Cross-Region Data Transmission Channel State Prediction Method Based on Service Traffic Thermal Characteristics

At present, the cross-region end-to-end communication of a power distribution network relies on heterogeneous networks such as carrier networks and power fiber networks to support the specialization of relay nodes, and the corresponding state volatility increases abruptly. Furthermore, the corresponding state information is often difficult to obtain in full in real-time, which leads to difficulty in predicting the channel state, and it is not possible to reasonably guide the joint deployment of sensing, communication, and computing resources. Therefore, this study proposes a cross-region data transmission channel state prediction method based on the thermal characteristics of the service traffic. Firstly, this study establishes a cross-region end-to-end data transmission system model for a new type of power distribution system, then refers to the relevant theory of thermodynamics, proposes an evaluation method of the thermal characteristics of the service traffic adapted to the multi-timescale, and finally designs the channel anomaly state evaluation function. Then, based on the thermal characteristics of the historical service traffic, the future state of the specified channel based on the LSTM (Long Short-Term Memory) neural network is predicted. Finally, the channel anomalous state assessment function is designed to accurately predict the future channel state under the specified time scale based on the thermal characteristics of the historical service traffic and based on the LSTM neural network. Simulation analysis shows that the proposed service traffic characteristic model can reasonably quantitatively characterize the change of channel state and accurately predict the state of inter-regional data transmission.