PaperlayerView exact revision on arXiv

Source-linked AI summary

Cancellation of Power Amplifier Induced Nonlinear Self-Interference in Full-Duplex Transceivers

Lauri Anttila, Dani Korpi, Ville Syrjälä, Mikko Valkama

arXiv:1401.3129v1cs.ITcs.NImath.IT

TL;DR

Strong transmitter self-interference and unmodeled front-end nonlinearities complicate full-duplex reception. The paper models the nonlinear SI path with a parallel Hammerstein structure and cancels its regenerated output digitally. Simulations show higher usable transmit power than linear cancellation, or the use of a lower-IIP3 PA.

  • Problem

    Prior SI cancellation studies largely did not account for transmitter and receiver front-end nonlinearities, while PA-induced nonlinear SI can become a bottleneck at higher transmit powers.

    Source: §I · p. 1, §II · p. 2

  • Method

    The method models the combined PA, multipath SI channel, and RF canceller as a parallel Hammerstein structure, then estimates its parameters to regenerate and subtract SI digitally.

    Source: §I · p. 1, A. Baseband nonlinear… · p. 3

  • Results

    The nonlinear canceller allows approximately 10 dB higher transmit power than linear SI cancellation regardless of antenna separation, and up to 15 dB reduction in IIP3 compared with linear cancellation.

    Source: §IV · pp. 4, 6

  • Takeaways & Limitations

    The simulations indicate that nonlinear digital cancellation can extend the usable transmit-power range or permit a lower-quality PA, subject to PA linearity requirements set by spectrum emission standards.

    Source: §IV · p. 6, §V · p. 6

    At very high transmit powers, achievable SI suppression starts decreasing due to quantization noise and model mismatch between the Wiener PA and parallel Hammerstein estimation model.

    Source: §IV · p. 5

Abstract

from arXiv · show

Recently, full-duplex (FD) communications with simultaneous transmission and reception on the same channel has been proposed. The FD receiver, however, suffers from inevitable self-interference (SI) from the much more powerful transmit signal. Analogue radio-frequency (RF) and baseband, as well as digital baseband, cancellation techniques have been proposed for suppressing the SI, but so far most of the studies have failed to take into account the inherent nonlinearities of the transmitter and receiver front-ends. To fill this gap, this article proposes a novel digital nonlinear interference cancellation technique to mitigate the power amplifier (PA) induced nonlinear SI in a FD transceiver. The technique is based on modeling the nonlinear SI channel, which is comprised of the nonlinear PA, the linear multipath SI channel, and the RF SI canceller, with a parallel Hammerstein nonlinearity. Stemming from the modeling, and appropriate parameter estimation, the known transmit data is then processed with the developed nonlinear parallel Hammerstein structure and suppressed from the receiver path at digital baseband. The results illustrate that with a given IIP3 figure for the PA, the proposed technique enables higher transmit power to be used compared to existing linear SI cancellation methods. Alternatively, for a given maximum transmit power level, a lower-quality PA (i.e., lower IIP3) can be used.

I. INTRODUCTION

Full-duplex communication sends and receives on the same carrier, but strong transmitter self-interference can overwhelm the received signal. This article proposes nonlinear digital cancellation for power-amplifier-induced interference and reports improved operating range over linear cancellation.

  • Earlier FD demonstrations used RF and digital baseband cancellation, with transmit powers in the local-area range of ≤20 dBm –[7].Source: §I · p. 1
  • PA nonlinearity can make self-interference a bottleneck at higher transmit powers, potentially exceeding the weak received signal and preventing successful detection.Source: §I · p. 1
  • The proposed nonlinear digital canceller models and mitigates PA-induced SI; simulations report up to 10 dB higher transmit power than traditional linear cancellation without degrading receiver SINR, or use of a lower-IIP3 PA.Source: §I · p. 1

II. PA NONLINEARITY CHALLENGE IN FD TRANSCEIVER

The section examines why PA nonlinearity limits full-duplex operation despite passive, analog, and digital SI cancellation. Under the stated system-calculation assumptions, PA-induced nonlinear SI becomes the strongest interference at 14 dBm transmit power.

  • With the assumed separate transmit and receive antennas, antenna isolation provides natural SI mitigation; the proposed techniques can also apply to single-antenna circulator devices.Source: §II · p. 1
  • Passive cancellation can reach 40–45 dB, active RF cancellation adds 25–45 dB, and reported digital baseband cancellation adds up to 20–30 dB, but existing digital methods considered only linear SI.Source: §II · p. 2
  • The example assumes 40 dB passive cancellation, 30 dB active analog cancellation, and variable linear digital cancellation keeping linear SI below thermal noise; the desired signal is 5 dB above sensitivity.Source: §II · p. 2
  • At 14 dBm transmit power, PA-induced nonlinear SI is the strongest interference despite somewhat optimistic assumptions; more realistic passive isolation and RF cancellation would emphasize it further.Source: §II · p. 2

A. Baseband nonlinear self-interference channel model

The model represents the nonlinear PA and the subsequent SI path with a parallel Hammerstein structure. FIR responses capture the multipath channel and RF canceller, yielding an overall model whose effective coefficients can be estimated efficiently.

