I work in the smart agriculture project WeideInsight in which we develop an energy-efficient localization system for cattle’s in an indoor (barn) and outdoor (pasture) environment. We are using the technologies Bluetooth Low Energy (BLE) and Mioty, a Low Power Wide Area Network (LPWAN) technique. We do not use GPS and we are aware that we do not have exact coordinates the whole time. My personal research goal is to develop a hybrid location model that returns the most meaningful location information. This can either be a symbolic Point of Interest (POI), a relation between different animals (social groups/activities) or geometric coordinate based on multilateration.
To optimize the process I want to inform you about my personal application process. You find a list of available theses in the next section. Pick one topic and write me an abstract about one or two pages. This helps you to get familiar with the topic and you can check if you like the topic or not. If there no interesting topics for you currently available you can also pitch your own idea and write your proposal.
List of free theses. Theses that are currently in progress are listed with a remark that you can think about individual theses.
Evaluation of beacon data
In various experiments, we gathered sensor and ground truth data. Now we need methods to compare our BLE sensor data with the camera observations of the barn. The challenge in this theses is to use map mathching to compare observations and raw data. Research question is here to develop a suitable testing method.
Development of a player for location data
We visualize location data with the system of a project partner. This system is optimized for live data. Although it is possible to visualize historical data, changing it is impossible. For this, we need our own system to manipulate the data. The player should have a connection to our partners product via API. Research questions here are a bit dependent on the further direction of this topic.
Literature surveys are possible topics. approaches like kitchenham are highly suggested to design a survey. The topics should be optimized for our smart aggriculture scenario. Possible topics here are
- Map Matching
- Information Visualization
A further iteration of a data generator for smart spaces and agriculture projects
In the first theses, we investigated whether it is possible to use the SmartSpec framework to generate location data for our scenario. Our implementation provided an initial prototype for the simulator with still limited capabilities, which do not fulfill all requirements just yet. Nevertheless, we were able to generate some ble location data inside the barn. We hinted towards several possible improvements, which are partially summarised here:
- Ability to configure farm layout, sensors and spaces using the map in the Web UI instead of JSON files
- Connection to our Backend in the project to pull learning data directly
- Data generation outside on the pasture with LPWAN
Filter for location beacon data
Our lateration approach to determine the indoor location based on beacon data is capable for errors. Beacon data as well as labeled ground truth data is available. Goal of the theses is to use filters to reduce noise and other errors.
Suitable as bachelor and master theses. The master requirement is the usage of a particle filter and bachelor students are free to choose.
Data Generator for smart spaces and agriculture projects
The percom best paper award 2022 provides an approach to simulate smart spaces. In these master theses, the first step would be to rebuild the models to generate data. Goal of the theses is then to ably this approach to the WeideInsight scenario to generate data.
Code Refactoring for environmental sensors
We are using environmental sensors to measure the air quality in Bamberg. For these sensors we are also using the LPWAN technology Mioty to transfer data. The C++ Code is not that efficient and need improvement. This thesis is more suitable as Bachelor theses. In master theses, there is the design of an experiment needed. Priority of this topic is also not that high.
Topic is still here, but just for experts with high experience.