One of the most powerful and important applications that the brain accomplishes is solving the sensory "cocktail party problem:" to adaptively suppress extraneous signals in an environment. Theoretical studies suggest that the solution to the problem involves an adaptive filter, which learns to remove the redundant noise. However, neural learning is also in its infancy and there are still many questions about the stability and application of synaptic learning rules for neural computation. In this thesis, the implementation of an adaptive filter in the brain of a weakly electric fish, A. Leptorhynchus, was studied. It was found to require a cerebellar architecture that could supply independent frequency channels of delayed feedback and multiple burst learning rules that could shape this feedback. This unifies two ideas about the function of the cerebellum that were previously separate: the cerebellum as an adaptive filter and as a generator of precise temporal inputs.

One of the most powerful and important applications that the brain accomplishes is solving the sensory "cocktail party problem:" to adaptively suppress extraneous signals in an environment. Theoretical studies suggest that the solution to the problem involves an adaptive filter, which learns to remove the redundant noise. However, neural learning is also in its infancy and there are still many questions about the stability and application of synaptic learning rules for neural computation. In this thesis, the implementation of an adaptive filter in the brain of a weakly electric fish, A. Leptorhynchus, was studied. It was found to require a cerebellar architecture that could supply independent frequency channels of delayed feedback and multiple burst learning rules that could shape this feedback. This unifies two ideas about the function of the cerebellum that were previously separate: the cerebellum as an adaptive filter and as a generator of precise temporal inputs.

The functionality provided by mobile devices such as cellular phones and tablets continues to increase over the years, with integration of an ever larger number of wireless standards within a given device. In several of these designs, each standard supported by a device requires its own IC receiver to be mounted on the device’s PCB. In multistandard and multimode radios, it is desirable to integrate all receivers onto the same IC as the digital processors for the standards, in order to reduce device cost and size. Ideally all the receivers should also share a single signal chain. Since each standard has its own requirements for linearity and noise figure, and each standard operates at a different RF carrier frequency, implementing such a receiver is very challenging. Such a receiver could be theoretically implemented using a broadband mixing receiver or by direct sampling by a high-speed analog-to-digital converter (ADC). Broadband mixing requires the use of a harmonic rejection mixer (HRM) or tunable band pass filter to remove harmonic mixing effects, which in the past have suffered from a large primary clock tuning range and high power consumption. However, direct sampling of the RF input requires a high-speed ADC with large dynamic range which is typically limited by clock timing skew, clock jitter, or harmonic folding. In this dissertation, techniques for programmable broadband radio receivers are proposed. A local oscillator (LO) synthesis method within HRMs is proposed which reduces the required primary clock tuning range in broadband receivers. The LO synthesis method is implemented in 130-nm CMOS. A clocking technique is introduced within the two-stage HRM, which helps in achieving state-of-the-art harmonic rejection performance without calibration or harmonic filtering. An analog frequency synthesis based broadband channelizer is proposed using the LO synthesis method which is capable of channelizing a broadband input using a single mixing stage and primary clock frequency. A frequency-folded ADC architecture is proposed which enables high-speed sampling with high dynamic range. A receiver based on the frequency-folded ADC architecture is implemented in 65-nm CMOS and achieves a sample rate of 2-GS/s, a mean 49-dB SNDR, and 8.5-dB NF. / text

One of the most powerful and important applications that the brain accomplishes is solving the sensory "cocktail party problem:" to adaptively suppress extraneous signals in an environment. Theoretical studies suggest that the solution to the problem involves an adaptive filter, which learns to remove the redundant noise. However, neural learning is also in its infancy and there are still many questions about the stability and application of synaptic learning rules for neural computation. In this thesis, the implementation of an adaptive filter in the brain of a weakly electric fish, A. Leptorhynchus, was studied. It was found to require a cerebellar architecture that could supply independent frequency channels of delayed feedback and multiple burst learning rules that could shape this feedback. This unifies two ideas about the function of the cerebellum that were previously separate: the cerebellum as an adaptive filter and as a generator of precise temporal inputs.

