The major challenge vehicle manufacturers are facing is how to produce cars that are feature-rich, cleaner and safer whilst also reducing costs. Much of the product innovation required to meet these objectives will depend upon electronic control. New innovative comfort, safety, economy optimizing or driver assistant functions will require distributed functional processing across multiple electronic control units (ECUs). 

The ECUs are interconnected via common communication networks, used for transferring data and controlling parameters. Most of the system functionalities have timing constraints that must be met in order that the function operates correctly. Often the timing requirements are inherited from the classification of its importance and user measurability.

The biggest challenge in architecting a robust network design is to schedule all communication in accordance with the functional execution within specified times. To meet the cost targets, different network technologies are often implemented to realize the distributed control function, which is adding another level of complexity to design and validation.

Vehicle multiplex networks are a key tool in minimising the complexity and cost impact on vehicle electrical and electronic architectures but also bear the highest risk for system quality and reliability.

Common Network Testing Methods and Tools
According to recent studies, network testing has been identified as an area where improvements are needed due to its direct linkage to costly impact. According to a study by McKinsey, network communications has been identified as the greatest contributing cost source when repairing defects in manufactured vehicles. The study indicated that the reliability of network communication is lacking and is not functioning as required throughout the life of the affected vehicles. This issue is ultimately linked to the control of network quality and reliability, and one primary control mechanism for these measures is network testing.
Despite each vehicle manufacturer,  and linked suppliers, having their own guidelines for quality assurance testing of components, many common network testing methods can be standardised and applied for achieving measurements of system network quality and reliability, leading to more robust networks. 

Bridging the gap to improve system network quality testing at this point is available by measuring different quality metrics.

• A Physical Layer Quality Metric covers how well cabling is completed and what impedance measurements exist at the connectors, stubs and termination points. The physical layer measurement of robustness is calculated based on the collection of all measured electrical characteristics shown to be in accordance with the design specifications.
• A Protocol Quality Metric covers the higher levels of the protocol high lighting problems such as semantics, incorrect packet data, and wrong or missing communication data. The measurement will vary for FlexRay, CAN and LIN as these protocols provide different levels of communication consistency.
• A System Timing Quality Metric covers the timing dependencies for correct system behavior. It will provide information that indicates the robustness of the distributed functions e.g. update intervals of signals or sequences of events are not correctly followed under all conditions. Such timing errors or race conditions lead to very hard to reproduce error situations. From an electrical point of view, the system works perfectly, however, functional errors do still occur and the system may not behave as planned.

These metrics include a level of “goodness”  measured through margins of error that provide a more reliable description about a systems performance than the most often used “pass” or “fail” label.

Physical Layer Quality Metric
The classical way to measure the quality of an electrical signal is through utilising an eye diagram. Oscilloscopes are used to obtain eye diagrams. The different protocol specifications define clear masks on how the eye has to look in order to be compliant with the specification.

Protocol Layer Quality Metric and System Timing Quality Metric
The Vehicle Protocol Test products address the needs of protocol and system validation with different communication characteristics. They are designed to minimise set-up time and reduce test set-up errors by providing a library of pre-defined testing procedures. The library includes testing algorithms needed for the evaluation of a system’s robustness. Each test is dynamically configured automatically according to existing database communication definitions. 
The robustness of LIN and CAN networks is typically determined by the timing characteristics of the communication schedule.  

Roland Jeutter is general manager of the automotive digital test division at Agilent Technologies