Eventtjänst för diagnos av tunga fordon

Abstract: The integration of small computer units in vehicles has made more complex in-vehicle functionality possible. To troubleshoot these more advanced functions, the diagnostic services need to read out more data from the integrated computer units, causing an increased network load. This has caused a demand on more efficient ways of performing vehicle diagnostics. One possible way to decrease the network load is to use an event based service. The aim of this thesis was to investigate the benefit of an event based service within the domain of vehicle diagnostics as well as to present a recommendation of how such a service should be designed. The thesis also aimed at eliciting obstacles and pitfalls connected with the implementation of the service in the current software architecture in heavy duty vehicles. An industrial case study was performed at the Swedish company Scania to elicit the advantages, problems and limitations with an event based service for vehicle diagnostics. First a set of experts representing different domains within vehicle diagnostics were interviewed to investigate the need for an event based service. Requirements were elicited and compared with the event based service ResponseOnEvent defined in the ISO standard 14229-1:2013. A decision was then made to diverge from the standard in order to increase the number of fulfilled requirements and flexibility of the service. A new proprietary service was thus created and evaluated through a proof of concept implementation where a prototype of the service was implemented in two control unit. The prototype implementation of the proprietary service highlighted multiple difficulties connected with the realization of an event based service in the current software architecture. One of the biggest problems was the fact that diagnostic services was assumed to always have a one-to-one relation between request and response, which an event based service would not have. Different workarounds were discovered and assessed. Another problem was the linking between an event triggered response message and the trigger condition. It was concluded that some restrictions would have to be made to facilitate the process of linking a response to its trigger condition. Non-determinism was another problem, since there were no guarantees that an event would not occur too often causing a network overload. In the final recommendation there are suggestions of how to solve these problems and some suggested areas for further research. The thesis argues that there is a need for a new way to diagnose vehicle functionality due to their increased complexity and the limited bandwidth of today’s in-vehicle networks. The event based service ResponseOnEvent offers a good alternative but might lack some key functionality required by an event based service. Therefore it is valuable to consider a proprietary service instead to maximize the benefits of an event based service. Due to its nature, an event based service might require a restructuring of the system architecture and limitations in the hardware might limit the usability and flexibility of the service.