Rural and remote communities in Canada often rely on satellites to access the internet, but those connections are fraught with many glitches and service interruptions because the technology can be unreliable. The inequity in internet access between these communities and those who live in cities is an ongoing problem with myriad consequences for Canada's economic productivity. A team of researchers from the University of Waterloo and the National Research Council (NRC) are tackling this long-standing issue using machine learning. The team's method, the Multivariate Variance-based Genetic Ensemble Learning Method, merges several existing AI-driven models to detect anomalies in satellites and satellite networks before they can cause major problems. "For remote areas in Canada and around the world, satellites are often their best option for maintaining internet access," said Peng Hu, an adjunct professor of computer science and statistics and actuarial science at Waterloo and the corresponding author of the study.

In this paper, we use a variance-based genetic ensemble (VGE) of Neural Networks (NNs) to detect anomalies in the satellite's historical data. We use an efficient ensemble of the predictions from multiple Recurrent Neural Networks (RNNs) by leveraging each model's uncertainty level (variance). For prediction, each RNN is guided by a Genetic Algorithm (GA) which constructs the optimal structure for each RNN model. However, finding the model uncertainty level is challenging in many cases. Although the Bayesian NNs (BNNs)-based methods are popular for providing the confidence bound of the models, they cannot be employed in complex NN structures as they are computationally intractable. This paper uses the Monte Carlo (MC) dropout as an approximation version of BNNs. Then these uncertainty levels and each predictive model suggested by GA are used to generate a new model, which is then used for forecasting the TS and AD. Simulation results show that the forecasting and AD capability of the ensemble model outperforms existing approaches.