VLCdata

Mirror arrays have been applied to indoor visible light communication (VLC) as a passive the reconfigurable intelligent surface (RIS) which has no signal processing to solve the problem of indoor visible light line-of-sight obstruction, however, after channel modeling, it is found that the reflected channel in this scenario has a serious multipath effect, to this end, we introduce deep learning techniques into channel estimation of VLC systems with mirror arrays for the first time, and propose a hybrid new model of Transformer and the bidirectional longshort-term memory model (Transformer-BiLSTM) for estimating the reflected channel passing through the mirror arrays. Different from the general data processing in the network, the data in this network has been modeled by the cyclic structure and the self-attention mechanism to model the sequences, and the multi-attention mechanism can learn the characteristics of the sequences under different time and dimensions. In this paper, the channel model under prior research is used to process the signal to obtain the required channel data to train the network offline and fully extract the channel features in the training samples. Experiments are conducted on the basis of three independent variables and show that the mean square error is greatly reduced compared to the conventional channel estimation algorithm. Simulation results show that this Transformer-BiLSTM model is robust to variations in system parameters, i.e., the transmit signal-to-noise ratio, the number of subcarriers and the number of RIS elements. In addition, compared to the use of Transfomer, BiLSTM, Deep Neural Networks (DNN) model and, BiLSTM, Multilayer Perceptron (MLP) with transformer tandem structure (BiTran), has superior estimation performance.