Time Series Analysis, Forecasting and Control
Time Series Analysis, Forecasting and Control
| Hi-index | 35.68 |
In this paper, we perform the prediction of the signal-to-interference-plus-noise ratio (SINR) root mean square (RMS) in the IEEE 802.16 system. We obtain the time series of the SINR RMS using system-level simulations. The SINR RMS is a heteroscedastic stochastic process. We propose a nonlinear transform of the SINR RMS that generates a linear homoscedastic stochastic process, which may be considered Gaussian in practice. We construct the prediction model of this Gaussian stochastic process using the linear autoregressive process. We propose three predictions of the SINR RMS, that is, the minimum mean square, the median, and the maximum-likelihood predictions, using the prediction of the linear autoregressive process. Our minimum mean-square error (MMSE) prediction of the SINR RMS is more reliable than the prediction based on averaging of the SINR RMS.