|
Sensor Synchronization: Wireless sensor networks have become an
important research area over the past few years. These collections of low-power
sensors can monitor their environment and communicate with each other. However,
many applications such as motion detection require time awareness; a bullet’s
trajectory can be accurately traced to its point of origin by comparing the
times at which different sensors detected the blast’s shockwave. A high accuracy
algorithm must be implemented so that the sensor clocks closely mimic each
other. Several such techniques exist, including Reference Broadcast
Synchronization (RBS) and the Timing-sync Protocol for Sensor Networks (TPSN).
For RBS, a reference broadcast is sent from a root sensor. The receivers record
their internal clock time at which they received the broadcast, and then
exchange it with each other. In TPSN, a hierarchical tree is established from
the root sensor, and pair-wise synchronization is performed. Both of these
algorithms offer accuracy within a few micro-seconds, with TPSN being more
effective. However, there are some issues that must be addressed as well.
Wireless sensors are small and are limited to using low-capacity batteries. The
most power consuming action is transmitting data through the RF radio. This is
especially true if the nodes transmit at a relatively high power, such as in a
sparsely populated network. Neither RBS nor TPSN take power requirements into
account when transmitting synchronization signals. Furthermore, the concept
behind effective wireless sensor networks works best with a large number of
sensing nodes, which requires cheap internal parts. Cheap crystal oscillators
are used as the sensors’ internal clocks, so clock drifts and skews are
significant factors. Neither of the protocols takes drifts into account.
Our work involves analysis of RBS and TPSN energy consumption. We are designing
a hybrid algorithm that focuses on conserving energy and maintaining the
network’s topology for as long as possible. The main interface window is shown
in Figure 1. The network grid and sensor parameters are fully customizable
(Figure 2), allowing for a large number of scenarios to test. A flood signal is
broadcast from a designated root node, and is forwarded to subsequent nodes as
they pick up the broadcast. The transmission strength decreases according to the
wireless path-loss equation. The algorithm being developed operates recursively
through each node, focusing on minimizing the number of transmissions; an RBS-type
communication is used for nodes that transmit few signals while TPSN is used for
denser situations. Each node has a clock drift generated using a Gaussian
distribution, so the sensors will require periodic resynchronization. These
resynchronizations will eventually deplete some of the sensors, which will
create holes in the topology. A new root node is selected once the number of
unattached nodes increases beyond a specified threshold. This action will help
increase the number of nodes being used, which maintains a larger physical area
being monitored.
This hybrid algorithm reduces energy consumption, while at the same time
maximizing the physically monitored area. It also compensates for internal clock
drifts to further improve upon currently used time synchronization algorithms.
 |
|
Sensor Energy-efficient Routing:
Demo
In recent times, battery-powered ad-hoc
networks called wireless sensor networks have become popular. In such networks,
sensor nodes that are scattered over a region connect to each other and form
multi-hop networks. These nodes are equipped with sensors such as temperature
sensor, pressure sensor, etc. and can be queried to get the corresponding values
for analysis. However, since they are battery operated, care has to be taken so
that these nodes use energy efficiently. One of the areas in sensor networks
such an energy analysis can be done is routing. This paper explores grid-based
coordinated flooding in wireless sensor network and compares the energy left in
the network over time for different grid sizes as well as for same grid size.
Simulations run on different grid sizes show that grid of 100 meters conserves
energy better and extends network lifetime longer than any one of 200, 150 or 50
meters grid size networks for the same node location.
A test area is divided into square-shaped grids of certain side length. Fresh
battery powered nodes are randomly placed in the area with the source and sink
nodes at fixed positions. One node per grid is elected as the coordinator which
does the actual routing. Non-grid coordinators do not participate in the routing
process. The source node starts the flood in the network with every coordinator
joining in routing. Once the flood reaches the sink node, flooding ceases and
information is continuously sent to the source by finding the back route to the
source. This process is continued until the first node (coordinator) along that
route runs out of energy. New coordinators are elected to replace the dead ones.
The source node refloods the network so that the sink can find new back route to
send information. This entire process continues until the network partitions and
the connectivity between the source and the sink nodes is lost.
Coordinator election is simple. The node with the maximum ID in the grid is
elected as the grid coordinator. When this coordinator runs out of energy, the
node with the second maximum ID becomes the new grid coordinator. This election
takes place when the flood hits a dead grid coordinator.
As the source refloods the network, every coordinator goes through three states
based on its remaining energy. If the remaining energy is greater than 25 % of
battery life (amount of energy when simulation was started), coordinators are in
routing state. If the remaining energy is between 5-25% of battery life,
coordinators are in energy alarming (warning) state. Finally, they are dead when
the remaining energy is less than 5% of the battery life. Energy calculations
are done as ratio of total energy of all nodes with coordinator nodes and total
energy of all nodes at the start. Such energy calculation is nothing but
normalized energy. Simulation results are plotted as normalized energy versus
simulation time. The results indicate that networks with grid sizes of 100
meters show better overall performance compared to 200, 150 or 50 meter grids.
|
Robert
Akl, D.Sc.
