C4.4 - Cost Modeling of Radio Sensor Systems for Industrial Applications
- SENSOR+TEST Conferences 2011
2011-06-07 - 2011-06-09
- Proceedings SENSOR 2011
- C4 - Sensor Electronic I
- M. Niedermayer - Fraunhofer Institut (IZM), Berlin (Germany)
- 445 - 450
The implementation of wireless sensor networks in complex machinery and plants paves the way for advanced concepts of condition monitoring and remote maintenance. Smallest irregularities on rolling bearings and gears can be recognized by sensitive vibration sensors to detect the damage status. Radio sensor systems provide cost-efficient solutions in broad range of industrial application ranging from process monitoring to material logistics.
Radio sensor systems exhibit a lot of design trade-offs regarding system behaviour, hardware architecture and fabrication technology. Conventional design flows typically allow cost evaluations only at the end of the development process when component geometries and the technology choices are fully specified. Whereas, an efficient cost minimization requires a systematic approach which guides the designer to optimize the cost-relevant elements during the whole design process. Hence, a new design methodology for cost-efficient radio sensor systems is introduced. The suggested flexible design approach is based on cost models. The optimization focus is dynamically set by the costdriving components and processes. The corresponding cost models and implemented tools are described. The abstraction level of the cost modeling depends on the stage of the design process according to the necessary estimation accuracy. Initially, the relevant cost trade-offs are modeled and parameterized on a high abstraction level. Some fundamental cost relationships are discussed for this purpose. During the design process the cost models will be refined as required by the complexity of the component choice and the technology selection.
For the verification of the implemented design platform, a couple of prototypes based on different architectures are presented. The cost-optimal implementations are illustrated for several application scenarios. Valuable insights were gained from the example of a sensor network for condition monitoring in paper plants. Important conclusions were drawn regarding which technology alternatives are particularly suitable for the cost situation of today. The fundamental influence of the device quantity on the technology choice is shown. Cost reduction strategies were derived for universal and modular sensor nodes in moderate quantities for niche applications. Furthermore, the current cost boundaries for very simple sensor network architectures have been determined. Finally, a perspective of our future work will be given.