In the automotive industry, a steadily growing number of mono-functional electronic control units (ECUs) with increasing complexity on the one hand and restrictive requirements for power consumption and mounting space on the other hand are forcing changes in car electronics. This tendency is enforced by increasing requirements on security, comfort, fuel consumption and emission. In luxury cars, we are now at up to 80 more or less independent electronic units with low potential for synergies between functions (pan-functional services). The actual assembly of electronic units is certainly easily expandable, has very low error propagation but it also involves lots of logistic and bulky cabling with unwanted weight as well as extensive space and power consumption.
If trends are properly interpreted, synergies between functions for more comfort, additional safety and security as well as minor air pollution are required in the future at least at unchanged costs in the vehicle's electronics. To be cost-effective this requires as far as possible software implementation. Therefore, the current multi-unit electronics solution might be replaced in a long-term range by a fully networked client/server-architecture built up from component technology and an IT conform layering structure with basic functions, middleware and applications.
The client/server-architecture uses the clients for real-time and safety features and to provide limp-home functionality; the server(s) implement(s) higher level services, i.e. pan-functional services, telematic services and a common user interface. An increasing number (depending on server capabilities) of the former non-real-time control units can be totally replaced by pure software solutions on the server(s). This new electronic structure promises easy installation of new functions and additional pan-functional services on a settled architecture and opens an attractive after sales market. On the other hand, the requirement for a settled new electronics architecture introduces an ambitious systems-engineering challenge to prepare an integration concept based on a commonly agreed Reference Architecture in the automotive industry.
In this paper, the proposed client/server architecture and corresponding software architecture will be described in detail and the distribution of functions will be discussed. The problems coming up with a possible new electronics architecture will be addressed and proposals to overcome them will be given.