Johannes Pillmann

Car-to-Cloud Communication Traffic Analysis Based on the Common Vehicle Information Model

Although connectivity services have been introduced already today in many of the most recent car models, the potential of vehicles serving as highly mobile sensor platform in the Internet of Things (IoT) has not been sufficiently exploited yet. The European AutoMat project has therefore defined an open Common Vehicle Information Model (CVIM) in combination with a cross-industry, cloud-based big data marketplace. Thereby, vehicle sensor data can be leveraged for the design of entirely new services even beyond traffic-related applications (such as localized weather forecasts). This paper focuses on the prediction of the achievable data rate making use of an analytical model based on empirical measurements. For an in-depth analysis, the CVIM has been integrated in a vehicle traffic simulator to produce CVIM-compliant data streams as a result of the individual behavior of each vehicle (speed, brake activity, steering activity, etc.). In a next step, a simulation of vehicle traffic in a realistically modeled, large-area street network has been used in combination with a cellular Long Term Evolution (LTE) network to determine the cumulated amount of data produced within each network cell. As a result, a new car-to-cloud communication traffic model has been derived, which quantifies the data rate of aggregated car-to-cloud data producible by vehicles depending on the current traffic situations (free flow and traffic jam). The results provide a reference for network planning and resource scheduling for car-to-cloud type services in the context of smart cities.

Ultra-Wideband Aided Precision Parking for Wireless Power Transfer to Electric Vehicles in Real Life Scenarios

Ultra-wideband wireless positioning technologies based on IEEE 802.15.4a have gained attention for various use cases requiring highly precise localization. In this paper an ultra-wideband based approach is proposed and validated for a vehicular use case requiring highly precise local positioning in a dedicated parking lot. For efficient power transfer in wireless charging of electric vehicles scenarios, an accurate alignment of the coils of vehicle and ground is essential. The specific challenge addressed in this paper is that the proposed approach achieves the required lateral accuracy with only two ground- based anchor nodes in combination with one vehicle-based node. For the in-depth validation of the proposed system concept, two experiments are performed: the first one is conducted with an electrical vehicle representing a real park-to-charge-case, whereas the second experiment is executed in a controlled environment to allow for further alignment error analysis. The experiments show, that an error lower than 10cm can be achieved.

Ultra-Wideband Antenna-Induced Error Prediction Using Deep Learning on Channel Response Data

Ultra-wideband wireless positioning technologies based on IEEE 802.15.4a have gained attention for various use-cases requiring highly precise localization. In this paper the orientation dependent characteristics of commonly used ultra-wideband modules are experimentally determined and analyzed. The specific challenge addressed in this paper is the prediction of orientation induced ranging errors through channel response analysis. For the in-depth validation of the proposed methodology, two experiments are performed: the first one is conducted indoors to demonstrate the system behavior in dense multipath environments, whereas the second experiment is executed in an outdoor environment to allow for detailed analysis with as few multipath components as possible. In a second step, a deep learning neural network is applied to the channel response data, showing that the orientation induced ranging error estimate can be improved significantly using the proposed method.

Novel common vehicle information model (CVIM) for future automotive vehicle big data marketplaces

Even though connectivity services have been introduced in many of the most recent car models, access to vehicle data is currently limited due to its proprietary nature. The European project AutoMat has therefore developed an open Marketplace providing a single point of access for brand-independent vehicle data. Thereby, vehicle sensor data can be leveraged for the design and implementation of entirely new services even beyond traffic-related applications (such as hyper-local traffic forecasts). This paper presents the architecture for a Vehicle Big Data Marketplace as enabler of cross-sectorial and innovative vehicle data services. Therefore, the novel Common Vehicle Information Model (CVIM) is defined as an open and harmonized data model, allowing the aggregation of brand-independent and generic data sets. Within this work the realization of a prototype CVIM and Marketplace implementation is presented. The two use-cases of local weather prediction and road quality measurements are introduced to show the applicability of the AutoMat concept and prototype to non-automotive applications.