Current advances in Artificial Intelligence (AI) technologies pave the way to consider new services that assist aircrew, possibly in embedded systems. Semantic reasoning using ontologies provide both opportunities and challenges for these services. Recent studies showcase that those reasoning approaches suffer from long and unpredictable execution times, and high memory consumption. Such limitations currently refrain the use of this approach in embedded systems. The objective of this work is to explore ways to deploy such reasoning in embedded architectures focusing on optimisations of time/memory envelope and benchmarking. We first explore the related work on existing optimisations of some implementations to get a clear view on reasoning techniques, then we investigate the variability sources to get a clear view on what is going inside a reasoner. The results are (i) a better overview of reasoner methodologies, (ii) a notable categorisation of optimisation families and (iii) a clearer view of impacts concerning reasoning time determinism.