Predicting touch operations by using hover information in smartphones for data prefetching

This paper presents new methods to predict input operations on smartphones. Frequently occurring poor network conditions such as low signal strength or congestion of the radio link can increase the latency of smartphone operations. Prediction can improve the latency by prefetching the content that is requested by the predicted operation. We developed methods to predict the input operation by using hover detection on the touch screen. Smartphone functions in recent years have included a function to detect the position of a finger several centimeters above the touch screen. The touch position and the type of operation to be performed are predicted by using the detected positions of the finger before the finger touches the screen. We evaluated this method on a smartphone launcher application. The results indicated that our methods were effective for predicting the operation and improving the latency.