Comparison of Edge Computing Methods for Environmental Monitoring IoT Sensors Using Neural Networks and Wavelet Transforms

IoT sensor solutions have become ubiquitous in recent years. This rapid growth has resulted in large amounts of real-time data being transmitted to a cloud. Cloud computing paradigms have extensive requirements on data rates in a communications channel. The current trend is to use edge computing techniques to process data files primarily on the edge of a network. The present paper compares neural networks and wavelet transforms which apply compression methods to environmental parameter datasets. Three selected parameters (temperature, air pressure and wind speed) were compressed at ratios of 1:2, 1:4, 1:8 and 1:12. In addition, three different types of wavelet were compared in the wavelet transform. The results showed that the best RMSE compression result can be achieved by using a Biorthogonal wavelet, which compressed up to 93% of the volume of data. In using a neural network to compress data, much depends on the nature of the data and also the amount of training data. With appropriately selected data, a volume