Towards Automated Configuration of Multi-Sensor Systems Using Evolutionary Computation—A Method and a Case Study

Sensor technology is an active field that enjoys substantial growth and continuously delivers more capable and versatile sensing elements. In particular, the advance of micro- and nanotechnologies provides the leverage for the introduction of powerful novel sensing elements, such as e.g., bio-sensors, as well as integrated multi-sensor systems. However, exploitation of sensors in applications requires sophisticated methods, from conventional signal processing to computational intelligence, to extract the resulting information. Currently, a significant part of a multi-sensor system still has to be assembled manually by experienced designers in a time and a labour consuming process. Clearly, an automatic method for design and configuration of such multi-sensor systems would be salient. In this paper we propose a general methodology for automated design and configuration of multi-sensor systems employing methods from evolutionary computation. The methodology includes feature and classification model optimization. As case study, we investigate a particular gas sensor system, which additionally encompasses issues of sophisticated sensor heating control and has a wide range of applications in nanotechnoly, biotechnology, chemical, and control engineering, environmental protection, etc. The results show, that using the proposed methodology competitive results in terms of classification accuracy and flexibility can be achieved.