Performance Evaluation of Time-Critical Data Transmission in Automotive Communication Systems
Year: 2009
Type: dissertation
Abstract: For almost 30 years, the airbag has been part of a car’s passive safety components to protect occupants in a crash. In 1980, in the former S-Class from Mercedes-Benz, the system triggered only one driver airbag, though it was composed of 170 parts. Today, highly integrated electronic control units for deployment of front, side or window airbags are standard equipment even in compact cars. At all times, the airbag control was built as an autarkic system for self-contained detection of crash severity and deployment of protection means. With continuing improvements in sensors, electronics and algorithmic concepts for sensor data processing in the airbag control, the time spans between impact and airbag deployment could be reduced to about 30 milliseconds. Every millisecond in reaction time which can be saved, directly contributes to a reduction of injury severity for the occupants. Thus, even shorter time spans for deployment would be desirable. However, the performance of passive safety systems, which react on a collision, and of constructive means, in terms of increased stiffness of the vehicle body, is widely exhausted. The networking and interplay of passive and active safety components, like the electronic stability program or the radar-based headway control, appears to be a promising instrument to compensate for this. Active safety components engage actively into the vehicle dynamics in case of a dangerous driving situation in order to avoid a crash, e.g. if the car turns into a side-slip or the distance to a vehicle in front falls below a certain threshold. If sensor data, which are computed by active safety components in a dangerous driving situation, are communicated to the airbag control, it will be able to react faster and more effective in case a crash cannot be avoided in the following. For example, previous knowledge of a risky situation might be used to lower algorithmic thresholds for airbag deployment in order to achieve a shorter reaction time in case of emergency compared to a purely passive reaction on the impact. Such a networking of active and passive safety has to be performed using the in-car communication infrastructure, consisting of bus systems like CAN or FlexRay and gateways to interconnect single bus segments. Real-time capabilities and timeliness of data transfer – which means data arrives at the receiver within a predefined time interval – are of essential importance for the effectiveness of networked applications for occupant protection. Otherwise, the crash might occur before the passive components were able to receive and process sensor data from active safety systems. For development of networked, highly time-critical vehicle safety systems, sophisticated evaluation of data transmission is indispensable at an early stage of system design. In this dissertation, concepts and methods are presented to conduct comprehensive performance evaluation studies of in-car communication and to support efficient safety systems in future vehicles. A hardware- and software-based measurement infrastructure was built, which allows to figure out effects of data transmission in normal operation mode, the so-called use-case, by evaluation of recorded communication data and to derive important system parameters from the samples. These parameters serve as realistic input for a system model, which resembles the functional and timing behavior of relevant system components using UML-based statecharts and discrete event simulation. The simulation allows for capturing various communication scenarios and for obtaining meaningful performance measures, like end-to-end transmission delays. For the field of time-critical safety applications, measures for the worst-case of system operation are also vitally important. Therefore, the analytical method of Network Calculus was extended to meet the requirements of automotive communication systems and applied to this area for the first time ever. The analytic results resemble guaranteed upper bounds for delays in data transmission and allow, together with the outcomes from real-life measurements and from use-case simulation, for a distinct evaluation of the networked system in early design phases. Performance measures could be obtained for both the use-case and the worst-case of system operation, supporting a comprehensive evaluation, which can be adapted to the actual criticality of the application and leads to a significant increase in performance of future networked vehicle safety systems.
Author: Herpel, Thomas
Language: German
Topic: Real-Time Systems Scheduling
Subfield: Hardware and Architecture
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2404887322