Beschreibung:
The present manuscript introduces a method for evaluating and forecasting the quality of vehicle-to-infrastructure (V2I) communication channels in urban settings. This method precisely classifies and predicts channel quality levels in V2I scenarios based on long-range (LoRa) technology. This approach aims to accurately classify and predict channel quality levels in V2I scenarios. The concept of channel quality scoring was first introduced, offering a more precise description of channel quality compared to traditional packet reception rate (PRR) assessments. In the channel quality assessment model based on the gated recurrent unit (GRU) algorithm, the current channel quality score of the vehicular terminal and the spatial channel parameters (SCP) of its location are utilized as inputs to achieve the classification of channel quality levels with an accuracy of 97.5%. Regarding prediction, the focus lies in forecasting the channel quality score, combined with the calculation of SCP for the vehicle’s following temporal location, thereby achieving predictions of channel quality levels from spatial and temporal perspectives. The prediction model employs the Variational Mode Decomposition-Backoff-Bidirectional Long Short-Term Memory (VMD-BO-BiLSTM) algorithm, which, while maintaining an acceptable training time, exhibits higher accuracy than other prediction algorithms, with an R2 value reaching 0.9945. This model contributes to assessing and predicting channel quality in V2I scenarios and holds significant implications for subsequent channel resource allocation.