Antenna De-Embedding Using Deconvolution With Tikhonov Regularization for mmWave Channel Measurement

Abstract

Antenna-free channel models can reflect real multipath propagation and can be applied widely for performance analysis, simulation, and physical emulation, combined with specific antennas used in the communication systems. However, the millimeter-wave (mmWave) channel measurement is usually performed by steering horn antennas, and the measured channel responses are actually spatial convolution of the channel propagation models and antenna pattern, which is commonly referred to as the antenna embedding effect. In this work, we propose a novel antenna de-embedding algorithm based on the deconvolution with Tikhonov regularization. By suppressing parts of the observed responses which are disguised by noise, the Tikhonov regularization facilitates the deconvolution of antenna pattern and enables the extraction of propagation models. In particular, in order to minimize the impact of deconvolved noise, we design an optimization algorithm to obtain the appropriate regularization factor with low computational complexity. To validate the proposed approach, we have performed an indoor mmWave channel measurement campaign using two different steering horn antennas. The principal peaks in the synthesized channel responses are accurately reconstructed, and the signal-to-noise ratio (SNR) is improved. The experiments verify that the proposed scheme de-embeds effectively the antenna effect and leads to the antenna-free channel models.