Sensor Technologies - Artificial Human Sensors: Science and Applications - page 94

Figure 4.8. A sensor (n by n) array data response depicted as a coloured sensor response
level of information picture.
information flow connected to the individually placed sensors, that provide sep-
arate data flow at different time sequences to clarify measurement inconsistency.
A large number of significant applications depend on how and where the sensor
elements are placed, that interface with the real world, e.g., Petersson (2008). These
distributed sensor networks is an emerging topic gaining increased interest and
applications may include military, medical, manufacturing, transportation, safety,
and environmental planning systems.
A number of multi-sensor systems have made emphasis to be fully realised,
but not completely succeeded. This is because of problems involved with collect-
ing the right data from the specific sensors in a predetermind measuring space
directly into automated systems, thus providing the requested and often time-
dependend information. Sensor fusion algorithms have been a useful technique
of choice for resolving these problems and to fully make use of the collected data.
4.4.3.1 MULTI-SENSOR DATA FUSION
The terminology in the process of association, correlation and combination of
information seems confusing, Wu (2000). It seems that the phrase “sensor fusion”
has shifted to “data fusion” and the trend is now moving towards “information
fusion”. Sensors usually provide location and time dependent flow of data that
have to be further processed and interpreted in the context to provide meaningful
information. A multi-sensor system may perform a data fusion process that carries
out the combining and integrating activity in order to extract an added value from
a diversity of measurement data, Hall (1992), Klein (1999).
A multi-sensor data fusion system has the main task to:
— organise data collection and signal processing of a number of sensors that may
be of similar or different types,
— present the local and global representations, by using the received and earlier
gained information,
— integrate the information from the different sensors into a dynamic and time
related real time model,