Sensing

Observing various aspects of the environment is repeatedly done everyday. For instance humans are capable of sensing temperature, which may stimulate certain actions to accommodate their preferences such as switching on/off the air conditioning system. In a large number of industrial plants, monitoring of equipment is done by humans and their judgement determines when a component is to be replaced. Sensors are used to substitute humans for monitoring events and for sensing physical properties of an environment. Clearly one sensor on its own can not give precise information about the overall attributes of an environment or the state of an industrial equipment. Hence by increasing the number and the types of sensors used, not only is the coverage increased but also the gathered data more closely resembles properties of the actual physical environment. Using a large number of sensor has a number of valuable advantages;

• By using more sensors, radical events are detected easier. e.g. if a large number of sensors were used as opposed to one or two sensors to monitor a house for fire detection, it would allow detection of fire before it spreads whereas if only a few sensors were used the fire would have to spread until it reaches the location of a particular sensor, by which time it might be too late to take any preventive action.
• Gathered data is less biased. e.g. If an office building temperature is being monitored with many sensors and one particular sensor reading is very high due to its particular location, for example it could be placed next to a light. The odd readings of this sensor can be filtered out when compared with the rest of the sensor readings taken in the same room.
• Incorrect reading due to sensor failures can be detected. Sensors are prone to failures and tend to degrade over time. This results in incorrect readings being obtained. Having many sensors allows for early detection of sensor failures.

Management of large number of sensors on per sensor basis is not practical, which is why using a networked set of sensors is appealing. However networked sensors have some constraints. The major shortcomings of networked sensor networks are:

• Cost of wiring and deployment
• Difficulty of maintenance, mainly as a result of cabling issues and lengthy replacement procedure.
• Lack of flexibility in node deployment due to wired nature of sensors both to provide power and retrieve sensed phenomena.

Advances in technology has allowed for development of very small devices capable of processing, wireless communication and also sensing phenomena(physical attributes and properties being sensed are commonly referred to as phenomena in context of sensor networks). In comparison to networked sensors, the two most notable points about these devices are that i) they are very power efficient, which means they can run on batteries for a long time and ii) they can communicate wirelessly, which means there is no longer a need for cumbersome and expensive cabling. The setup of such networks have become popular in various observation applications(e.g. habitat monitoring) and spawn the field of wireless sensor network(WSN).