Jürgen Götze

An algorithm and architecture based on orthonormal m-rotations for computing the symmetric EVD

Abstract In this paper an algorithm and architecture for computing the eigenvalue decomposition (EVD) of a symmetric matrix is presented. The EVD is computed using a Jacobi-type method, where the angle of the rotations is approximated by an angle α κ , corresponding to an orthonormal μ-rotation. These orthonormal μ-rotations are based on the idea of CORDIC and share the property that performing the rotation requires a minimal number of shift-add operations. We present various methods of contsruction for such orthonormal μ-rotations of increasing complexity. Moreover, the computations to determine which angle α κ to use in the approximation of the optiimal angle, can itself be expressed purely in orthonormal μ-rotations on the matrix data. The complexity of the used type of orthonormal μ-rotation decreases during the diagonalization of the matrix. A significant reduction of the number of required shift-add operations is achieved. All types of fast, orthonormal μ-rotations (and the computation to determine the optimal angle) can be implemented as a cascade of only four basic types of shift-add stages. These stages can be executed on a modified sequential floating-point CORDIC architecture, making the presented algorithm highly suited for VLSI-implementation.

New applications for wide-area monitoring, protection and control

As operation of electrical transmission systems faces new challenges due to liberalization and integration of renewable energies, novel solutions for power system management are needed. At the same time, the development of smart grids also pushes high performance communication and information technology that contribute decisively in enabling a dynamic monitoring, protection and control of large-scale and wide-spread power systems. The potential offered by smart grids and new fast controllable power systems equipment needs to be exploited by development of valuable applications for power system operation. Moreover, first experiences with selected dedicated applications call for the design of an integrated wide-area monitoring, protection and control (WAMPAC) system. This paper presents recent progress of research unit FOR1511 in the development of WAMPAC applications for stability monitoring, protection schemes based on wide-area information, and real-time congestion management. The research unit aims at developing a coherentWAMPAC system taking into account interdependencies and synergies of newly developed applications and conventional local systems in place.

Camera-assisted localization of passive RFID labels

For indoor localization systems, Radio Frequency Identification (RFID) is an often chosen technique. A passive UHF RFID label can be localized via Received Signal Strength Indicator (RSSI) values using a Constrained Unscented Kalman Filter (CUKF). A camera-based localization technique which employs back projection method and movement estimation is combined with the RFID-based localization. This camera-assisted localization technique leads to an increase in localization accuracy by a factor of two compared to the Constrained Unscented Kalman Filter without camera assistance which already performes twice as good as the Unscented Kalman Filter (UKF).

Common Architectures for TD-CDMA and OFDM Based Mobile Radio Systems Without the Necessity of a Cyclic Prefix

In this article an OFDM system without the necessity of a cyclic prefix is presented. First a generalised data model that describes different mobile radio systems by a set of parameters is developed. That model is taken as a basis to perform data detection general enough to be applicable in different mobile radio systems. The differences and similarities of different systems with regard to the data model are elaborated. It comes out that in order to enable common detection strategies the cyclic prefix in OFDM based systems can be discarded, which implies that the information data rate is increased. The price is a higher computational requirement of the receiver. Following the desire not to increase the computational requirements for data detection significantly a technique is found to perform data detection in OFDM systems without cyclic prefix causing only a moderate increase in computational requirements.

Increased Robustness and Security of Digital Watermarking Using DS-CDMA

Digital Watermarking is a popular method for discouraging illicit copying of multimedia data. In this paper a blind, robust and secure watermarking technique based on DS-CDMA is presented. By using this technique a watermark can be extracted at very high PSNR value with very low BER. An adaptive approach is used to select the coefficients in the DCT domain, where the watermark is to hide. This significantly reduces BER. Multiple spreading codes can be used to make this scheme secure and more robust against JPEG compression, scaling and cropping.

Localization of UHF RFID labels with reference tags and Unscented Kalman Filter

To accomplish location-awareness in indoor scenarios Radio Frequency Identification (RFID) tags are often used for the localization. This paper only uses Received Signal Strength Indicator (RSSI) values from passive UHF RFID labels for the localization process. The RSSI values of reference tags (RTs) for which the locations are known are employed as side information for the localization of an RFID label (LT). For the localization of this LT an Unscented Kalman Filter (UKF) in combination with the RTs information is used. Simulations based on measured data achieve a localization accuracy of about 17cm with this localization method, compared to a localization error of 20cm without RTs and 83cm when only using k nearest neighbor (kNN) algorithm.

Parameterization and implementation of orthogonal wavelet transforms

A method is presented that parameterizes orthogonal wavelet transforms with respect to their properties (i.e. compact support, vanishing moments, regularity, symmetry) and also takes into consideration a simple implementation of the transform. The parameter space is given by the rotation angles of the orthogonal 2/spl times/2 rotations used in the lattice filters that realize the different stages of the wavelet transform. The different properties of an orthogonal wavelet transform can be expressed in this parameter space. Then, restricting the parameter space to the rotation angles of simple CORDIC-based approximate rotations leads to a reduced parameter space. The wavelet transforms in this reduced parameter space are amenable to a very simple implementation (only a small number of shift and add operations).

Low-complexity multi-purpose IP Core for quantized Discrete Cosine and integer transform

In this paper a low-complexity and highly-integrated IP Core for image/video transformations is presented. It is possible to perform quantized 8×8 DCT and quantized 8×8/4×4 integer transforms on the presented reconfigurable architecture using only shift and add operations. The XVID experimental and FPGA synthesis results show that the proposed architecture not only achieves multiplierless video transformations efficiently, but also retains good transformation quality. It is worth noticing that the proposed IP Core is very suitable for low-complexity and multi-purpose Video CODECs in SoC designs.

Multiwavelet transforms based on several scaling functions

An algebraic method for the design of discrete wavelet transforms based on several scaling functions is presented. Solving systems of partly nonlinear equations is necessary to compute the discrete coefficients. Wavelet transforms based on several scaling functions enable properties that are impossible in the single-wavelet case. Wavelet transforms based on several scaling functions can also be designed wavelet-like, what leads to a better approximation of the continuous bases. In this paper we show how to construct the discrete wavelet transforms based on several scaling functions with the algebraic design method and discuss the properties of the resulting wavelet bases.<<ETX>>

Efficient data detection algorithms in single- and multi-carrier systems without the necessity of a guard period

To enable efficient detection strategies in wireless communications a guard period is frequently inserted into the transmitted data symbols. This is the case in OFDM systems and in single-carrier systems with frequency domain equalization. Especially for transmissions over wireless channels with long delay spreads the required guard period is quite long. Here we present efficient data detection algorithms that are applicable if a guard period is not inserted after every symbol or if it is omitted completely. The increase of the computational requirements in the base station is moderate and can be implemented via parallel processing. Finally, an OFDM system without guard periods is presented as an example.