Chapter 1
Introduction

Most advanced biological and man-made physical systems require reliable sensing to function properly. The more sensors they integrate, the more complete and comprehensive is the information they can gather from their surroundings. For a long time, practical challenges have set limits to the number of sensors that can be embedded into physical systems, processes, or environments. Among these challenges are limited space, the difficulty and obtrusiveness of wiring, heat dissipation, and power supply. Miniaturisation of sensors has fundamentally changed the way we deal with these challenges. Furthermore, integrating processing and wireless communication capabilities into sensing systems has enabled not only dynamic programmability but also networking, so that data can be processed (aggregated, filtered, compressed) in a distributed manner or can be packed in packets and transferred to a different location where they can be processed by employing advanced signal-processing algorithms.

The past decade has witnessed an explosion of interest in wireless sensors and wireless sensor networks, for which there are a variety of applications. In civil engineering, these sensors and networks can be employed to monitor the integrity of infrastructure, such as pipelines, bridges, and buildings. In the medicine and healthcare domain, they have already proved to be indispensable, but they are also finding new applications in augmenting existing diagnosis and monitoring infrastructure and in enabling more independent and flexible lifestyles for patients. In agriculture and environmental science, wireless sensors and wireless sensor networks are useful for precision agriculture, for monitoring the quality, amount, and flow of water, and for studying wildlife without the need to interfere with it.

However, the usefulness of the applications that employ sensors depends on the depth of understanding pertaining to the design and operation of the sensors as well as the quality of the data-processing algorithms employed. The faith an application developer puts in a sensor should be based on a quantitative understanding of its reliability, accuracy, precision, sensing range, sensitivity, and lifetime as well as on the strength of the assumptions supporting the data-processing schemes. Otherwise, the relevance of the application will necessarily be limited to laboratory settings, or prototypes. On the other hand, a fundamental understanding of sensors and their design leads to innovative ideas and the identification of totally new application domains.

Interestingly, the basic concepts of sensors are straightforward to grasp and the electrical circuits required to realise a sensing system are relatively simple and comprehensible, for example, compared to the design of high-frequency communication systems. This is because, in most practical situations, sensors have to deal with low-frequency signals that can be detected and processed by relativelysimple electrical components. The purpose of this book is to acquaint the reader with:

• the fundamental principles of electrical, ultrasonic, optical, and magnetic sensing
• the broad range of issues pertaining to the design and manufacturing of microelectromechanical sensors (MEMS)
• the principles of energy harvesting and sensor integration
• the fundamental assumptions and methodologies pertaining to the processing of sensed data.

I have tried to make the book self-contained by discussing the necessary prerequisites within the book itself, so that the reader is not obliged to refer to other materials in order to understand the text.

1.1 System Overview

Figure 1.1 displays the most essential building blocks of a self-contained sensing system. These are the sensing system, the conditioning system, the processor, and the wireless transceiver. The figure is intended to give a complete picture, but we shall not be dealing with the wireless transceiver here. Whether or not these blocks are distinct from each other depends on many factors, such as the quality of the signal that can be sensed by the sensor, the targeted energy and space efficiency, and the ease with which the wireless sensor should be integrated into and interact with other systems. The processor and the wireless transceiver are usually connected with the rest of the system using standard buses that use standard or quasi-standard protocols. Hence, the main design issue is how to integrate the remaining building blocks. I shall briefly summarise the function of each building block and highlight the different trade-offs that influence its integration with the other blocks.

Figure 1.1 The main building blocks of a wireless sensor.

1.1.1 Sensing System

The process we wish to observe or monitor is called the measurand. It either releases some form of energy that describes its condition in some way, or external energy in the form of an electrical (radio-frequency), optical, or acoustic signal is applied to it, so that from the way it modifies some of the characteristics of the signal (magnitude, phase, frequency or a combination of these), its condition can be determined or inferred. The human body is a typical example of a measurand, because it is a remarkable signal generator. The human brain generates electromagnetic signals that can be sensed by electroencephalogram or magnetoencephalogram. Likewise, the human cardiovascular and muscular systems generate electric potentials that can be sensed by employing an electrocardiogram or electromyogram. In contrast, ultrasound systems release ultrasonic waves into a human organ and the spectrum of the reflected signal is analysed to determine the organ's condition. If a measurand's condition is determined from the signal it releases, then the sensing method is called "passive sensing". Otherwise, it is called active sensing. Passive sensing introduces less intrusion into the measurand compared to active sensing, but the amount of energy that can be collected through passive sensing is normally small.

1.1.2 Conditioning System

Regardless of the sensing mechanism employed, there are important conditions the sensor and the signal it produces should fulfil before useful features can be extracted from the signal. One of them is appropriate interfacing. When a sensor is employed to a measurand, apparently the measurand "perceives" the presence of the sensor, because the sensor draws some amount of power from it. This power must not affect the measurand's proper operation. A related issue to interfacing is impedancematching. If the impedance of the measurand as seen by the sensor is not matched by the sensor's own input impedance, maximum power does not transfer from the measurand to the sensor. Instead, power dissipation in the form of heat may be experienced at the interface, disturbing the measurand and reducing the efficiency of the sensor. Therefore, the impedance of the measurand (human body, water, air) should be taken into account when the sensor circuits are designed.

Even with the interfacing problem solved, the signal produced by the sensor may not accurately reflect the measurand's true condition for a number of reasons. Noise may be added to the sensed signal from the surrounding environment or from the internal circuits of the sensor itself. Likewise, some portion of the signal may be removed, suppressed, or distorted, because the sensor circuits act as filters. Therefore, the bandwidth of the desired signal and the bandwidth of the sensing circuits should be matched. It must be noted that in most real-world cases, the signal produced by a measurand contains a range of frequency components. The purpose of the conditioning system is to deal with all these issues. A conditioning component typically consists of a filter circuit and a differential amplifier, the order in which they appear usually depending on the nature of the measurand as well as the strength of the signal produced by the sensing system.

1.1.3 Analogue-to-digital Signal Conversion

This component is not directly shown in Figure 1.1 because it may be a part of the conditioning system or the processor or it may be a distinct entity. Regardless of its specific position, the analogue signal the sensor produces and the conditioning system pre-processes should first be converted to a digital bit stream before it can be further processed by a microcontroller or a digital signal processor (DSP). In some sensors, the analogue-to-digital converter (ADC) is an integral part of the conditioning system, while in others it is a separate block. Modern microcontrollers also integrate multiple general-purpose ADCs, to one of which the analogue signal coming from a conditioning circuit can directly be fed. Next to the transceiver and the processor, the ADC is the largest power consumer and hence care must be taken in choosing a suitable ADC. Several factors determine the choice of an ADC. For example, if the sensor signal is noisy, it is better not to use a powerful pre-amplifier lest the noise is amplified together with the useful signal. In this case it is better to use a high-resolution ADC, so that an efficient DSP algorithm can eliminate the noise. But a high-resolution ADC consumes a large amount of power and generates a large amount of data, which require a sizeable resource to process, store, and communicate. If, on the other hand, there is a small noise component in the signal, then it is better first to amplify the signal and use a low-resolution ADC. If the ADC is not an integral component of the sensor or conditioning system, then it is possible to use the sensor for...