Sunteți pe pagina 1din 11

Wireless Sensor Network System Design for Environmental Monitoring:

Authors :
Ninad Anjali Waman
Barnali Roy
National Institute of Technology Silchar

Abstract:
We propose generic wireless sensor network systems for two types of environmental
monitoring needs:
1.Time Critical Applications for Early warning.
2.Non-Time critical for Long-Term data Collection for Trend Analysis.
The construction and deployment of the wireless sensor network along with the
challenges of the terrain for design of various components of the network application.
Both the hardware and software technology considerations are taken care of. The
communication protocols are selected as per the requirements of the application.
The Sensor data may consist of environmental parameters like temperature,
precipitation,Water Flow, Pressure and the dynamic variations of these physical
quantities.

Keywords:
Wireless Sensor Networks, Environmental Monitoring, Time Critical
1. Introduction:
Wireless Sensor Networks(WSN's) are formed by a set of nodes that gather information
and forward it to the sink. Sensor nodes in general are extremely small,low cost, low
energy that possess sensing, signal processing and wireless communication
capabilities. Sensor networks are emerging as a new tool for habitat monitoring in nature
preserves, monitoring and gathering events in hazardous environment , surveillance of
buildings and surveillance of enemy activities in a battlefield environments. Wireless
Sensor Networks have emerged as one of the dominant technology of this decade that
has a potential use in defence and scientific applications.
Many Environmental applications of sensing are not possible using satellite imagery. At
the same time long term trend data might be required. Therefore it requires continuous
monitoring .Due to Harsh terrain it is not feasible for humans to directly collect data
from the sensors .This is an ideal scenario for the deployment of wireless sensor
network to monitor the environment and provide data over a long period of time.
A real life scenario in this case is the monitoring of glaciers in the Indian Himalayas.
There has been rapid decrease in the volume of glaciers and receding of glaciers. At the
same time, there is a constant danger of Glacial Lake Overflow events(GLOF) for
people living near the source of the rivers.
There is thus a necessity of constantly monitoring for long-term trend analysis and an
early warning system for GLOF.

2. Hardware Technology:

This needs consideration of two types of requirements for hardware.


A. Nodes for Sensing
1. Application Specific
2. Varying in number , Small, low-cost, smart.
3. Low Power
4. Static nodes but ad-hoc topology

B. Base Station, SINK Nodes and Cluster Heads:


1. Control of Network
2. Data Aggregation and Interpretation.

C. Relay Nodes:
1. Relay data from source to sink or to Base station

Due to communication energy constraints and long network lifetime required a system-
on-a-chip solution is best suited for long term data calculation. The ZigBee SOC
hardware platform is more suited for this application. Various chipsets are available from
the ZigBee alliance partners. Among these CC2420 has the lowest Transmission and
reception supply current. It has sensitivity of -95 dbM and a transmission and reception
current of 19 mA and 17.4 mA respectively. System on a chip solution includes the
CC243x series of chipsets having a sensitivity of -94 dbM and receiver and transceiver
current of 27 and 25 mA respectively
Another Option is FreeScale Semiconductor’s Integrated System-in-Package SiP
MC312x series of chipsets.
For real-time monitoring and reporting of events time-critical response is more
important.This implies latency time has to be minimized.
The NA5TR1 NanoTrontm real-time location CSS chip is an ideal solution for real-time
application requirements .It uses the ZigBee Protocol Stack(802.15.4) has a sensitivity
of -94 dbM and current 27 mA and 25 mA in the receiver and transmitter respectively.
Berkeley motes family-MICA motes based on the ATMEGA family of MCU’s from
ATMEL are a widely used sensor node series.
Number of Nodes: This will depend on the terrain and the Application requirements.
For high volume data a large number of nodes with data aggregation capability is
required. This also helps in multi-hop transmission and in case of failure of nodes.

For an Event monitoring application, the nodes need to be deployed strategically.


Topology of the network is more important than number of nodes. Where man made
deployment is not possible, post-deployment topology needs to be formed and network
configured optimally. A distributed optimization needs to applied on the network as a
whole. The availability of geographical information and its resolution also affects the
design of the sensor network.
Following is a table describing the hardware capabilities of the sensor network chip
hardware considered:

RF Module Sensitivity(dbm) Current(Rx,Tx,mA) Manufacturer


CC2420 -95 17.4,19 TI
CC243X -94 25,27 TI
NA5TR1 -94 25,27 NanoTron
MC312x -92 31,38 FreeScale

3. Medium Access Control (MAC) Protocols:

While the ZigBee Standard Communication protocol stack is most suitable for
most Wireless Sensor Network applications. However , the special constraints of
environmental monitoring in harsh terrain requires the consideration and use of some
other protocols .The S-MAC or Sensor-MAC (sleeping nodes) protocol serves best the
needs of long-term data monitoring for trend-analysis.

Features of S-MAC:
It is contention-based protocol i.e. the various nodes compete for the opportunity to
transmit in the medium. It is a modified version of 802.11n.Each node will periodically
go to sleep-listen cycles. Control packets such as SYNC,RTS,CTS are sent only during
listen cycles. The Duty Cycle for a WSN using this protocol is 10% .

