Phase Noise Estimation with 1-bit Quantization and Oversampling

A promising approach to avoid the bottleneck of the analog-to-digital converters (ADCs) high power consumption at high sampling frequencies is to use only 1-bit quantization resolution. By using temporal oversampling at the receiver, a high resolution in the time-domain can be achieved, which can partly recover the losses in terms of rate caused by a reduced amplitude resolution. Channel estimation and synchronization has to be performed on 1-bit quantized receive samples, which poses a new challenge. This work is concerned with the phase estimation of a 1-bit quantized system with phase noise in the low signal-to-noise (SNR) range. A block-based least squares (LS) estimator interpolated by a Kalman filter is presented to track the phase noise. To enhance the performance especially for the case of a large spacing between pilot blocks we study the Rauch-Tung-Striebel (RTS) algorithm. Both algorithms are adjusted to the system characteristics and bounds for the steady state performance are derived. Using these bounds and numerical results it is shown, that the RTS algorithm achieves a lower error variance than the Kalman filter at the price of increased latency as it works on entire frames