Olga Saukh

Assistant Professor at TU Graz, Institute of Technical Informatics (ITI)
Resident scientist at Complexity Science Hub Vienna (CSH Vienna)

Research Projects

This page lists funded research projects I have been working on since my Ph.D. years. The projects are listed chronologically. I also spontaneously work on side projects that deviate from my main research line but that I find fun. Please see the research page for more details.

D4Dairy: Digitalisation, Data integration, Detection and Decision Support in Dairying

Duration: 2018-2022, I'm a PI, a key researcher and an area manager in this project

D4Dairy project addresses the challenges of the stakeholders along the dairy value chain, in particular of the farmers and the economic partners contributing to this project. The project is funded by the FFG (Austrian Research Promotion Agency). The overall goal of D4Dairy is the generation of added value for herd management as well as the improvement of animal health, animal welfare and product quality by creating a well-developed (data) network and by exploiting the opportunities offered by new (digital) technologies and analytical methods.

The specific objectives of D4Dairy therefore are a) to capture the enormous amounts of diverse data (theoretically) available on the farm and from other partners along the milk chain; b) to aggregate these data into one central database, assess different data communication methods in compliance with legal requirements and develop a concept of interoperability; (c) to perform complex and advanced analyses in order to detect risk factors and identify early predictors of health problems using big data approaches, mid-infrared spectra, genetic and genomic studies, mycotoxin detection and information about the impact of housing climate on animal health and welfare; (d) to develop data-based strategies to reduce the use of antimicrobials and implement quality assurance programs and (e) to provide the information obtained from the analyses for decision support using newly developed complex and innovative tools that are easy to apply, operate in real-time in an automated fashion, and whose results are easy to interpret.

OpenSense II: Crowdsourcing High-Resolution Air Quality Sensing

Duration: 2014-2016, I led Zurich part of the project

OpenSense II explores the dimension of crowdsourcing and human-centric computing in measuring and validating air pollution data. We study possibilities to incentivize users to make available states based on physical measurements, such as location, motion and pollution, through their mobile personal devices or monitoring assets that they can install in their homes or on their cars. The project leverages the previous research projects results to improve the quality of gathered data and push spatial resolution of constructed models even further.

  • Our IPSN'15 paper proves that Geometric Mean Regression (GMR) used to calibrate gas sensors over multiple hops yields very low error accumulation.
  • Journal paper describing SCAN, a generalization of GMR to arrays of cross-sensitive sensors, published by ACM IMWUT'17.

inUse: Increasing Usability of Sensor Generated Data

Duration: 2013-2014, I was a co-PI in this project

In order to make sensor data useful, despite the lack of expert supervision in the loop, context annotations, analysis and modeling become key components in setting up sensor data-based applications: only if sensors and sensor data are annotated and enriched by information describing their meanings, quality, validity scope, measurement procedure, and connections with closely correlated data, they can be understood and used by the general public. The kind of useful context ranges from sensor and measurement descriptions (time, location, sensor type, validity, time of last calibration, measurement quality, etc.) to advanced context derived by aggregating, combining, analyzing, and enriching raw data, e.g., in the form of analytical models, annotations, and correlations.

inUse demonstrates the usability of air pollution data currently being gathered by the OpenSense network in Zurich. This data is used to calculate pollution maps with a higher temporal and spatial resolution than available from state-of-the-art maps based on the data from conventional measurement stations.

OpenSense: Open Sensor Network for Air Quality Monitoring

Duration: 2010-2013, I led Zurich part of the project

OpenSense is a research project dedicated to monitoring air quality in urban areas with mobile wireless sensor nodes to better understand spatial variation of main air pollutants in cities. Several groups from ETH Zurich and EPF Lausanne are involved in the project. Our group at ETH Zurich successfully operated a network of ten OpenSense stations installed on top of VBZ trams in Zurich to measure urban air quality. In particular, we measured O3, NO2, CO, and ultra-fine particles along with temperature and humidity. Please visit the OpenSense Zurich project page for more details on our activities in the project. The project is funded by Nano-Tera.ch.

There are many research challenges directly and indirectly related to setting up and operating a network of low-cost sensors on top of trams. For example, how to select a set of trams for installing the stations, to quantify the impact of mobility on sensor measurements and how to do the sensor calibration in the mobile setting.

AWARE: Autonomous Operation of Wireless Sensor-Actuator Networks

I took part in the execution of AWARE project as a Ph.D. student at the University of Bonn, Germany. The goal was to design, develop and experiment with a platform providing the middleware and the functionalities required for the cooperation among aerial flying objects, i.e. autonomous helicopters, and a ground sensor-actuator wireless network, including mobile nodes carried by people and vehicles. The platform enables the operation in sites with difficult access and without communication infrastructure. In order to verify the success in reaching the objectives, the project considers the validation in two different applications: civil security / disaster management and filming dynamically evolving scenes with mobile objects. Three general experiments in a common scenario have been conducted in order to integrate the system and test the functionalities required for the above validations.

Structural Health Monitoring with Wireless Sensor Networks

During my Ph.D. and shortly after I was working on a couple of projects dedicated to structural health monitoring with WSN technology. In civil engineering practice, monitoring of civil structures had always been done with conventional wired systems. These mature systems combine high-fidelity sensor values with a reliable and robust system performance. However, the installation, primarily the cabling, is very time-consuming and expensive. Many application scenarios benefit from using wireless sensor networks due to their attractive properties of being cable-free and easy to deploy, which allows minimizing installation cost and time. Although WSN technology allows considerable cost reduction, actual deployments of wireless systems are still very limited due to numerous challenges that need to be solved: handling reliably high sampling rates and large data volumes, assuring system reliability, and achieving an acceptable system lifetime. Moreover, WSNs still need to offer enough transparency for civil engineers to concentrate on assessing the state of the structure by easily tuning acquisition parameters and reconfiguring parts of the measurement system.