Chapter 12Estimation

Designing a good sensor is the first step towards interfacing the physical world with the virtual world. Two additional steps are required before the data obtained from a sensor can be useful. The first step deals with determining whether the data represent the physical reality and the second step deals with understanding the meaning of the sensed data. Erroneous representations or interpretations of sensor data often have detrimental consequences and great care must be taken over the two stages. This chapter deals with the intermediate stage. I will not be dealing with the last stage in this book, as it is application dependent. I shall begin this chapter at the simplest level; you may already know some of the concepts and techniques I will be treating, but my goal is to lay a sound foundation, so that the book is self-contained.

The subjects treated in this chapter are by no means exhaustive. Moreover, I regard estimation from a single viewpoint, which is the processing of sensed data. I refer readers wishing to further enrich their knowledge on random variables, stochastic processes, and estimation techniques to the excellent books by Papoulis and Pillai (2002), Ross et al. (1996), Gardiner (1985) and Grewal (2011).

I shall begin this chapter by making a somewhat sensitive statement: we shall never be able to construct a sensor (or any system, for that matter) that captures reality as it is. We can approach reality but never touch it. Even at a quantum ...

Get Principles and Applications of Ubiquitous Sensing now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.