Abstract | ||
---|---|---|
This paper compares parametric and nonparametric channel es- timation methods in a multipath environment. For both meth- ods, the MSE on the received symbol pulse is shown to con- sist of a modeling error term and an estimation error term. We lower bound the latter term by computing the associated Cramer-Rao bounds. We point out that the MSE can be min- imized by a proper selection of the number of parameters to be estimated. In addition, we present simulation results from practical estimators that confirm our theoretical analysis. Our numerical results indicate that the parametric method outper- forms the nonparametric approach. Multipath channels give rise to inter-symbol interference (ISI) at the receiver. The traditional approach to counter the effects of ISI is to equalize the received symbol pulse. In order to successfully mitigate the ISI, an accurate knowledge of the re- ceived signal pulse is paramount. Hence, channel estimation is an important task of the receiver in digital communication systems. A parametric channel estimation method exploits the mul- tipath structure by estimating the path gains and delays, and computing from these estimates the corresponding received symbol pulse. In a nonparametric channel estimation method the samples of the received symbol pulse are estimated with- out taking the multipath structure into account. Empirical re- sults have indicated that the estimation accuracy improves by exploiting the underlying channel structure (1, 2, 3). In this contribution, we investigate the mean squared error (MSE) on the received symbol pulse, resulting from the two estimation methods. We present a lower bound on the MSE, which is based on the Cramer-Rao lower bound (CRB), and provide simulation results that confirm our theoretical deriva- tions. Our numerical results illustrate that the parametric esti- mation method yields the smaller MSE. |
Year | DOI | Venue |
---|---|---|
2007 | 10.1109/PIMRC.2007.4394436 | PIMRC |
Keywords | Field | DocType |
cramer rao bound,model error,lower bound | Multipath channels,Multipath propagation,Upper and lower bounds,Algorithm,Communication channel,Real-time computing,Nonparametric statistics,Parametric statistics,Statistics,Mathematics,Estimator | Conference |
ISBN | Citations | PageRank |
978-1-4244-1144-3 | 0 | 0.34 |
References | Authors | |
4 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Dieter Van Welden | 1 | 23 | 4.58 |
F. Simoens | 2 | 64 | 5.90 |
Heidi Steendam | 3 | 423 | 48.24 |
M. Moeneclaey | 4 | 1471 | 274.44 |