SIR estimation and closed-loop power control for 3G

Several estimators of the received SIR (signal-to-noise and interference ratio), a key parameter for closed-loop power control, are considered. Three estimators are considered, two of which use coherent processing and one uses non-coherent processing. The first two methods require an estimate of the multipath fading channel. Both causal and non-causal estimation methods are considered. While non-causal channel estimation improves the fading channel estimation quality (defined in terms of a small MSE between the true fading channel and its estimate) it is shown that the resultant signal energy estimate is biased. Furthermore, the SIR estimator using non-causal channel estimates introduces additional delay in the power control loop. On the other hand, the SIR estimator using causal channel estimates is less accurate (in terms of a larger channel estimation MSE) compared to the one using non-causal channel estimation, however it avoids the bias in the energy estimation. This method of estimating the SIR also reduces the latency in the power control loop. The third estimator is noncoherent, i.e. it does not make use of the knowledge of the fading channel profile. This estimator of the signal energy introduces a noise-bias which is subtracted using a noise estimation technique. It is shown that the non-coherent estimator performs better than the two coherent methods and is robust to the channel fading conditions. The effect of these estimators on the block error rate (BLER) performance of a 3GPP radio link is carried out in a multipath Rayleigh fading channel.