Reference Receiver Based Digital Self-Interference Cancellation in MIMO Full-Duplex Transceivers

Transcription

1 Reference Receiver Based Digital Self-Interference Cancellation in MIMO Full-Duplex Transceivers Dani Korpi, Lauri Anttila, and Mikko Valkama Tampere University of Technology, Department of Electronics and Communications Engineering, Finland 10th International Workshop on Broadband Wireless Access

2 Outline Introduction RF impairments in a full-duplex transceiver Reference receiver based structure Initial analysis with system calculations Reference receiver aided digital cancellation procedure Waveform simulations Conclusion 2

3 Uplink Downlink Introduction Simultaneous transmission and reception at the same center frequency is an appealing scheme Increased data rate, MAC level benefits, etc. t FDD Single channel full-duplex f t t Uplink Downlink TDD Uplink & Downlink f f 3

4 Introduction (cont.) However, it has also its downsides Increased complexity due to self-interference cancellation, RF impairments, etc. Self-interference Tx Rx PA Attenuation & delay RF cancellation Transmitter chain DAC Receiver chain ADC SI regeneration Digital cancellation 4

5 RF impairments The self-interference (SI) signal is distorted in numerous ways within the transceiver chain, especially in the transmitter This affects the accuracy with which the SI signal can be regenerated for cancellation in the digital domain Advanced modeling is required to achieve a sufficient amount of SI cancellation 5

6 Reference receiver based structure A simple way of avoiding the complex modeling is to take the reference signal always from the PA output Then, the digital cancellation signal would include all the TX impairments Self-interference Tx Rx PA Attenuation & delay RF cancellation Transmitter chain Reference receiver DAC Receiver chain ADC SI regeneration Digital cancellation 6

7 Reference receiver based structure (cont.) The downside of this approach is the need for additional receivers In essence, a trade-off between the computational cost of digital cancellation and RF hardware It is not obvious what is the optimal scheme of utilizing the additional receivers 7

8 Detailed analysis TX1 Transmitter chains TX2-30 db PA VGA IQ Mixer LPF DAC Transmit bit stream 1 2x2 multipath coupling channel -30 db PA VGA IQ Mixer LPF Reference receiver chains IQ Mixer LPF VGA ADC DAC SI regeneration Transmit bit stream 2 SOI RX1 Amplitude & phase matching IQ Mixer LPF VGA ADC Receiver chains To detector RX2 Amplitude & phase matching RF cancellation BPF LNA IQ Mixer LPF VGA ADC Digital cancellation To detector RF cancellation BPF LNA IQ Mixer LPF VGA ADC Digital cancellation 8

9 Power of different signal components (dbm) Power of different signal components (dbm) Initial performance comparison with system calculations 0 Linear Digital Cancellation with Reference Receivers 0 Traditional Linear Digital Cancellation SI mirror image -70 Noise floor Quantization noise -80 RX-induced nonlinearities Signal of interest Transmit power (dbm) -60 SI mirror image Noise floor -70 Quantization noise TX-induced nonlinearities -80 RX-induced nonlinearities Signal of interest Transmit power (dbm) With the reference receiver based structure, even linear processing can provide sufficient SI suppression with transmit powers up to dbm

10 Reference receiver aided cancellation procedure The digital cancellation procedure is very straight-forward in this case The received SI signal in the digital domain during the training period can be expressed as s i,adc,tr = N T j=1 X j,pa,tr h ij + z i,tr = X PA,tr h i + z i,tr. Each X j,pa,tr is constructed from the samples obtained via the reference receivers

11 Parameter estimation For example, least squares can be used to estimate the SI channel: H h i = X PA,tr 1 H X PA,tr XPA,tr s i,adc,tr Using h i, a cancellation signal can be generated for any given transmit signal, which is then used to cancel the SI: y i,dc = s i,adc + r i X PA h i r i + z i, where r i is the signal of interest

12 Waveform simulations More comprehensive performance analysis using the same transceiver model as earlier Explicit basebandequivalent model for each component, resulting in realistic distortion of the SI signal 2x2 multipath coupling channel SOI TX2 RX2 TX1 RX1 Transmitter chains -30 db PA VGA IQ Mixer LPF -30 db PA VGA IQ Mixer LPF DAC Reference receiver chains IQ Mixer LPF VGA ADC IQ Mixer LPF VGA ADC Amplitude Receiver chains & phase matching RF can- BPF LNA IQ Mixer LPF VGA cellation Amplitude & phase matching RF can- BPF LNA IQ Mixer LPF VGA ADC cellation DAC ADC SI regeneration Digital cancellation Digital cancellation Transmit bit stream 1 Transmit bit stream 2 To detector To detector

13 SINR (db) Waveform simulations (cont.) The proposed scheme cancels SI efficiently With the higher transmit powers, it is limited by the impairments of the receiver chains Reference RX aided cancellation -15 Linear cancellation Widely-linear cancellation Nonlinear cancellation Transmit Power (dbm)

14 Conclusion The proposed reference receiver based scheme achieves a significant improvement in the SI cancellation ability utilizing only simple linear processing in the digital domain However, it requires additional hardware in the RF domain Additional research is required to assess whether this is the best way of using the additional receivers

15 Thank you Questions and/or comments are welcome!