. Communication Systems & Networks Group University of Bristol Outline 2 ➢ Background ➢ Electrical Balance Duplexer ➢ User interaction antenna measurements ➢ Simulated adaptive self-interference canceller ➢ Results ➢ Conclusions
- 3. Communication Systems & Networks Group University of Bristol Background: Spectral Issues 3 24.3 Exabytes per month of mobile data traffic by 2019 [1] 10 Times Mobile Data Traffic ↑ Mobile Connections in 5G… 2020 ↑ V2V Useable Frequency Allocations Spectrum Shortage Explosive Requirements ? United States’ Radio Spectrum Frequency Allocations [3] China Mobile 2014 [2]: 39.4 million net additional customers. This figure should goes up, 720,000 4G BS put in use in 2014. Smart City [1] Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2014–2019. [2] “China Mobile’s annual report 2014”. [Online]: www.chinamobileltd.com/en/ir/reports/ar2014.pdf [3] [Online] http://www.ntia.doc.gov/files/ntia/publications/spectrum_wall_chart_aug2011.pdf
- 4. Communication Systems & Networks Group University of Bristol Bi-directional “Duplex” Communication 4 >20dBm <-90dBm Transmitter Receiver Antenna Self Interference Current systems: • Time Division Duplexing (TDD) • Frequency Division Duplexing (FDD)
- 5. Communication Systems & Networks Group University of Bristol Time Division Duplexing 5 Fc Frequency Time
- 6. Communication Systems & Networks Group University of Bristol Frequency Division Duplexing 6 Frequency Time F2 F1
- 7. Communication Systems & Networks Group University of Bristol In-Band Full-Duplex 7 Frequency Time Fc Transmitting and receiving on the same frequency at the same time
- 8. Communication Systems & Networks Group University of Bristol Full Duplex – New Approach? 8 Single Frequency Tactical Manpack, 1980 Chris Richardson (Roke), UK patent PhD Student: S Chen, Bristol 1997 ‘Duplex Free’ S. Chen, M. Beach, and J. McGeehan, “Division- free duplex for wireless applications," Electronics Letters, vol. 34, no. 2, pp. 147-148, jan 1998.
- 9. Communication Systems & Networks Group University of Bristol In-Band Full-Duplex Research 9 Rice University Stanford University Tx chain Tx Signal Rx Signal A τ Rx chain Yonsei University Tx chain Tx Signal Rx Signal A τ Rx chain Tx chain Tx Signal Rx Signal A τ Rx chain EBD DUPLO Project University of Bristol Tx chain Tx Signal Rx Signal Rx chain EBDTx chain Cancellation Signal
- 10. Communication Systems & Networks Group University of Bristol Electrical Balance Duplexer 10 PA LNATx Rx ZBAL 0ᵒ0ᵒ 0ᵒ 180ᵒ ZANT High isolation achieved when ZBAL = ZANT Self Interference Cancellation
- 11. Communication Systems & Networks Group University of Bristol EBD isolation – frequency domain. 11 Limited isolation bandwidth due to frequency domain antenna impedance variation.
