Taming Uncertainties in Real-Time Routing for Wireless Networked Sensing and Control

Taming uncertainties in real-time routing for wireless networked sensing and control

“…For instance, Figure 12 shows path in the NetEye medium traffic scenario, the time series of the actual cumulative distribution function (CDF) values of the estimated 90 percentiles using the P 2 algorithm. We see that it takes more than 200 samples for the estimation to converge, and this observation holds in general [38].…”

“…In our experiments, we use 90% as the required real-time guarantee probability by default, but we have also experimented with the real-time guarantee probability of 99% and observed similar phenomena [39]. For differentiating the performance of different protocols, the deadline for each traffic scenario is chosen so that it is neither too stringent (that no protocol can support) nor too loose (that all protocols can support).…”

“…Since nodes of WSC networks tend to be resource-constrained (e.g., in CPU and memory), it is infeasible to directly compute the distribution of d P (t) in general. One alternative is to first sample d P (t) and then use non-parametric approaches (e.g., the P 2 algorithm) to estimate delay quantiles based on these delay samples [31,38]. Nonetheless, non-parametric quantile estimation usually converges slowly.…”

“…For comparison purpose, we present the probabilistic delay bounds estimated via the optimal-partition-minimum-delay (OPMD) method [38] or by assuming path delay is normally distributed [34]; we also present the maximum path delay. Given that Markov Inequality usually gives a looser bound than one-tailed Chebyshev Inequality, the probabilistic delay bound by Markov Inequality is greater than that by Chebyshev Inequality.…”

Enhanced Markov-Difference Based Power Consumption Prediction for Smart Grids

“…There are various available forecasting methods for power data in Smart Grid, such as trend extrapolation forecasting (TEF), regression analysis forecasting (RAF), artificial neural network forecasting (ANNF), gray theory forecasting (GTF), time series forecasting (TSF), wavelet analysis forecasting (WAF) and so on. Recently, many researches have studied different kinds of mathematics models for the electricity prediction [32]- [36]. These works make great contributions to improving prediction accuracy in Smart Grids.…”