Performance of send-on-delta sampling schemes with prediction

In the paper, the performance of the sampling algorithm as a measure that evaluates together the sampling cost and the sampling accuracy is introduced and analyzed. We define the sampling cost as a number of samples in a unit of time and respectively the sampling accuracy as the mean intersampling error. On the basis of the sampling scheme performance measure, we compare the pure send-on-delta scheme and the send-on-delta scheme with prediction because these sampling algorithms are developed on controlling the linear intersampling error. As shown on the basis of simulation results, the send-on-delta schemes with first or second-order prediction outperforms the classical periodic sampling, and the pure send-on-delta scheme.