Josef A. Nossek

Comparison between unitary ESPRIT and SAGE for 3-D channel sounding

The detailed knowledge of the directional characteristics of the mobile radio channel is required to develop directional channel models and to design efficient smart antenna concepts for future mobile radio systems. In this paper, we examine the applicability and performance of two popular high-resolution parameter estimation schemes for directional channel sounding, namely subspace-based unitary ESPRIT and the maximum-likelihood-based SAGE algorithm. To this end, an investigation of the resolution capability using a synthetic scenario is carried out for both schemes by means of Monte-Carlo simulations. The performance is also compared to the Cramer-Rao lower bound (CRLB). Moreover, we apply both parameter estimation schemes to propagation data generated by a ray-tracing model. This model is based on indoor channel measurements generated by the wideband channel sounding system ECHO 24. The channel measurements were taken inside a radio frequency shielded room (RFSR), an environment dominated by severe specular multipath and, therefore, can be considered a worst case scenario.

Adaptive space-frequency RAKE receivers for WCDMA

Adaptive space-frequency RAKE receivers use maximum ratio combining and multi-user interference suppression to obtain a considerable increase in performance in DS-CDMA systems such as WCDMA. To this end, the signal-plus-interference-and-noise and the interference-plus-noise space-time covariance matrices are estimated. The computational complexity is reduced significantly by transforming the covariance matrices into the space-frequency domain and by omitting noisy space-frequency bins. The optimum weight vector for symbol decisions is the largest generalized eigenvector of the resulting matrix pencil. By iteratively updating the optimum weight vector slot by slot, real-time applicability becomes feasible while the fast fading is still tracked. The performance and the computational complexity depend on the number of space-frequency bins, antenna elements, and iterations. Therefore, the performance can easily be scaled with respect to the available computational power.

Calibration of smart antennas in a GSM network

Smart antennas can achieve a considerable gain in spectral efficiency provided that errors due to mutual coupling and amplitude or phase errors of the antenna elements are negligible. Calibration by the knowledge of the directions of the wavefronts transmitted by a second base station lessens the influence of these imperfections. This study describes the calibration results of general planar antenna arrays in the existing GSM network of the German network operator Mannesmann Mobilfunk GmbH. The presented direction based calibration algorithm does not only take into account the multipath propagation of discrete wavefronts, but also the considerable angular spread. Such a procedure is possible when the base station is carrying operational traffic and is not limited to offline or even preinstallation calibration. A discussion gives some insights on how to obtain best calibration results.

Low complexity space-frequency RAKE receivers for WCDMA

Adaptive space-frequency RAKE receivers use diversity combining and multiuser interference suppression to obtain a considerable increase in performance in DS-CDMA systems such as WCDMA. The main advantages of operation in the space-frequency domain include a reduced computational complexity and an improved noise suppression. To this end, the signal-plus-interference-and-noise (SIN) and the interference-plus-noise (IN) space-frequency covariance matrices are required. The optimum weight vector for symbol decisions is the largest generalized eigenvector of the resulting matrix pencil. If we decouple spatial and frequency processing with respect to interfering users, the IN space-frequency covariance matrix can be approximated by the Kronecker product of the frequency and the spatial covariance matrix. That this IN covariance matrix is estimated by using the outputs of the antenna elements before correlation and the output of the conventional RAKE fingers of the antenna elements may be utilized to approximate the SIN covariance matrix. Thus, the required correlations are reduced to the number of RAKE fingers. Moreover, the computational complexity which is required to estimate the optimum weight vector may be reduced significantly.

Device level based cell modeling for fast power estimation

In this work, a method for fast power estimation in complex digital circuits is presented. Properties like delay and power consumption of a circuits basic cells are extracted precisely by circuit level simulations. The cell model then includes delay and power consumption values for all possible transitions at the cell inputs. The total power consumption is finally determined by logic simulation at higher architectural levels. The cell model also includes the glitching behavior at the outputs resulting from different path delays inside a cell. It is shown experimentally that this delay model, including glitches generated by the basic cells, leads to good power estimation results of complex circuits within an accuracy of 8% in the worst case and needs 4 and 2 orders of magnitude less simulation time than SPICE and PowerMill respectively.

Orthogonally-anchored adaptive CDMA interference rejection in space-time rake processing

The paper presents an orthogonally-anchored adaptive linear interference rejection technique in space-time rake processing. With the use of a super short training sequence, the initial space-time adaptive filtering vector composed of two mutually orthogonal components is calculated by using a projection-based filter optimization technique. The adaptive filter optimization based on the pre-optimized initial weight vector is implemented to efficiently suppress the multiuser access interference in the filtering procedure. In contrast to the blind space-time rake processing, the computation efficiency exhibited in the suggested approach is achieved by implementing the linear optimization for the filtering vector in reduced-dimensional (or full-dimensional) space-time complex vector spaces instead of using eigenvector-based processing, in which the optimization of the filtering vector in full-dimensional data vector space is required.