Wireless sensor networks consist of a multitude of small sensors knots imbedded in the environment. The sensors knots are able to observe phenomena, e.g. temperature or soil moisture fields, with high temporal and spatial resolution. The sensor knots are wirelessly networked and can transmit information from one to another (multi-hop communication), until the information centre is reached. The wireless sensor network design has to consider channel errors, delays, packet losses, and power and topology constraints. In order to conserve battery power, a reactive routing protocol is used that determines a route only when it is required. The sensor network can react on external influences, e.g. a network for monitoring soil moisture should adapt its measurement and transfer frequency depending on rainfall occurrences.
In order to account for these problems, several requirements on the sensor network have to be fulfilled:
• since a multitude of sensors is needed, the single sensors have to be cheap
• since the sensors may have to be buried, they have to be very robust
• a bi-directional communication between the sensor units and the coordinator has to be possible
• in order to enable long-term undisturbed monitoring, the sensors have to be as low-current as possible
The network may have a flat or a hierarchical architecture. Both have advantages and disadvantages. The hierarchical architecture is more survivable since it does not have a single point of failure and it also allows multiple routes between nodes. The flat architecture provides a simpler network management.