Jakub St'astny

Modelling and Recognition of Movement Related EEG Signal

Our previous study was aimed at the classification of right index finger movement direction by means of the movement-related EEG signal. The EEG database we used was originally recorded for a physiological research; from our point of view it has one significant drawback: there is no continuous non-movement related (resting) EEG of sufficient length (>10 sec) in the database. To overcome this limitation we decided to generate artificial resting EEG signal. This article describes method and process of non-movement related (resting) EEG signal generation along with the reached new classification results including the false alarm rate. Artificial EEG signal was generated by AR modelling. The AR model parameters were estimated from short segments (3 sec) of resting EEG already present in the database using the autocorrelation method. Owing to large intra-and inter-personal variability one set of parameters had to be estimated for each person and electrode. The artificial EEG was used for the classification with satisfactory classification results.

Subband optimization for EEG-based classification of movements of the same limb

The contribution investigates the impact of frequency feature optimization on discriminating between movement-related EEG realisations associated with right shoulder elevation and right index finger flexion movements. Exhaustive search of subbands in the range from 5 to 45 Hz is performed. A classifier based on Hidden Markov Models is utilised. The results show a large variability of optimal settings among subjects and electrodes. Using subband optimization an average 3.5% increase in classification accuracy of EEG filtered using 8-neighbor Laplacian filter was achieved, reaching an overall score of 81.2±1.2%, individual improvements ranging from 1.2 to 9.9%. The best general setting common for all subject was confirmed as 5-40 Hz.

Stochastic arithmetic complex number operators

This work presents a simple study of stochastic arithmetic complex number operators for addition and multiplication. Their usage is demonstrated by design of a sum of product circuit As the stochastic complex number operators need more control random streams than stochastic rational number operators, we optimized the number of random generators used in the real circuit. In the end our sum of product circuit contains two LFSRs thus we analyzed the impact of the choice of the seeds for the LFSRs on the quality of the calculated results. By using exhaustive search over the LFSR state space we were able to reduce the output RMSE by 34% in comparison to choice of the equally spaced seeds over the LFSR state space.

Movement EEG classification using parallel Hidden Markov Models

In this contribution we examine the use and utility of parallel HMM classification in single-trial movement-EEG classification of index finger reaching and grasping movement. Parallel HMMs allow us to easily utilize the information contained in multiple channels. Using HMM classifier output in parallel from examined EEG channels we have been able to achieve as good a classification score as with single electrode results, further we do not rely on a single electrode giving persistently good results. Our parallel approach has the added benefit of not having to rely on small inter-session variability as it gives very good results with fewer classifier parameters being optimized. Without any classification optimization we can get a score improvement of 11.2% against randomly selected physiologically relevant electrode. If we use subject specific information we can further improve on the reference score by 1%, achieving a classification score of 84.2±0.7%.

Programmable PWM modulator optimized for high speed for OPWM test platform

Optimal Pulse Width Modulation (OPWM) is an established technique to generate PWM waveforms with low base-band distortion. The technique requires fast PWM generator to minimize base-band distortion. The aim of this paper is to present a programmable PWM modulator optimized for high speed satisfying the OPWM technique demands. The PWM modulator stores pulse sequence data in an internal memory which is facilitated by SDRAM memory. The design was implemented on Register Transfer Level (RTL) using VHDL language. The design verification was conducted on both RTL and gate levels as well as tested on two development boards.

Filter Hardware Cost Reduction by Means of Error Feedback

The article presents an uncommon application of the error feedback-improved IIR filter. A simple method to reduce the hardware cost (silicon area) of the biquadratic section implementation by means of error feedback (EF) is described. The optimization method utilizes the fact that the filter with an EF is more resistant to roundoff noise than a filter without it. An iterative method is used to reduce the occupied silicon area. First the standard IIR filter is designed with the requested quantization properties. Then the EF-improved biquadratic section is designed to attain the same roundoff noise properties. The occupied silicon areas of both solutions are compared then. Although implementation of EF results in more arithmetic components and more complex filter control, the resulting structure attaining the same quantization noise is smaller under defined circumstances (filter with poles close to the unit circle). Results show it is possible to spare up to 22% of the occupied silicon area. Our findings are valid for FPGA as well as ASIC implementation of the IIR filters. Our method has an advantage in using a standard and already verified filtering IP core which results in design time reduction.