Database Reference
In-Depth Information
2. Limitations of Sensor Nodes
A sensor is a small device, whose volume is only a few cubic centime-
ters [7]. It is capable of sensing the environment, communicating with
other sensors, and performing simple computations. Section 2.1 intro-
duces energy constraints on sensors, while Section 2.2 summarizes other
restrictions such as unreliable transmission, and limited computational
resources.
2.1 Energy Constraint
Sensors are powered by batteries. A sensor is dead once it runs out of
battery power. The energy constraint is usually the bottleneck for most
WSNs. A sensor node consumes extremely low power during the sleep
mode, when most of its components are inactive. However, the use of
radio component, sensing unit and CPU is extremely power-consuming
[8]. A practical WSN, which is expected to be functioning for months [2
3], requires algorithms that properly use the components of sensors. For
instance, a Mica node, powered by two AA batteries, runs out of energy
in a few days if its components are constantly active. On the other
hand, it achieves six months of lifetime when it is properly programmed
[7]. The radio component is the most energy-consuming. It serves the
three functions of sending messages, receiving messages, and listening
for transmission requests from other nodes.
In order to execute a query that has been submitted to the base
station, the network protocol spreads it to the (selected) sensor nodes,
which transmit their measures back to the base station. The commu-
nication cost of query execution includes the total number of messages
transmitted in the network in order to answer this query. In particu-
lar, it covers functions i) and ii) listed in the above paragraph. There
are various algorithms to reduce the communication cost [9][10][11][12].
However, recent research shows that the time that the radio is active
dominates the energy cost, rather than the number of messages trans-
mitted. When the radio is on, the sensor is either sending/receiving
messages, or listening for incoming transmissions.
Table 3.1
shows the
typical power consumptions of the radio in different modes. We observe
that message transmitting and listening cost substantial energy. There-
fore, to reduce the energy consumption, WSN algorithms should have
the radio in sleep mode as long as possible.
