2009/3

Browse

Recent Submissions

Now showing 1 - 5 of 11
  • Thumbnail Image
    Item
    Experimental Characterization of System Parameters for Ranging in IEEE 802.15.4a using Energy Detectors
    (Společnost pro radioelektronické inženýrství, 2009-09) Geiger, Bernhard; Gigl, Thomas; Preishuber-Pfluegl, Josef; Witrisal, Klaus
    The IEEE 802.15.4a standard for impulse radio ultrawide band (IR-UWB) communication systems defines a ranging scheme which relies on the measurement of the round-trip propagation time of electromagnetic pulses. Accuracy is strongly dependent on the estimation of the timeof-arrival (TOA) of the pulse that is spread in time due to multipath propagation. The major concern therefore is the proper detection of the leading edge. In this work, the ranging capabilities of the standard are analyzed for an energy detector receiver. Emphasis is put on the influence of transmitter and receiver parameters, which are evaluated for a set of measured scenarios. It is shown that sub-meter ranging accuracy can be achieved with fixed parameter settings.
  • Thumbnail Image
    Item
    Imaging of the Building Contours with Through the Wall UWB Radar System
    (Společnost pro radioelektronické inženýrství, 2009-09) Aftanas, Michal; Drutarovsky, Milos
    Any actual information about a building interior can be very useful before entering a dangerous area. It can be used to plan strategies in many rescue and security applications. The paper deals with imaging of the inner and outer building contours from the outside using through the wall UWB radar. The whole processing chain for obtaining the contours of a scanned building is explained. The image processing method of highlighting the building walls using Hough transform with assumed knowledge of the direction of walls is presented. The algorithm was tested on real measurement data acquired from a M-sequence UWB radar system.
  • Thumbnail Image
    Item
    Adaptive Digital Image Watermarking Based on Combination of HVS Models
    (Společnost pro radioelektronické inženýrství, 2009-09) Foris, Peter; Levicky, Dusan
    In this paper two new blind adaptive digital watermarking methods of color images are presented. The adaptability is based on perceptual watermarking which exploits Human Visual System (HVS) models. The first method performs watermark embedding in transform domain of DCT and the second method is based on DWT. Watermark is embedded into transform domain of a chosen color image component in a selected color space. Both methods use a combination of HVS models to select perceptually significant transform coefficients and at the same time to determine the bounds of modification of selected coefficients. The final HVS model consists of three parts. The first part is the HVS model in DCT (DWT) domain. The second part is the HVS model based on Region of Interest and finally the third part is the HVS model based on Noise Visibility Function. Watermark has a form of a real number sequence with normal distribution.
  • Thumbnail Image
    Item
    A Novel First-Order Current-Mode All-Pass Filter Using CDTA
    (Společnost pro radioelektronické inženýrství, 2009-09) Lahiri, Abhirup; Chowdhury, Ankush
    This paper presents a new realization of a firstorder current-mode (CM) all-pass filter (APF) using the recently proposed modern active building block (ABB), namely the current differencing transconductance amplifier (CDTA). The CM APF is made using minimum number of components, namely a single CDTA and one grounded capacitor. The circuit does not use any external resistors and offers the advantages of current-tunable pole frequency, low input impedance and high output impedance. Non-ideal analysis and sensitivity analysis are provided and PSPICE simulation results are included to verify the workability of the circuit.
  • Thumbnail Image
    Item
    Comparison of Generative and Discriminative Approaches for Speaker Recognition with Limited Data
    (Společnost pro radioelektronické inženýrství, 2009-09) Silovsky, Jan; Cerva, Petr; Zdansky, Jindrich
    This paper presents a comparison of three different speaker recognition methods deployed in a broadcast news processing system. We focus on how the generative and discriminative nature of these methods affects the speaker recognition framework and we also deal with intersession variability compensation techniques in more detail, which are of great interest in broadcast processing domain. Performed experiments are specific particularly for the very limited amount of data used for both speaker enrollment (typically ranging from 30 to 60 seconds) and recognition (typically ranging from 5 to 15 seconds). Our results show that the system based on Gaussian Mixture Models (GMMs) outperforms both systems based on Support Vector Machines (SVMs) but its drawback is higher computational cost.