- 12. Communication Systems & Networks Group University of Bristol EBD isolation – time domain. ➢ Antenna impedance is time variant due to environmental interaction. ➢ Balancing impedance must track antenna impedance to maintain isolation. ➢ This process is performed by a balancing algorithm. EBD adaptation requirements have been investigated by measuring dynamic antenna reflection coefficient and embedding this in circuit simulations Measure antenna S11 Simulated duplexer Antenna S11 data
- 13. Communication Systems & Networks Group University of Bristol Antenna measurements ➢ Antenna is enclosed in mobile phone housing ➢ Antenna S11 measured every 1.25 ms ➢ “User” is making texting and browsing motions to emulate hand-held scenario
- 14. Communication Systems & Networks Group University of Bristol EBD isolation – time domain. ➢ EBD simulation uses measured antenna and simulated balancing network to calculated Tx-Rx isolation 𝐺 ω, 𝑡 = 1 4 ( Γ𝑎𝑛𝑡𝑒𝑛𝑛𝑎 ω, 𝑡 − Γ𝑏𝑎𝑙𝑎𝑛𝑐𝑒 ω, 𝑡 ) Measured antenna S11 Calculated S11 of Ideal tunable RC balancing circuit Calculated Tx-Rx gain
- 15. Communication Systems & Networks Group University of Bristol Simulated adaptive balancing behaviour 15 ➢ Isolation depends on tracking behaviour of balancing circuit ➢ Three adaptation behaviours are shown here ➢ Ideal balancing. The optimal balancing impedance is recalculated for each antenna measurement. ➢ Static balancing. Balancing impedance is calculated based on the first antenna measurement and then not updated. ➢ Idealized Adaptation (LRA). Balancing impedance is updated at a given interval. This is equivalent to running an ideal balancing algorithm ➢ this work the simulated EBD is using a 5 ms adaptation interval ➢ This is sufficient to maintain isolation >50 dB in the user interaction scenario
- 16. Communication Systems & Networks Group University of Bristol Simulated EBD with digital cancellation 16 Simulated EBD with measured antenna dynamics Adaptive digital canceller with different adaptation behaviours Tx signal Residual SI • Dynamic SI channel through EBD • Digital canceller must adapt to maintain isolation • Residual SI is observed for difference digital canceller adaptation behaviours
- 17. Communication Systems & Networks Group University of Bristol Simulated EBD with digital cancellation: Idealised digital cancellation 17 ➢ Updates digital canceller coefficients to optimum values at a given interval.
- 18. Communication Systems & Networks Group University of Bristol Simulated EBD with digital cancellation: Least means squared (LMS) algorithm 18 ➢ Runs the LMS algorithm at a given rate. ➢ Noise floor is at -100 dB for both types of cancellation
- 19. Communication Systems & Networks Group University of Bristol Results: EBD with Ideal Digital Canceller 19 ➢ EBD is re-balancing at 5ms intervals ➢ EBD isolation varies across 39-53 dB (for this particular antenna measurement) ➢ Ideal digital canceller shows large degradation in cancellation performance between coefficient updates. ➢ A very fast update cycle of 50 us is required to achieve ~80 dB isolation. ➢ This is a substantial processing overhead, may not be feasible
- 20. Communication Systems & Networks Group University of Bristol Results: EBD with Ideal Digital Canceller 20 ➢ CDF of isolation for different coefficient update rates gives insight into adaptation requirements ➢ A very fast update cycle of 50 us is required to achieve >80 dB isolation for 98% of the time. ➢ This is a substantial processing overhead, may not be feasible. ➢ Achieving higher isolation would require even faster adaptation.
- 21. Communication Systems & Networks Group University of Bristol Results: EBD with LMS Digital Canceller 21 ➢ EBD rebalancing causes a step change in the SI channel. ➢ This causes a large spike in SI, and the LMS algorithm takes time to reconverge. 50 ns step interval
- 22. Communication Systems & Networks Group University of Bristol Results: EBD with LMS Digital Canceller 22 50 ns step interval 25 us step interval
- 23. Communication Systems & Networks Group University of Bristol Results: EBD with LMS Digital Canceller 23 ➢ 1 us LMS step interval is required to achieve >80 dB isolation. ➢ Also a substantial processing overhead.
- 24. Communication Systems & Networks Group University of Bristol Conclusions 24 ➢ The effect of the users hand movements on a handheld device causes substantial variations in the self-interference channel. ➢ This imposes stringent requirements on the adaptation of digital canceller coefficients. ➢ In-Band Full-duplex may not be realistic in this scenario – however, these devices are not always hand-held, and the users hand may also be static for much of the time.
- 25. Communication Systems & Networks Group University of Bristol Thanks & Questions? 25