MCINet: Multi-Channel Informer Networks for Long Sequence Time-Series Forecasting in Ocean-Going Ship Satellite Communication Networks

Long sequence time-series forecasting (LSTF) in ocean-going ship satellite communication networks (OSSCN) is a critical task for effective network planning, resource allocation, capacity planning, and operation management. However, traditional prediction models face challenges such as insufficient data extraction, high complexity, and difficulty in handling heterogeneous datasets, leading to lower prediction accuracy and reliability. To address these issues, this paper proposes a novel data-driven approach, called Multi-Channel Informer Networks (MCINet), for LSTF in OSSCN. Specifically, the OSSCN data is processed using the multi-head self-attention mechanism and convolutional layers of MCINet, allowing for the full utilization of the data information. Additionally, we leverage the gradient centralized (GC2) optimizer and AdamW optimizer to reduce the computational complexity and training time of the model, thereby better fitting the complex OSSCN data. Furthermore, The MCINet is connected across layers to better handle heterogeneous data and avoid overfitting, leading to improved prediction accuracy and reliability. We compared MCINet with other prediction models on two public datasets and a real ship network traffic dataset OSSND, and the results proved that our method is superior to the current state-of-the-art methods.