标题： 基于机器学习的5G NR PRACH混合接收机 显示英文标题

标题： A Machine Learning based Hybrid Receiver for 5G NR PRACH

摘要： 随机接入是一种关键程序，用户设备（UE）通过它向基站（BS）标识自己。 随机接入始于UE在物理随机接入信道（PRACH）上发送一个随机前导码。 在传统的BS接收机中，UE的特定前导码通过与所有可能的前导码进行相关性检测来识别。 PRACH信号还用于估计由传播延迟引起的定时提前量。 基于相关性的接收器在高衰落和低信噪比的场景下会遭受虚假峰值和漏检的问题。 本文描述了一种混合接收器的设计，该接收器首先使用一种AI/ML模型进行前导码检测，然后通过传统峰值检测进行定时提前量估计。 所提出的接收器结合了多个天线上的相关窗口的功率时延轮廓，并将其作为神经网络模型的输入。 该模型预测特定前导码窗口内是否存在用户，之后通过峰值检测估计定时提前量。 结果显示，对于模拟数据集和真实硬件捕获的数据集，该混合接收器的表现优于传统的接收器。 显示英文摘要

摘要： Random Access is a critical procedure using which a User Equipment (UE) identifies itself to a Base Station (BS). Random Access starts with the UE transmitting a random preamble on the Physical Random Access Channel (PRACH). In a conventional BS receiver, the UE's specific preamble is identified by correlation with all the possible preambles. The PRACH signal is also used to estimate the timing advance which is induced by propagation delay. Correlation-based receivers suffer from false peaks and missed detection in scenarios dominated by high fading and low signal-to-noise ratio. This paper describes the design of a hybrid receiver that consists of an AI/ML model for preamble detection followed by conventional peak detection for the Timing Advance estimation. The proposed receiver combines the Power Delay Profiles of correlation windows across multiple antennas and uses the combination as input to a Neural Network model. The model predicts the presence or absence of a user in a particular preamble window, after which the timing advance is estimated by peak detection. Results show superior performance of the hybrid receiver compared to conventional receivers both for simulated and real hardware-captured datasets.