Deep Learning Based Packet Detection and Carrier Frequency Offset Estimation in IEEE 802.11ah
share
Data-based machine learning methods are currently disrupting communication engineering physical layer research. Promising results have been presented in the recent literature, in particular in the domain of deep learning-based channel estimation. In this paper, we investigate deep neural network (DNN)-based solutions for packet detection and carrier frequency offset (CFO) estimation. We focus on preamble-based OFDM systems such as IEEE 802.11, and apply the investigated DNN-based methods in emerging IEEE 802.11ah standard. Our investigation, performed within a detailed standard-based simulated environment, demonstrates competitive performance of DNN-based methods as compared to the conventional ones. In particular, convolutional neural network and recurrent neural network architectures applied in packet detection and CFO estimation, respectively, demonstrated robustness and accuracy that matched and even surpassed the conventional methods.
READ FULL TEXT