Digital communication systems which use Orthogonal Frequency Division Multiplexing (OFDM) are now widely used and have many advantages. The main disadvantage is the requirement for highly linear analogue electronics including the high power amplifier (HPA). This requirement cannot be met in all circumstances because of the occurrence of symbols with high peak to average power ratio (PAPR). Such symbols may be non-linearly distorted by limiting. Approaches to solve this problem have been either to reduce the PAPR at the transmitter or to try to mitigate the effect of the non-linearity at the receiver. Soft-limiting, i.e. applying limiting in software prior to the HPA is a simple way to reduce the PAPR. It produces non-linear distortion which will cause an increase in the bit-error-rate (BER) at the receiver. This thesis surveys existing alternatives ways of reducing the effect of non-linearity and proposes some new ones. Two iterative receiver techniques, based on statistical analysis of the nature of the non-linearity, have been implemented and investigated. These are the ‘Bussgang Noise Cancellation’ (BNC) technique and the ‘Decision Aided Reconstruction’ (DAR) techniques. As these techniques are valid for any memory-less nonlinearity, an alternative form of limiting, named as Inverted-Wraparound (IWRAP) has been included in the BNC investigation. A new method is proposed which is capable of correcting the received time-domain samples that are clipped, once they have been identified. This is named the ‘Equation-Method’ and it works by identifying constellation symbols that are likely to be correct at the receiver. If there are a sufficient number of these and they are correctly identified, the FFT may be partitioned to produce a set of equations that may be solved for the clipped time-domain samples. The thesis proposes four enhancements to this new method which improve its effectiveness. It is shown that the best form of this method outperforms conventional techniques especially for severe clipping levels. The performance of these four enhancements is evaluated over channels with additive white Gaussian noise (AWGN) in addition to clipping distortion. A technique based on a ‘margin factor’ is designed to make these methods work more effectively in the presence of AWGN noise. A new combining algorithm referred as ‘HARQ for Clipping’ is presented where soft bit decisions are combined from multiple transmissions. ‘HARQ for Clipping’ has been combined with the best version of the Equation-Method, and the performance of this approach is evaluated in terms of the BER with different levels of AWGN. It has been compared to other approaches from the literature and was found to out-perform the BNC iterative receiver by 3dB at signal to noise ratios around 10dB. Without HARQ, the best version of the Equation-Method performs better than the BNC receiver, at signal-to-nose ratios above about 17dB.

No description available.

One of the most powerful and important applications that the brain accomplishes is solving the sensory "cocktail party problem:" to adaptively suppress extraneous signals in an environment. Theoretical studies suggest that the solution to the problem involves an adaptive filter, which learns to remove the redundant noise. However, neural learning is also in its infancy and there are still many questions about the stability and application of synaptic learning rules for neural computation. In this thesis, the implementation of an adaptive filter in the brain of a weakly electric fish, A. Leptorhynchus, was studied. It was found to require a cerebellar architecture that could supply independent frequency channels of delayed feedback and multiple burst learning rules that could shape this feedback. This unifies two ideas about the function of the cerebellum that were previously separate: the cerebellum as an adaptive filter and as a generator of precise temporal inputs.

This work deals with digital signal processing in the field of adaptive filtering. Fundamental basics of adaptive filtering are described and primary aim is to create executable laboratory examples, using adaptive filtering, in LabView programming language. These laboratory examples are intended to be used by students fo studying and during laboratory lessons. Objective is to connect the examples with external devices, such as microphone. A microphone is used as an user's speech input acquiring interface. In the thesis is depicted Wiener's filter and problem of adaptive filtering is discussed. Contemporary adaptive algorithms are described and their applications as well. Most mentioned is the LMS algorithm and it's forms. Laboratory examples use following concepts: Adaptive Echo Cancellation, Active Noise Control and System Identification. Each of these examples is solely executable (need for LabView or Run-time engine), consisting also of theory with diagrams. Examples therefore are usable even without manual.

No description available.

The use of audio and video communication has increased exponentially over the last decade and has gone from speech over GSM to HD resolution video conference between continents on mobile devices. As the use becomes more widespread the interest in delivering high quality media increases even on devices with limited resources. This includes both development and enhancement of the communication chain but also the topic of objective measurements of the perceived quality. The focus of this thesis work has been to perform enhancement within speech encoding and video decoding, to measure influence factors of audio and video performance, and to build methods to predict the perceived video quality. The audio enhancement part of this thesis addresses the well known problem in the GSM system with an interfering signal generated by the switching nature of TDMA cellular telephony. Two different solutions are given to suppress such interference internally in the mobile handset. The first method involves the use of subtractive noise cancellation employing correlators, the second uses a structure of IIR notch filters. Both solutions use control algorithms based on the state of the communication between the mobile handset and the base station. The video enhancement part presents two post-filters. These two filters are designed to improve visual quality of highly compressed video streams from standard, block-based video codecs by combating both blocking and ringing artifacts. The second post-filter also performs sharpening. The third part addresses the problem of measuring audio and video delay as well as skewness between these, also known as synchronization. This method is a black box technique which enables it to be applied on any audiovisual application, proprietary as well as open standards, and can be run on any platform and over any network connectivity. The last part addresses no-reference (NR) bitstream video quality prediction using features extracted from the coded video stream. Several methods have been used and evaluated: Multiple Linear Regression (MLR), Artificial Neural Network (ANN), and Least Square Support Vector Machines (LS-SVM), showing high correlation with both MOS and objective video assessment methods as PSNR and PEVQ. The impact from temporal, spatial and quantization variations on perceptual video quality has also been addressed, together with the trade off between these, and for this purpose a set of locally conducted subjective experiments were performed.