Easing the Reuse of ML Solutions by Interactive Clustering-based Autotuning in Scientific Applications

Software is disrupting one industry after another. Currently, the automotive industry is under pressure to innovate in the area of software. New, innovative approaches to vehicles and their HW/SW architectures are required and are currently subsumed under the term “SW-defined vehicle”. However, this trend does not stop at the vehicle boundaries, but also includes communication with off-board edge and cloud services. Thinking it through further, this leads to a breakthrough technology we call “Reliable Distributed Systems”, which enables the operation of vehicles where time and safety-critical sensing and computing tasks are no longer tied to the vehicle, but can be shifted into an edge-cloud continuum. This allows a variety of novel applications and functional improvements but also has a tremendous impact on automotive HW/SW architectures and the value chain. Reliable distributed systems are not limited to automotive use cases. The ubiquitous and reliable availability of distributed computing and sensing in real-time enable novel applications and system architectures in a variety of domains: from industrial automation over building automation to consumer robotics. ... mehrHowever, designing reliable distributed systems raises several issues and poses new challenges for edge and cloud computing stacks as well as electronic design automation.