Synopsis
Microsensors are particularly buoyant sector in the industry of man-made complex machines. Traditionally, the main sensor requirements (linearly transferred from the macro sensors industry to the micromachining technologies) were in terms of metrological performance, i.e. the (most often) electrical signal produced by the sensor needed to match relatively accurately the measurand. Such basic sensor functionality is no longer sufficient. The nature of industry demand, and therefore the research goals of the sensing community are presently shifting, away from aiming to design perfect mono-function transducers towards the utilization MEMS based sensors as system components. A new set of requirements for sensing systems and more generally for measurement systems is therefore being generated. Such requirements ultimately imply that components are enhanced with increasingly autonomous functional capabilities. It is here, in the area of data processing and extraction of information, that the author proposes to situate the core of the tutorial, expanding both ways:
- towards the sensing devices themselves and the MEMS technology which enables their production and
- towards the application end of the enhanced sensing systems.

The presentation clarifies the strands of development in sensing, some of which are linked with the industry demand for “replacement products” ( process/instrumentation sensors designed for high accuracy or cheap/minimum size& weight/minimal electronics sensors for liberal use in appliances and automotive industry for example), whilst other strands are under development either to enable new applications or to support the dreams of future machines ( for example large networks of sensors exhibiting collective behaviour and ultimately cogent sensing to enable cogent actuation and eternal vehicles). The evolution process is discussed from a system requirements perspective and supported by an analysis of the components which make a sensor/sensor system, from the simplest such sensor performing straight forward metrology through the self-testing sensor to the fully fledged cogent sensor which can autonomously make informed decisions on the data and perform complex information transformations. The hardware and software requirements of the sensors along this line will be discussed and example implementations will be shown.
The newer pool of potential “big” sensors applications need more than MEMS device technology perfection - the inherent, natural MEMS properties of size and potentially low cost encouraged the liberal usage of these devices in applications (smart skin with thousands of devices embedded, deployable sensor webs, etc) which in turn lead to the need to rely on/add efficient and clever processing of data generated by the sensing device, before such data reaches the outer world. Technology perfection might not, therefore, be, in the new light, the primary aim in developing successful MEMS sensors and particularly sensor systems. Since signal processing is needed anyway by the sensing application, most imperfections could also be, potentially, compensated for in the software/hardware associated/integrated with the sensor, as long as the integration of sensor and processing is resolved.
There are many current proposals for hugely ambitious information gathering systems based on very large networks of autonomous intelligent sensors. Typically, the proposals envisage thousands or millions of such sensors, collaborating together to make an overall system with highly advanced functionality (although, often, the precise nature of this functionality is not explored). This tutorial addresses such applications and presents a view on the specification and design of such large sensor systems. The architecture proposed in this tutorial provides standardised interfaces at appropriate levels of abstraction in the system, which allow the initiation and continuation of design and development at those levels of abstraction.
Attendees will gain the perspective and context of the field in order to make design decisions which optimally utilize current and forthcoming developments in these technologies.


Content
The tutorial will consist of four sessions (50 minutes/session). Sessions will be run as a mixture of presentation and open discussion.

1. Oversold dreams and practical potentials
     what was promised by MEMS and what was actually delivered – current sensor markets
     new dream applications (Smart Dust, GEMS, the ageless machines)
     MEMS attributes, technological requirements and integration challenges
          transducers: mechanical components and electronic interfaces
          signal conditioning and integration

2. Advanced Microsystems - where the top down meets the bottom up
     enhanced sensor functionality – compensation, calibration, self-test, power management
     smart sensors
     intelligent sensors
     cogent sensors – achieving autonomy

3. Top down system functions in sensor systems
     the data to information translation – what, where and how
     self contained and information delivering systems
     systems framework for very large networks of cogent sensors
     a review of potential applications of large networks of sensors
     skills and expectations of the design communities with respect to such systems.

