NeXOS contribution to the adaptation of system analysis engineering tools for mature and reliable ocean sensors

Oceanography was started by Navy engineers and the references of readiness and functional specifications were military. Since the end of the 80s, a new generation of instruments was able to promote more cost-efficient technical solutions. They cover now the needs of scientific ocean research as well as operational oceanography and environmental monitoring or assessment of the coastal areas. The ambition in the EC FP7 NeXOS project is to proceed in this direction in order to improve the temporal and spatial coverage, resolution and quality of marine observations. The Technology Readiness Levels are now successfully used for oceanographic equipments. NeXOS promotes a specific approach for the sensors themselves and for sensor systems. It happens to be very useful to detect weak points both at the beginning of the development and at high level of maturity. Some criteria are more often weak in the ocean sensor development world such as: follow-up of cost drivers at an early stage of the design, exact scope of the market, safety, dependence on few component providers, etc. The practice of functional analysis of sensor systems shows also a need to focus on specific aspects. Marine environment constraints are known to be critical. The designer has to take into account surrounding functions dealing with data availability, interoperability, modularity, robustness which are in fact major objectives of the NeXOS project. Reliability analysis in the context of marine sensor systems is in many cases a key issue. Some sensors will be deployed for long term autonomous missions, some of them, for instance on-board Argo Floats, will never be recovered. It then needs to be very performed with the rather small amount of failure rated available. The fear events are not only coming from the operations at sea but also from several steps of the data dissemination process: metrology, associated metadata, processing, etc. In order to achieve this goal, is necessary to consider several alte- native configurations of the system design in such a way that functional specifications remain unchanged but enhance dependability. This is framed in the so-called reliability allocation problems [1], usually addressed by firstly obtaining Fault Tree models of the system and then performing cost-constrained optimization of whole system reliability. The most common criteria used to overcome reliability issues consist in apply redundancy on critical components to provide backup in case of failure of some component, use diversity (i.e. components from different manufacturers) in redundant parts so as to avoid common cause failures and employ physical dispersion (i.e in a redundant configuration, locate components in different parts of the system).