Energy-Efficient Protocols for communication in Biosensor networks

1
Energy-Efficient Protocols for communication in
Biosensor networks
2
What are biosensors?
Introduction
These are microsensor devices placed in or on the
body. A network of such biosensors can be used to
monitor health or to act as prosthesis.
Why do we need wireless communication?
The data obtained by each sensor by itself is not
important. It is the aggregation of data that is
important. For this communication between sensors
is necessary. However, a network may have 10s or
even 100s of sensors and wiring them inside the
body is not possible.
3
An example of the use of biosensor networks
Introduction
Even though their nerve-endings are fine, a lot
of people cannot see because the rods and cones
in their retina have degenerated. We can induce a
sensation of sight by artificially stimulating
the nerve-endings.
How do we do that?
We place an array of biosensors in front of the
nerve-endings at the back of the eye. An external
processor compresses images it receives from a
camera placed on the spectacles and transmits
them to this array. We also need feedback from
this array and hence reverse communication should
also be possible.
4
The Retinal Prosthesis Project
External Processor
Transceiver
Sensor network
Optic Nerve
Spectacles
5
1. Low but constant energy supply.2. Low
bandwidth.
Characteristics of a Biosensor Network
Since biosensors are typically implanted in the
body, they cannot be powered by internal
batteries that will run out of energy. To
overcome this problem the battery packs are
placed outside the body and the power is supplied
to the sensor via infrared radiation.
These sensors have to communicate in an
error-prone environment with high path loss
exponents. Increasing signal power is not an
option as energy supply is scare.
6
3. Robustness and fault tolerance.4.
Size
Characteristics of a biosensor network
As biosensors are surgically implanted into the
body, replacing failed sensors is difficult.
Fault tolerance is also important because the
failure of one node should not incapacitate the
entire network. This calls for a dynamic,
self-stabilizing network.
As these sensors have to be implanted into the
body, they have to be as small and lightweight as
possible.
7
Observations
Wireless Communications

1. Each sensor has to collect and transmit data
   every cycle. This suggests a TDMA approach to
   medium access. A TDMA approach will ensure
   contention-less access and hence no energy is
   wasted in collisions.
2. Neighboring sensors will collect more or less
   similar data. This suggests that we can use local
   data compression by removing redundant data.
   Doing this will reduce the number of bits to the
   transmitted and hence the transmission cost.

8
Cluster-Based Approach
Our Approaches to the Problem
We can reduce the power consumption by
limiting the long distance transmissions. We do
this by organizing the network into clusters.
Each cluster has one leader and several
members. The members transmit their data to the
leader. The leader compresses the data and
transmits it to the the base station. The
selection of leaders is done by the external
processor. Several factors such as nodal density
and transmission range will be taken into account
during the selection process. Once the
leaders are selected and the nodes have
subscribed to a cluster, all communication
follows TDMA.
9
Cluster-based Approach
10
Tree-Based Approach
Our Approaches to the Problem
.
This is the tree-based routing that is
common to todays networks. The base station is
at the root of the tree. Nodes that are reachable
from it become its children. These children in
turn have their own children and so on.
Data from any node has to hop across several
links to reach the base station.
11
Tree-based Approach
12
We observed that the power consumption depended
on (a) Number of nodes(b) Percent of nodes that
are cluster-heads(c) Propagation loss
exponent(d) Distance between nodes.
Results
13
Our analysis shows that the cluster-based
approach provides better energy-efficiency than
the tree-based approach. This is true for a
wide range of path loss exponents, as long as we
adjust the number of clusters in the network.
For high path loss exponents, a smaller number
of clusters is needed.
Results
14
Results (contd.)
15
Results (contd.)
16
Results (contd.)
17
Results (contd.)
Explanation of Results In the tree-based scheme,
each data packet passes via several levels of the
tree and transmission/reception costs are
incurred at each level. The cost of such
multi-hop routing then becomes greater than the
cost of single-hop routing that is present in the
cluster-based scheme.