A. Baseband nonlinear self-interference channel model

  • The PA is modeled in discrete-time baseband with a parallel Hammerstein structure, using basis functions ψ_p(x_n) = |x_n|^(p−1)x_n and FIR branch responses.The memory length is M and the nonlinearity order is P.Source: A. Baseband nonlinear… · pp. 2–3
  • The multipath SI channel and frequency-dependent RF canceller response are each modeled with FIR filters acting on the PA output.Source: A. Baseband nonlinear… · p. 3
  • The overall SI channel combines the nonlinear PA, multipath SI channel, and RF canceller as a parallel Hammerstein model with effective coefficients and causal and non-causal memory depths.Source: A. Baseband nonlinear… · p. 3
  • Although nonlinear in the signal, the model is linear in its effective coefficients, enabling efficient estimation such as linear least squares.The model can also represent cascades involving polynomial or Hammerstein nonlinearities and an LTI system.Source: A. Baseband nonlinear… · p. 3

B. Digital SI canceller parameter estimation

The canceller estimates coefficients for a nonlinear SI model, regenerates the interference, and subtracts it from the received signal. Least squares minimizes the canceller output power while treating the desired signal as noise; estimation can use live data or embedded pilots.

B. Digital SI canceller parameter estimation

  • The canceller estimates coefficients for nonlinear terms across its memory depths, then regenerates the SI estimate for subtraction at digital baseband.Pc is the nonlinearity order; Mc1 and Mc2 are the non-causal and causal memory depths.Source: B. Digital SI canceller parameter… · pp. 3–4
  • Least squares selects the model coefficients to minimize digital SI-canceller output power, treating the received signal of interest as noise.The received signal also includes additive noise and modeled self-interference.Source: B. Digital SI canceller parameter… · p. 4
  • With full column rank in Ψ, the coefficient estimator has the stated solution, and its pseudoinverse can be precomputed for pilot-based estimation.The basis matrix is known inside the device.Source: B. Digital SI canceller parameter… · p. 4
  • Estimation can run during data transmission or use embedded pilots; recursive least squares avoids matrix inversion and is more appealing for live-data estimation.Precomputing the pseudoinverse for pilot-based estimation reduces real-time computational load.Source: B. Digital SI canceller parameter… · p. 4

IV. NUMERICAL RESULTS AND ANALYSIS

Full-waveform simulations show that nonlinear SI cancellation supports higher transmit power and lower PA IIP3 than linear cancellation. Its advantage grows as PA distortion becomes more significant, but high-power cancellation is constrained by quantization noise and model mismatch.

IV. NUMERICAL RESULTS AND ANALYSIS

  • The waveform simulations use OFDM, a multipath FIR SI channel, and a Wiener PA model, which the parallel Hammerstein estimator cannot model perfectly.The PH channel has order 5 with length-5 filters on each branch; estimation uses 3,200 samples, and residual SI can result from model mismatch.Source: §IV · p. 4
  • ≈10 dB higher transmit power is supported than with linear cancellation, regardless of antenna separation; at 50 dB separation, fidelity matches the no-SI case up to 20 dBm.The reported simulations use only 3,200 samples for estimation.Source: §IV · pp. 4–5, Figure 4 · p. 5
  • In the PA’s linear region, linear processing achieves more digital cancellation because nonlinear processing estimates more parameters; nonlinear cancellation gains increase as PA distortion grows.At very high transmit powers, quantization noise and model mismatch reduce achievable suppression.Source: Figure 5 · p. 5, §IV · p. 5
  • Up to 15 dB lower PA IIP3 is possible than with linear cancellation; at higher antenna separation, 14 dBm IIP3 achieves the ideal 15 dB SINR.PA linearity may still be bounded by spectrum-emission standards.Source: Figure 6 · p. 5, §IV · p. 6
  • At 20 dBm transmit power, antenna separation has little effect on achievable digital attenuation; the linear-versus-nonlinear gap is larger for less-linear PAs and narrows as PA linearity improves.The corresponding comparison is shown for the cancellation amounts in Fig. 7.Source: Figure 7 · p. 5, §IV · p. 6

V. CONCLUSION

Waveform simulations show that nonlinear digital SI cancellation extends the usable transmit-power range by at least 10 dB or permits a lower-quality PA. System power calculations identify PA nonlinearities as significant above about 10 dBm in an LTE uplink-like scenario.

V. CONCLUSION

  • At least 10 dB of usable transmit-power range is added by the proposed canceller, or a lower-quality transmitter PA can be used.Waveform simulations demonstrate both alternatives.Source: §V · p. 6
  • PA nonlinearities become a significant problem above about 10 dBm in an LTE uplink-like scenario, according to transceiver system power calculations.Source: §V · p. 6
  • The proposed digital SI canceller handles linear and nonlinear self-interference simultaneously.Source: §V · p. 6
  • Evaluation under realistic conditions with an implemented full-duplex transceiver and typical nonlinear PA is left for future work.Source: §V · p. 6
Loading 1401.3129v1…