Prediction techniques for haptic communication and their vulnerability to packet losses

We introduce three different uses of linear regression for improving the prediction of samples in haptic communication and compare this technique to the commonly employed zero-order and first-order linear predictors. We couple the prediction techniques with (error resilient) perceptual data reduction approaches and evaluate their robustness when losses in the network are present. Experimental results show that the proposed prediction technique improves haptic data reduction while keeping lower signal distortion compared to the traditional prediction methods when facing adverse network conditions.