Sensor Networks’ Integration - Application and Multidisciplinary Aspects of Wireless Sensor Networks - page 92

JWebDust
jWebDust [ 15 ] provides a software framework that allows web-based applications to
query and control multiple potentially heterogeneous wireless sensor networks.
WebDust is based on a multitier architecture, splitting the overall system into five
tiers, namely sensor, control, data, middle, and presentation tier. All framework
components are implemented in Java, apart from the ones contained in the sensor
tier. The sensor tier is formed of one or more wireless sensor networks consisting of
sensor nodes (motes) operating a TinyOS-based jWebDust firmware. The jWebDust
firmware enables multi-hop routing within the WSN and provides support functions
such as query subsystem, discovery services, monitoring service, and time synchro-
nization. Sensor networks are connected to control centers that form the control tier ,
acting as gateways between the sensor tier and data tier. The control centers are
responsible for the gathering of all readings coming from the sensor network and the
forwarding of queries from the data tier to the sensor nodes. Control centers periodi-
cally poll the data tier for new available queries and store all available sensor readings
into the data tier. Control centers are able to handle temporal disconnections of the
sensor tier from the data tier by buffering of sensor readings until reconnection sub-
ject to local capacity constraints. The data tier is based on a relational database
system and hosted by SQL servers. Information is stored in tables that can be
grouped into three categories. Mote-related tables store information on the hardware
characteristics and sensing services of sensor networks. Query-related tables hold
information on currently active queries in the jwebDust system. Sensor-reading
tables hold sensor readings that have been performed by each particular mote in the
system. The middle tier provides a set of reusable components that allows the map-
ping of the information stored in the tables of the data tier, the manipulation of the
information, e.g., creation of queries in the data tier and implementation of rule-
based actions and notifications. The presentation tier implements the user interface
components visible to the end user, providing interaction controls and sensor data
visualization tools.
jWebDust decouples applications from the sensor networks via the n -tier archi-
tecture and supports horizontal composition of WSAN applications of different
domains. Applications can learn about the availability of different sensor networks
and their capabilities by querying/browsing the data tier. Individual motes within
a WSN are assumed to have unique IDs. In order to make them globally distin-
guishable across multiple WSNs, each WSN is assigned a unique sensorNet-
workID. The addition of a new sensor network requires some initial configuration,
e.g., assignment of sensorNetworkID and inclusion of novel mote and sensor
types. Afterwards, discovery of new motes and sensors in a sensor network exe-
cutes automatically. jWebDust supports interactions with different heterogeneous
WSNs, where control centers are able to hide the heterogeneity of WSNs from the
data tier. Although the framework claims to support a variety of different query
types, applications currently require explicit knowledge of a sensor network and
sensor types (learned from the data tier) in order to create various queries for sensor
information. Service broker-like components with semantic query support could