Evaluating Data Marshalling Approaches for Embedded Real-Time Systems on the Example of Autonomous Scaled Cars

Abstract: This paper conducted a comparison between four di erent data mar- shaling approaches (Netstrings, LCM, Protobuf, and ROS) that were im- plemented in the OpenDaVINCI environment and evaluated from di erent perspectives. In this study we compared the data marshaling approaches on the low-level, focusing on the overhead in terms of control data and en- coding/decoding for data structures that are common to OpenDaVINCI. Further, the data marshaling approaches were compared in terms of de- scriptive expressiveness with the current interface description language ODVD.