More electric vehicles are necessary to reach the goals set by international greenhouse gas legislation. However, low range, long charging time, and a sparse infrastructure compared to fuel-based individual mobility reduce customer acceptance of individual electric mobility. Energetic planning functions as range prediction can ease dealing with those drawbacks. These predictive functions need detailed information on energy demand along the route. To enable this prediction, we present a novel analytical method to energetically represent velocity profiles along a map’s link by four parameters. This method has 63 % less error in energetic driving profile representation than other methods. We present a framework to aggregate the four parameters from geolocation records provided by a fleet of different vehicles. We analyze this framework concerning the energetic error introduced to range estimation. The analysis shows that the errors of the framework are lower or of equal magnitude than the ones introduced to range estimation by recent residual high-voltage-battery energy estimation algorithms. A precise route energy demand prediction based on the presented findings is the basis for a more reliable range prediction. This prediction will help to increase the acceptance of individual electric mobility.

Common use cases in the Industrial Internet of Things (IIoT) deploy massive amounts of sensors and actuators that communicate with each other or to a remote cloud. While they form too large and too volatile networks to run on ultra-reliable, time-synchronized low-latency channels, participants still require reliability and latency guaranties. We elaborate this for safety-critical use cases. This paper focuses on the effects of networking protocols for industrial communication services. It analyzes and compares the traditional Message Queuing Telemetry Transport for Sensor Networks (MQTT-SN) with the Constrained Application Protocol (CoAP) as a current IETF recommendation, and also with emerging Information-centric Networking (ICN) approaches, which are ready for deployment. Our findings indicate a rather diverse picture with a large dependence on deployment: Publish-subscribe protocols are more versatile, whereas ICN protocols are more robust in multi-hop environments. MQTT-SN competitively claims resources on congested links, while CoAP politely coexists on the price of its performance.

No abstract available.

This paper investigates the improvements from an intelligent self-adaptive modification to a Global Navigatior Satellite System (GNSS)-Based Rear Axle Kinematic Parametei Estimator (SA-RAKPE) for an automatic-driving-system in a passenger vehicle. The required highly accurate dead-reckoning localization can be achieved by a well-calibrated kinematic odometry model. For this purpose, the presented Extended Kalman filter approach combines a Differential-Velocity system model and a GNSS measurement model. Subsequently the intelligent self-adaptive modifications are introduced to allow the SA-RAKPE to work even under difficult conditions The self-adaptive modifications include a GNSS-Delay-Finder Module that calculates variable delays of the signals used in complex vehicle architectures. The newly developed SA-RAPKE deals with changes in the system and measurement mode accuracies and even works during interruptions caused by GNSS-shortages. To do this, it changes the update equation and fills the interruptions with virtual parameter measurement to avoid estimation inaccuracies from observability loss and even to store the level of learned parameters. After passing the GNSS-shortages, the filter compensates the error in the system model depending on the length of the GNSS-shortage This makes it possible to continue the parameter learning while passing a great number of bad condition passages. This newly developed self-adaptive filter learns the true axle parameter faster than a restartable filter. The results show that despite numerous high-rise zones, tunnels and bridges, outstanding performance and a short learning phase ensue, especially in urban areas.

This paper presents an agile approach to facilitate the rapid development of traffic sign classification algorithms in heavy vehicles under a wide range of visibility conditions. A vision-based traffic sign recognition system makes a significant contribution to improving the transportation safety by enhancing the driver's awareness on important road signs in an automotive cockpit environment. It has therefore been conducive to the ongoing innovation of Advanced Driver Assistance Systems (ADAS) which paves the way to autonomous driving. The paper introduces a real-time framework which can benchmark the performance of the image processing and machine learning algorithms at the electronic system level using an open-loop hardware-in-the-Loop (HiL) simulation. The presented research provides a generative model with real-world datasets to improve the classification performance of machine learning algorithms. A driving scenario considering speed limit traffic signs describes an automatic parameter selection to find the best separating hyperplane for the support vector machine (SVM) classifiers. The framework supports an evolutionary verification process for the traffic sign classification algorithms on the electronic control unit (ECU) in the laboratory.

This paper deals with an advanced approach for the verification of ADAS algorithms in complex automotive sensor networks. We introduce a real-time modular framework that can benchmark the performance of the ADAS algorithms at the electronic system level using a Hardware-in-the-Loop (HiL) co-simulation. The presented research provides a generic architecture of ADAS sensor error injection for robustness testing of the System under Test (SuT). Effects of sensor failures can be added in a continuous manner to achieve increased degrees of realism. For efficient verification, an astute selection of relevant test scenarios is conducted using model-based testing to avoid repetitive test scenarios. The industry-proven framework facilitates a systematic testing of ADAS algorithms precisely and more efficiently on the target electronic control unit (ECU) in the laboratory.

In this paper we present an online learning approach to predict driver behavior and resulting vehicle states. The driver is represented by driver states x→ and a control function fc. Kernel Density Estimation is used for online estimation of current driver states. Data sampling methods are introduced to observe the virtual driver states. The driver states are used as input for the control function to predict resulting vehicle states. To consider environmental influence on driver behavior a context-separated learning approach is presented. The system is tested with real drive test data from different drivers on a specified test route. Different settings regarding learning speed and type of context-separation are investigated. Results show, that consideration of environmental influences on the driver states lead to a better identification of the current behavior but prediction on a longer time horizon does not necessarily improve correspondingly.

Many new research areas such as the Internet-of-Things (IoT) propose the interconnection of every days objects, changing the way how we interact with the environment and our world. One key building block of these new research areas are wireless multi-hop networks (WMHNs). Particularly, providing an energy efficient routing is one of the most challenging topics. Algorithms inspired by behavioral patterns found in nature, such as swarm intelligence algorithms, promise to be capable to handling these past and new challenges. We propose an energy-aware ant routing algorithm based on the foraging behavior of ant colonies. While many energy-efficient ant colony based routing algorithms only consider the energy value of future next hops, we provide a new mechanism for estimating the fitness of a path and energy information dissemination. We study the performance of our new algorithm in simulation in comparison to one of the de-facto standard ant routing algorithms.

Call connection probability in an all-optical network using wavelength division multiplexing (WDM) depends on the number of WDM wavelengths and on the network capability for wavelength conversion at network nodes. New closed-form expressions for call connection probability are derived which reflect these dependencies. These expressions are combined in a converter availability model to study the impact of wavelength conversion on call connection probability and the tradeoff between network wavelength conversion capability and the number of WDM wavelengths.

A flexible model called the group-of-pictures gamma-beta auto-regression (GOP GBAR) model based on gamma-distributed variables is proposed for variable-rate MPEG video sources. This model is an extension of Heyman's (see IEEE/ACM Trans. Networking, vol.5, p.554-60, 1997) GBAR model for video teleconferencing. The GOP GBAR model explicitly accounts for MPEG GOP cyclicity and has convenient analytical properties with easily estimated parameters. Possible extensions of the GOP GBAR model are proposed to capture special features of MPEG video.

A robust generalized likelihood ratio test (GLRT) detector is derived for targets in multiplicative noise with gamma-Weibull statistics. The target is assumed to have a spatial distribution (burstiness) modeled by the incomplete binomial distribution. The GLRT detector is found to be similar in performance to the optimized power law detector.