Mobile Intelligence

27.2 PRINCIPLES OF EVENT RECOGNITION FOR SENSOR NETWORKS

In this section, we provide the foundation and discuss the concepts applied in our approach for event detection using sensor nodes.

27.2.1 Event Patterns

The occurrence of events of interest is captured using patterns. A pattern may be defined as the opposite of chaos [26]. For example, severe weather conditions such as thunderstorms have weather maps or patterns that can be used to predict their occurrence. Similarly, in sensor networks, an event pattern could be a weather map, the behavior data of sea birds, or stress levels in a bridge.

Sensory data convey information about the conditions of the environment. Using patterns instead of handling individual sensory data points provides a practical method for capturing multidimensional and complex events of the physical environment. We assume that an event pattern is represented by M × N cells. Each cell is associated with one sensor node and corresponds to a specific region of the monitored area. We assume that the number of active sensor nodes is equal to the number of cells (M × N). In scenarios where the number of deployed sensor nodes might be greater than M × N, cell leaders may be elected to represent the cells and aggregate the sensory data of sensor nodes belonging to their cell. Consequently, a matrix of sensor nodes of size M × N is created to correspond to event patterns. The matrix, however, can be of regular or irregular shape, depending on the deployment scheme. ...