Investigation of Performance Improvement of EAMVs in Diffuse Scattering Channels

Kriangsak Sivasondhivat, Jun‐ichi Takada · Tokyo Tech Research Repository (Tokyo Institute of Technology) · 2005

In order to reuse a channel model, it is requisite to separate the channel model from its measurement system and estimation procedures. For this purpose, the concept of a parametric channel model is introduced. Considering direction of arrival (DOA) estimation, a traditionally used signal model in a parametric channel model, which is a plane wave model, can be reasonable if and only if a measurement system has sufficient resolution to resolve multipaths. This condition is not the case in diffuse propagation channels, where a number of very close multipaths are incident at a measurement antennas with limited number of antenna elements. In terms of parameter estimation, the model mismatch between the resulting perturbed and expected signal models leads to the error in the estimated nominal DOAs of clusters in cluster scenarios. While in terms of signal reconstruction of the received signal, which can be more important than nominal DOA estimates in some applications, this model mismatch causes high residual power after the reconstructed signals are removed from the input signal. To deal with these problems, the application of the extended array mode vectors (EAMVs) based on the first and the second-order Taylor series expansion has been proposed [1][2]. In this paper, based on one- and two-cluster scenarios, the EAMVs’ performances from both viewpoints are investigated in the framework of successive interference cancellation (SIC) -based Maximum Likelihood Estimation (MLE), also known as SAGE with SIC [3][4]. 2 Signal Model and Concept of EAMVs In K-cluster scenarios with Dk scatterings, the output x (t) can be written as x (t) = K P k=1 v ksk (t) + n(t), where vk is the spatial signature (SS) of the k th cluster, sk (t) and n(t) denote the k th baseband and noise signals, respectively. Given the nominal azimuth DOA φ0,k and the angle deviation of the dth scattered multipath ˜ φd,k, vk is

Read the paper · More papers on PaperTik