The respective effect of average incorporated ion energy and impinging atomic oxygen flux on the deposition of silicon oxide (SiOx) barrier coatings for polymers is studied in a microwave driven low pressure discharge with additional variable RF bias. Under consideration of plasma parameters, bias voltage, film density, chemical composition and particle fluxes, both are determined relative to the effective flux of Si atoms contributing to film growth. Subsequently, a correlation with barrier performance and chemical structure is achieved by measuring the oxygen transmission rate (OTR) and by performing x-ray photoelectron spectroscopy. It is observed that an increase in incorporated energy to 160 eV per deposited Si atom result in an enhanced cross-linking of the SiOx network and, therefore, an improved barrier performance by almost two orders of magnitude. Furthermore, independently increasing the number of oxygen atoms to 10 500 per deposited Si atom also lead to a comparable barrier improvement by an enhanced cross-linking.

Estimating the positions of multiple speakers can be helpful for tasks like automatic speech recognition or speaker diarization. Both applications benefit from a known speaker position when, for instance, applying beamforming or assigning unique speaker identities. Recently, several approaches utilizing acoustic signals augmented with visual data have been proposed for this task. However, both the acoustic and the visual modality may be corrupted in specific spatial regions, for instance due to poor lighting conditions or to the presence of background noise. This paper proposes a novel audiovisual data fusion framework for speaker localization by assigning individual dynamic stream weights to specific regions in the localization space. This fusion is achieved via a neural network, which combines the predictions of individual audio and video trackers based on their time-and location-dependent reliability. A performance evaluation using audiovisual recordings yields promising results, with the proposed fusion approach outperforming all baseline models.

Smart intelligent sensor systems are key for many application areas in the domain of ubiquitous computing like embedded autonomous systems, wireless sensor networks, internet of things and wearables. In this article, we provide an overview of recent approaches to improve the design, structure and deployment of machine learning methods for smart sensor systems in order to make them as lightweight and energy efficient as possible. The techniques considered cover both hardware and software perspectives.

No abstract

No abstract