Acquisition of state and DOS features based channel estimation for VTV mmWave-Massive MIMO: A Deep nested with Layered LSTM approach

Abstract

Millimetre Wave integrated Massive Multiple Input and Multiple Output (mmWave-Massive MIMO) technology is an exciting area of research to enable Vehicle to Vehicle Communication (VTVCom) in a variety of fields. The combination of mmWave-Massive MIMO technology supports a wide range of applications, from 5G networks to autonomous vehicles. Based on this context, each VTV User communication Link (VUL) is expected to enable high-capacity and reliable communication between vehicles in a dense urban environment over rapidly time varying channels. However, the vehicular channel is degraded due to the presence of other vehicles in the vicinity of each VUL. This interference can cause tracking algorithms to fail or become less effective, which can lead to degraded VTV link quality and communication errors. To solve these problems, this paper proposes a deep learning based Direction Of Signal (DOS) tracking inspired channel estimation model using Deep Nested-Layered Long Short-Term Memory network (DN-LrLSTM) for VTVCominmmWave-Massive MIMO systems. In this, we employed a DN-LrLSTM model to acquire the state channel features of VUL, and to update various channel parameters by capturing DOS tracking features over varying channels.The DOS tracking based channel estimation model solves the vehicle degradation problems and accurately estimates various channel parameters. Experimental results are conducted on an urban road VTV communication scenario, which shows the performance of the proposed DN-LrLSTM framework achieves high capability in capturing complicated channel features of VUL and outperforms robust performance on accurate channel estimation in terms of several evaluation metrics.