4. Concepts which contribute to a generic and adaptable system design strategy for large sensor networks.
     architectural model to locate the system components required for building large microsensor networks
     possible implementations of some system level components and additional constraints that these place on the          overall system
          topology establishment and management
          fault management
          initialisation and syncronization
          communications
          application level functions such as data fusion

---

Elena Gaura selected publications:
1. Gaura, E. Rider, R.J. Steele, N. (2000) Closed-loop neural network controlled accelerometer, Proceedings of the I. Mech. E, Part I, Journal of Systems and Control Engineering, vol. 214, no.I2, pp.129-138.
2. Gaura, E.I.. Rider, R.J Steele N. and Naguib R.N.G. (2000) Neural Network Based Smart Accelerometers for Use in Telecare Medicine Journal of Systems Science, Vol. 26, No. 3, pp. 83-96.
3. Steele, N. Gaura, E. Godjevac, J. (2001). Neural Networks and Control. Int. Journal of Comp. Research, 10 (2) pp. 237-250
4. Gaura, E.I. Rider, R.J. Steele, N. Naguib, R. N. G. (2001). A Microsystem Approach to Acceleration Measurements in Telecare Applications. IEEE Trans. on Information Technology in Biomedicine, vol. 5, no. 3, pp. 248-252.
5. Lewis, C. P. Gaura, E. (1998). The Sigma-Delta Modulator - A Control Engineering Perspective, Symposium on Electronics and Telecommunications, Proceedings, vol. 2, pp. 110-118, Timisoara, Romania.
6. Gaura, E. Rider, R. J. Steele, N. Hesketh, T. G. (1998). Neural Networks Approach for the Design of Closed-loop, Micromachined Transducers, Microsystems - Sensors & actuators for industrial applications, Technological Meeting at MICRONORA’98, Becanson, France.
7. Gaura, E. Kraft, M. Steele, N. Rider, R. J. (1999). A comparison of approaches for the design of closed-loop micromachined accelerometers, Thirteenth International Conference on Systems Engineering, ICSE’99, Proceedings, pp.163-168, Las Vegas, USA.
8. Gaura, E.I. Rider, R.J. Steele, N. (2000). Intelligent model based control of acceleration sensors. Third International Conference on Modelling and Simulation of Microsystems, MSM’2000, Proceedings, pp.513-516, San Diego, USA.
9. Gaura, E.I. Rider, R.J. Steele, N. Beaumont, A. (2000) Development of a Prototype Neural Network Controlled Accelerometer. IFAC Mechatronic 2000, Proceedings, pp. 1039-1044, Frankfurt, Germany.
10. Gaura, E. Rider, R.J. Steele, N. (2000). Developing smart micromachined transducers using feed-forward neural networks: a system identification and control perspective Proceedings of the IEEE International Joint Conference on Neural Networks, IJCNN’2000, ISBN 0-7695-0619-4, Vol. IV, pp. 353-358, Como, Italy.
11. Gaura, E. Kraft, M. (2001). Comparison of two novel control strategies for a closed loop micromachined tunnelling accelerometer. Proceedings of the Fourth International Conference on Modelling and Simulation of Microsystems, MSM’2001, Proceedings, pp. 100-103, Hilton Head, USA.
12. Kraft, M. Gaura, E. (2001). Intelligent Control for a Micromachined Tunnelling Accelerometer Proc. Int. MEMS Workshop (IMEMS), pp. 738-742, Singapore.
13. Gaura, E. Kraft, M. (2002). Noise Considerations for Closed Loop Digital Accelerometers, Proceedings of the Fifth International Conference on Modelling and Simulation of Microsystems, MSM’2002, Proceedings, pp.154-157, Puerto Rico, USA.
14. Gaura E., Newman, R. (2003) Microsensors and arrays: measurement validation, monitoring and automatic diagnosis of sensor faults, Proceedings of the Fifth International Conference on Modelling and Simulation of Microsystems, MSM’2003, San Francisco, USA.
15. Elena Gaura, Robert M. Newman (2003) Microsensors, arrays and automatic diagnosis of sensor faults, Proceedings of the IEEE International Conference on Advanced Intelligent Mechatronic AIM2003, Kobe, Japan, pp.360-366.
16. Newman R. M., Gaura E. I. (2003) Using very large arrays of intelligent sensors, Proceedings of the IEEE International Conference on Advanced Intelligent Mechatronic AIM2003, Kobe, Japan, pp.356-359.
17. Gaura, E. Newman, R.M. (2004) Smart, Intelligent and Cogent Microsensors – Intelligence for Sensors and Sensors for Intelligence, Proc. of the 2004 Nanotechnology Conference and Trade Show, Nanotech04, Boston, USA, vol.1, Chapter9 – Smart MEMS and sensor systems, pp. 443-446.
18. Newman, R.M. Gaura, E. I. (2004) Systems issues in arrays of autonomous intelligent sensors, Proc. of the 2004 Nanotechnology Conference and Trade Show, Nanotech04, Boston, USA, vol.1, Chapter9 – Smart MEMS and sensor systems, pp. 410-413.