Review of "Viewpoints and Views in Hardware Platform Modeling for Safe Deployment"
Suggestion for acceptance
Strongly accept. I think this was one of the most interesting papers so far. It reached into two of my biggest areas of interest: automotive software/hardware using the CAN bus, and cyber-physical systems. Very cool.
Summary of the paper
Collaboration between cyber-physical systems is very complicated. This paper looks specifically at the hardware platforms in modern cars. At the time of writing (2014), there were no sufficient methods for defining a hardware platform that addresses different stakeholder's concerns. Describing a suitable hardware platform for the deployment of safety-critical software components is a difficult task. This paper outlines viewpoints for stakeholder's concerns and provides a method for multi-view modeling of hardware platforms. To test their method, they used an Arduino-based cooperative adaptive cruise control system (CACC). CACC systems adjust the car's speed by taking into account the car's distance from the car in front of it, and messages from the cars ahead of it on the road.
Positive points
This paper accomplished its goal to describe the viewpoints for stakeholder's concerns in hardware modeling for the safe deployment of critical software. They created a language for multi-view modeling that refers to viewpoints. My favorite contribution was the way they modeled an Arduino-based car. With more research on larger test cases, I believe their language can scale up to be very useful for designers in the industry.
Negative points
Their current model does not support the allocation of software components to hardware platform instances. They showed the diagrams that they were able to create with their Hardware Platform Description Language (HPDL), but they didn't show any of the actual code. Maybe they thought it was obvious for people who are familiar with UML? They say that their syntax and semantics can be found in a separate report that they included in their references. I guess this is ok, but at least one or two code snippets would have been interesting.
Potential future work
As mentioned earlier, their current model does not support the allocation of software components to hardware platform instances. In the future, they would like to provide a meta-model that supports this allocation. They also want to extend HPDL to couple it with analysis techniques. They also require further testing on larger test cases. User acceptance tests could also be performed so they can receive valuable feedback on HPDL.