Runtime Assessment of the Parameter Utilisation in Adaptive Systems
Martin Goller, Sven Tomforde · 2022 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops) · 2022
A pervasive computing system that provides users with a specific functionality often needs to adapt to dynamic conditions and changing requirements. In particular, if it is equipped with autonomous learning capabilities and consists of a large number of individual units, this leads to considerable complexity in monitoring and controlling the system’s behaviour at the macro level. Building on recent work on a measurement and evaluation framework for properties of such systems, in this paper we introduce a new metric for quantifying the parameter utilisation. The idea is to monitor the degree of freedom used and thus be able to make statements about the expected system behaviour. We analyse the characteristics of this metric using two simulated scenarios from the fields of surveillance networks and traffic control.