Dutch Ultra Low Power Conference

The medicine of the future you’ll need to take only once, and it’s a bioelectronic one

Wouter Serdijn

The Dutch Ultra Low Power Conference brings together Belgian and Dutch professionals and companies involved in the development and application of devices with ultra low power technologies. It targets engineers, designers and technical managers in the advanced field of energy harvesting and ultra low power and energy-efficient designs. The keynote will be given by Wouter Serdijn, professor of bioelectronics at Delft University of Technology.

Additional information ...

MSc CE Thesis Presentation

Energy Efficient Feature Extraction for Single-Lead ECG Classification Based On Spiking Neural Networks

Eralp Kolagasioglu

Cardiovascular diseases are the leading cause of death in the developed world. Preventing these deaths, require long term monitoring and manual inspection of ECG signals, which is a very time consuming process. Consequently, a wearable system that can automatically categorize beats is essential.

Neuromorphic machines have been introduced relatively recently in the science community. The aim of these machines is to emulate the brain. Their low power design makes them an optimal choice for a low power wearable ECG classifier.

As features are crucial in any machine learning system, this thesis aims at proposing an energy efficient feature extraction algorithm for ECG arrhythmia classification using neuromorphic machines. The feature extraction algorithm proposed in this thesis consists of the merger of a low power feature detection and a feature selection algorithm. Also, different network configurations have been investigated to achieve classification using an LSM architecture. The resulting system can accurately cluster seven beat types, has an overall classification rate of 95.5%, and consumes an estimate of 803.62 nW.

MSc SS Thesis Presentation

The cocktail party problem: GSVD-beamformers in reverberant environments

Derk-Jan Hulsinga

Hearing aids as a form of audio preprocessing is increasingly common in everyday life. The goal of this thesis is to implement a blind approach to the cocktail party problem and challenge some of the regular assumptions made in literature. We approach the problem as wideband FD-BSS. From this field of research, the common assumption of continuous activity is dropped. Instead a number of users detection is implemented as a preprocessing step and ensure the appropriate number of demixing vectors for each time frequency bin. The validity of the standard mixing model used for STFT’s is challenged by looking at the response of a linear array.

Source separation is achieved by demixing vectors based on the GSVD, derived in a model-based approach. While most permutation solvers offer an a posteriori solution for all users, we looked at finding local solutions for a single user. Combining this with the user identification called the alignment step, we conclude that the permutation problem can be reduced to selecting a demixing vector for each discrete time-frequency instance. The correlation coefficient proves to be a sufficient metric to couple reconstructions to the original data as it selects most of the active time-frequency bins.

In simulations, our demixing vectors achieve comparable inteligibility, measured by STOI, as the compared techniques and it is more robust against smaller sample sizes than the theoretically SINR optimal MVDR.

Additional information ...

MSc TC Thesis Presentation

Blind Signal Identification

Dennis van der Geest

The capability to efficiently find signals of interest in a very dense electromagnetic spectrum is becoming increasingly important with the continuous increase in spectrum usage. In this research project, methods are developed to identify communication signals by estimating signal features (symbol rate, modulation scheme, etc.) in the absence of a-priori knowledge, i.e. blind. By modelling the received communication signal both as a stationary and a cyclostationary process, various feature estimation methods are evaluated based on their computational complexity, their estimation accuracy and their robustness in the presence of signal contamination, such as frequency offsets. By efficiently combining various estimation methods, a signal classification algorithm is derived which is aimed to provide an optimal tradeoff between computational complexity and classification performance.

Additional information ...

Signal Processing Seminar

An introduction to distributed signal processing

Richard Heusdens

Due to the explosion in size and complexity of modern data sets, it is increasingly important to be able to solve problems with a very large number of features or training examples. In industry, this trend has been referred to as ‘Big Data’, and it has had a significant impact in areas as varied as artificial intelligence, internet applications, computational biology, medicine, finance, marketing, journalism, network analysis, weather forecast, telecommunication, and logistics. As a result, both the decentralized collection or storage of these data sets as well as accompanying distributed solution methods are either necessary or at least highly desirable. In this talk, we will give an introduction to the design of distributed algorithms. We will discuss the basic requirements of these algorithms, like being simple, resource efficient, scalable, robust against changes in network topology, asynchronous, etc. We will demonstrate the design of such algorithm by considering the example of distributed averaging in a sensor network.

Additional information ...

Signal Processing Seminar

Synchronization for underwater communications based on dual Zadoff-Chu sequences

Yiyin Wang

Abstract: Underwater acoustic channels are not only characterized by multipath propagation but also by Doppler scaling effects. These characteristics challenge the preliminary tasks of an acoustic receiver, such as timing and frequency synchronization, and Doppler scale and channel estimation. In this talk, we propose a novel preamble design based on a dual Zadoff-Chu (ZC) sequence. With the help of the well design preamble, a cyclic feature based detector is developed to bypass the requirement of channel statistic information. The Doppler scale estimation is simplified as the frequency estimation adopting the ESPRIT type algorithm. Furthermore, the special structure of the preamble facilitates the estimation of the residual carrier frequency offset (CFO), and the good correlation properties of the preamble enable a low-cost channel estimation. Therefore, with a single preamble, multiple preliminary tasks of the receiver are accomplished. Simulation results indicate the superior performance of the proposed methods.

Additional information ...