This paper investigates methods for link quality measurement of an indoor, time-varying wireless link. Link quality estimates are used for a number of higher-layer functions, including rate adaptation, routing, and topology control. The objective is to rapidly and efficiently estimate the current reliability of an RF link in terms of its packet success probability in the presence of a time-varying channel. We compare various approaches to estimating the current packet success rate (PSR), including packet counting and several methods employing physical layer signal-to-noise ratio (SNR) measurements. Among the SNR-based estimators, we consider those using a simple moving average, an exponential moving average, and a Yule-Walker predictor of the current SNR. The analysis shows that the SNR-based estimators are more efficient than packet counting methods in terms of the number of measurements needed, are more accurate when link quality variability increases, and are more flexible in terms of predicting the PSR for various bit rates and packet sizes. An important requirement, however, of the SNR-based estimators is a priori knowledge of the SNR-PSR relationship, which is environment- and radio-dependent. An efficient methodology for obtaining this mapping is presented.
Proceedings Title: IEEE ICC 2009
Conference Dates: November 30-December 4, 2008
Conference Location: Dresden, LA
Pub Type: Conferences
Channel estimation, Indoor, Link quality, Packet counting