Search

In massive MIMO-OFDM systems, channel estimation is a significant module which can be utilized to eliminate multipath interference. However, in realistic communication systems, carrier frequency offset (CFO), which often exists in receive end, will deteriorate the performance of channel estimation. One of the effective solutions is to compensate CFO via the help of pseudo-noise (PN) sequence. At the beginning of this paper, to reduce system complexity and correctly compensate CFO, we propose an improved OFDM frame structure. Subsequently, we theoretically analyze the catastrophic influence of CFO on conventional PN-sequence-based compressed sensing (CS) channel estimation scheme. As our solution, based on the improved OFDM frame structure, a novel massive MIMO-OFDM channel estimation method under CFO environment is proposed. It first estimates CFO by utilizing differential correlation algorithm. Thereby, the interference caused by CFO can be eliminated. Then, relying on the PN sequence, the partial common support (PCS) information of each channel is obtained. Finally, using the PCS information as a priori information, we improve the CS reconstruction scheme to estimate the accurate channel. The simulation result shows that the proposed scheme demonstrates better MSE and BER performance than other mentioned schemes. The major advantage of our scheme is its anti-CFO ability and independence to channel sparsity level. Therefore, the proposed scheme is meaningful for practical use.

This article presents our initial results in deep learning for the complex multiple radar waveforms recognition. The method is composed of time-frequency analysis and deep residual network (ResNet). Firstly, we transform one-dimensional radar signals into two-dimensional time-frequency images (TFIs), which can reveal more characteristics of the signals. And then, we preprocess these images by grayscale, image opening operation and image resizing. Meanwhile, we design a ResNet and sent these preprocessed images into the network for off-line training. Finally, we use the trained model to recognize different modulation radar signals on-line and test the performance of the method. From our simulation results, the approach can achieve considerable performances that the overall recognition rate of 14 types radar signal is close to 96% when the SNR is −2dB.