Transcript Wireless Distributed Sensor Networks

Wireless Distributed
Sensor Networks
Special Thanks to:
Jasvinder Singh
Hitesh Nama
What is a Sensor Network??
• A sensor node is a small device complete with
sensing, data processing, and communication
components
• A sensor network is composed of a large number
of sensor nodes that are densely deployed either
inside the phenomenon or very close to it.
• Sensor nodes in a network work cooperatively,
meaning that they can use their processing abilities
to locally carry out simple computations, and then
transmit only the required data.
What is a Sensor Network??
• One big advantage for a sensor network is that the
position of the sensor nodes can be random; it
does not need to be predetermined.
• Sensor networks can be used for a myriad of
applications, including:
– Health
– Surveillance / Reconnaissance
– Disaster Relief
Information Distribution (Energy Efficient
Routes)
Information Distribution (Energy Efficient
Routes)
•
•
•
•
PA = power available
 = energy required
ME = minimum energy
MH = minimum hop
• Four possible approaches to choosing “energy efficient”
route – Maximum PA, Minimum Energy, Minimum Hop,
Maximum minimum PA node route
Information Distribution (Energy Efficient
Routes Continued…)
• Maximum PA searches for the largest total (sum of) PA
corresponding to a path w/out extending any routes (A-B-T
vs. A-B-C-T)
• Minimum Energy looks for the route that consumes the
least amt of energy. This is path A-B-T
• Minimum Hop selects the route that takes the fewest hops
(route D-T)
• Maximum minimum PA node route selects the route where
the minimum PA is larger than the minimum PA’s of the
other routes (route D-T)
Information Distribution (Data Aggregation)
Information Distribution (Data Aggregation)
• Based on Data-Centric routing, a process where sensing tasks are
assigned to sensor nodes based on what the other nodes are interested
in.
• D-C routing requires attribute-based phenomena, where users are
interested in the occurrence of a specific phenomenon, rather than the a
query of an individual node.
• Data aggregation is a technique used to solve implosion and overlap
problems in data centric routing.
– Implosion – duplicated messages are sent to the same node
– Overlap – neighboring nodes receive duplicated messages because
their overlapping “observation regions” sense the same stimuli at
the same time.
Information Distribution (Data Aggregation
Continued…)
• Data aggregation can be seen as a set of automated
methods of combining the data that comes from many
sensor nodes into a set of meaningful information.
Information Distribution (SPIN)
Information Distribution (SPIN)
• SPIN – Sensor Protocols for Information via Negotiation
• Energy Efficient combination of two packet sending
methods
– Flooding – each node receiving a data or management packet
repeats it by broadcasting, unless a max number of hops for the
packet is reached or the destination packet is the node itself.
Requires little maintenance, but subject to implosion and overlap.
– Gossiping – incoming packets are sent to a randomly selected
neighbor. Avoids implosion, but takes a long time.
• SPIN contains 3 messages: ADV, REQ, DATA
– ADV contain descriptor of DATA
– If neighbor is interested, it sends a REQ for the DATA
– DATA is sent
Information Distribution (Directed Diffusion)
Information Distribution (Directed Diffusion)
• Sink sends out interest to all sensors, containing a
timestamp field and several gradient fields. The interest
stored in the catch of each sensor node.
– Interest – description of a task
• As the interest is propagated through the network,
gradients from the source back to the sink are set up.
• When the source has data for the interest, the source sends
the data along the interests gradient path.
Information Distribution (other methods)
• Sequential Assignment Routing – enables sensor nodes to
discover their neighbors and establish transmission /
reception schedules without a central management system
(usually through a TDMA implementation).
– SAR algorithm creates multiple trees where the root of each tree is
a one-hop neighbor from the sink. Each node usually belongs to
multiple trees.
– Belonging to multiple trees allows a sensor node to select what
path to route its data back to the sink.
• A similar protocol, LEACH (Low-Energy Adaptive
Clustering Hierarchy), randomly selects sensor nodes as
clusterheads, so the high energy dissipation in
communicating with the base station is spread to all the
sensor nodes in the network.
Power Consumption
• A wireless sensor node can only be equipped with
a limited power source (it is a microelectronic
device).
• It is important that the battery lifetime for a sensor
node is long (on the order of several months)
because the malfunction of even a few nodes can
cause significant rerouting of packets in the sensor
network
Power Consumption - DVS
• Energy consumption in a static CMOS-based
processor has components – switching
(independent of time) and leaking.
– The switching energy is usually calculated as
Eswitch = Ctot * Vsupply2
• Dynamic Voltage Scaling (DVS) suggests that
reducing Vsupply can result in energy savings, at the
cost of additional propagation delay
Power Consumption – RF Hardware
• For short-range transmission at gigahertz carrier
frequencies, the radio’s power is in large part used
by the frequency synthesizer which generates the
carrier frequency rather than the actual transmit
power. Hence, data rate does not affect power
consumption of the transmitter.
• However, as packets become shorter, the radio’s
start-up time becomes significant.
• To reduce energy, the node’s radio module is duty
cycled.
Power Consumption – System Partioning
• Algorithm implementations for a sensor network
can take advantage of the network’s capability for
parallel processing to reduce energy.
• Partitioning a computation among multiple sensor
nodes and performing the computation in parallel
will allow energy savings through frequency and
voltage scaling.
Power Consumption – Software
• The overall energy efficiency of wireless sensor
networks its heavily dependent upon the software
that runs them.
– General purpose processors and DSPs offer more
flexibility for a sensor network than dedicated circuitry,
but this results in the increasing importance of
programmable solutions.
– Improving control and application software can
substantially reduce power consumption
Conclusion
• For more information:
http://www.cdt.luth.se/babylon/snc/References/Akyil
diz2002_SurveySensorNets_01024422.pdf
http://www-mtl.mit.edu/~anantha/