Image Processing Reference
In-Depth Information
only depends on the correct usage and participation in the sensor network suite of protocols and not
on the node's specific way of implementing software or hardware. An intuitive description given in
Ref. [] envisions a sea of sensor nodes, some of them being mobile and some of them being static,
occasionally containing tiny isles of relatively resource-rich devices. Some nodes in the system may
perform autonomously (e.g., forming the backbone of the network by executing network and system
services, controlling various information retrieval and dissemination functions, etc.), while others
will have less functionality (e.g., just gathering data and relaying it to a more powerful node). hus,
from the sensor node architecture point of view we can distinguish between several kinds of sensor
nodes.Asimpleyetsuicientinthemajorityofthecasesapproachwouldbetohavetwokindsof
nodes: high-end sensor nodes (nodes that have plenty of resources or superior capabilities; the best
candidate for such a node would probably be a fully equipped personal digital assistant (PDA) device
or even a laptop) and low-end nodes (nodes that have only the basic functionality of the system and
have very limited processing capabilities).
The architecture of a sensor node consists of two main components: defining the precise way in
which functionality is needed and how to integrate it coherently in a sensor node. In other words, sen-
sor node architecture means defining the exact way in which the selected hardware components con-
nect to each other, how they communicate and how they interact with the central processing unit, etc.
A large variety of sensor node architectures have been built up to this moment. As a general design
rule, all of them have targeted the following three objectives: energy efficiency, small size, and low
cost. Energy efficiency is by far the most important design constraint because energy consumption
depends on the lifetime of the sensor nodes. As the typical scenario of sensor networks deployment
assumes that the power supplies of nodes will be limited and not rechargeable, a series of trade-offs
need to be made to decrease the amount of consumed energy. Small size of the nodes leads to the
ability of deploying lots of them to study a certain phenomenon. The ideal size is suggested by the
nameofoneoftheirstresearchprojectsinthearea:SmartDust[].
4.2 Wireless Sensor Network Architectures
A sensor network is a very powerful tool when compared to a single sensing device. It consists of a
large number of nodes, equipped with a variety of sensors that are able to monitor different character-
istics of a phenomenon. A dense network of such small devices, gives the user the opportunity to have
a spatial view over the phenomenon and produces at the same time results based on a combination
of various sorts of sensed data.
Each sensor node has two basic operation modes: initialization phase and operation phase. In order
to guarantee good performance, the sensor network as a whole should function in a smooth way, with
the majority of the nodes in the operation mode and only a subset of nodes in the initialization phase.
The two modes of operation for the sensor nodes have the following characteristics:
•
Initialization mode
: A node can be considered in initialization mode if it tries to integrate
itself in the network and is not performing its routine function. A node can be in initial-
ization mode, e.g., at power on or when it detects a change in the environment and needs
to configure itself. During initialization, the node can pass through different phases such
as detecting its neighbors and the network topology, synchronizing with its neighbors,
determining its own position, or even performing configuration operations on its own
hardware and software. At a higher abstraction level, a node can be considered in initial-
ization mode if it tries to determine which services are already present in the network,
which services it needs to provide or can use.
•
Operation mode
: After the initialization phase the node enters a stable state, the regu-
lar operation state. It functions based on the conditions determined in the initialization
