One of the benefits expected by digitalization of production environments is a tremendous increase in flexibility. The vision is a volatile Industrial Internet of Things requiring only marginal engineering and installation effort.
Localization will become a key technology reducing engineering efforts but also giving a device location awareness, i.e., to assign its data to a context within a factory environment. This applies both to mobile devices such as automated guided vehicles (AGVs) or product identifiers as well as machinery and machine components, which will face the need to be (physically) reconfigured and frequently moved during their life-time.
Today’s localization algorithms rely on a perfect omni-directional antenna characteristic. In opposite, directional antennas offer great benefits in suppressing disturbances, increasing communication ranges and reducing energy consumption for constricted devices. In addition, many real omni-directional antennas (especially on integrated circuit boards) have a certain directional effect, which reduces the localization accuracy.
The research project PinpointIoT investigates novel localization algorithms that can perform localization using directional antennas. Two main challenges have to be faced to achieve the project goal: 1.) The ambiguity of received signal strength (RSS) values of directed antennas that an object can be either far away and in the main lobe of the antenna or close and outside the main direction needs to be solved. 2.) Data of antennas located at different positions at the factory floor needs to be fused to retrieve a high localization precision. The basic idea is to iteratively find a position estimate that matches the received RSS values at all directed antennas best. I.e., the error between a set of sampling points in the solution area and the measured RSS value is calculated and from the result a gradient field is generated that can be followed to the global error minima giving the positon of the device.
As the methodology, an iterative approach is selected that initially identifies overlapping reception areas and continuously increases the location precision and accuracy. By performing clustering of nodes, including quality parameters such as jitter or other variance metrics, information is additionally weighted to achieve a higher localization precision and robustness. In this way the project can address needs for both, quick decisions by providing rough position information and high location accuracy by continuous refinement, if required.
Investigations will focus on the development of such algorithms but also their ability to mitigate multipath propagation, shadowing and reflection, being main disturbances in today’s localization schemes. Simulation tools and finally an experimental laboratory validation will be base instruments for validation.