Data Gathering MAC D-MAC:


It is a schedule based wireless protocol. Its main objective is to achieve low latency
while maintaining the energy efficiency. It has been designed and optimized for tree
based data gathering. It uses a combination of TDMA and CSMA.It uses ACK within
each slot to transmit and receive one packet. The packet delay is of the order of
milliseconds. Results from simulation show that D-MAC outperforms S-MAC in a
variety of parameters including throughput, latency and energy efficiency.

The DASH7 wireless sensor network technology is another alternative. It is an ultra-low


power technology originally created for military applications but now adapted to
commercial applications. It is an open standard operating in the 433 MHZ unlicensed
spectrum with support for sensor, encryption and other features.It has a small protocol
stack thereby reducing packet processing overhead and hence reducing the energy
consumption per bit. Each node has a transmission range of 1 kilometer with an average
power draw of 30-60 microwatts. It has a latency time of 2.5-5 seconds which may be
sufficient for some time critical applications.

4. Clustering Protocols:

When large amount of data is to be processed then data aggregation needs to be


performed. Data Aggregation: The data from individual nodes is aggregated at a point
from where it is sent to the base station for processing. The Cluster-head needs to be
able to transmit to the base station .For this a multi-hop strategy may be deployed. If N
hops are in the whole transmission then this will be more energy efficient compared to
single-hop transmission by a factor of Nα-1, where α is a parameter based on
environmental factors .This is essential for large data.Here LEACH based hierarchical
clustering is appropriate for the purpose.
However for event monitoring time-critical response is more important and data transfer
is small and for a limited amount of time. Here the problem of flooding and node drain
needs to be considered

5. Routing Protocols:

These have to be selected and stacked upon the chosen MAC protocol.
For DATA data Aggregation PEGASIS will prove to be near optimal. This will ensure a
longer network life time.
However, the cluster heads need to be selected properly as it is vital to reach the
information to the SINK. TEEN can prove to be useful where only above or below
threshold data are to be recorded i.e. anomalies.
Directed Diffusion Protocol is a better option as it dynamically generates routes in the
sensor network. Its Interest-Gradient-Reinforcement based methodology also makes it
more robust and fault tolerant.

For Time Critical application APTEEN is the best solution as it can distinguish dead
nodes and event monitoring needs only event-triggered data. It is a LEACH(Low Energy
Adaptive Clustering Hierarchy)based protocol
This also allows for three types of queries to be performed viz.
1. Historical Queries
2. One-time Queries
3. Persistent Queries

Features of Various Routing Protocols:


A reliable and energy efficient protocol is very much important for supporting various
WSN applications. There have various routing protocols,bellow are some of them. We
have compared among them which will be best applicable for our system.
LEACH(Low Energy Adaptive Clustering Hierarchy):
It is a hierarchical clustering algorithm for sensor networks. Here operations are divided
into rounds. During each round a different sets of nodes act as cluster-heads. Nodes that
have been cluster-head can not become cluster-heads again for p rounds. Thereafter,
each node has a probability of 1/p of becoming a cluster-head in each round. At the end
of each round,each node that is not a cluster-head selects the closest cluster-head and
joins that to transmit data. The cluster-heads then aggregate and compress the data and
forward it to the base station,thus it extends the lifetime of the major nodes. In this
algorithm,energy consumption will distribute almost uniformly among all the nodes and
the non heads are turning off as much as possible. This routing protocol seems
applicable to our application,but this has some drawbacks. In each round,LEACH has
cluster-heads comprising of 5% of the total nodes. It uses TDMA(Time division
Multiple Access) as scheduling mechanism, which makes it prone to long decays when
applied to large networks. So its inadequate for time-critical applications. As the basic
algorithm assumes that any node can communicate with the sink provides it limited
scaling. Again it has got “Hot Spot” problem, i.e. nodes on the path from a congested
area to the sink may drain.

PEGASIS(Power efficient Gathering in Sensor Information System):


It is an enhenced LEACH protocol. It is based on optimal chain for extending the
lifetime of the network. Here each node communicates only with the closest neighbor by
adjusting its power signal to be heard only by his closest neighbor. Each node uses the
signal strength to measure the distance to neighbor nodes,in order to locate the closest
node. After chain formation PEGASIS elects a leader from the chain in terms of residual
energy, who has to collect data from the neighbors to be transmitted to the base station.
As a result the average energy spent by each node per round is reduced. Unlike
LEACH,PEGASIS avoids cluster formation and use only one node in a chain to transmit
to the base station,instead of multiple nodes. This method reduces the overhead and
lowers the bandwidth requirements from the base station. Here also It is considered that
any node can communicate with the sink. It is also assumed that nodes are stationary. As
information may have to travel many nodes,so it has a limited scaling. So this routing
protocol is also not that efficient for our system.

TEEN(Threshold Sensitive Energy Efficient Sensor Network):


It is a LEACH based routing protocol for reactive network, having a smart data
transmission which saves power. Here nodes have dynamically reconfiguring capability.
At any cluster change time,the cluster-head broadcasts the following
parameters:Attributes and Threshold values(Hard threshold and Soft threshold). The
nodes transmit only if the perceived value is greater than the Hard Threshold(HT), or
value differs from the last transmitted value(SV i.e. sensed value) by more than the Soft
Threshold(ST). After transmission SV is set to the currently transmitted value. Here the
time-critical data reaches the user almost instantaneously. So this scheme is entirely
suited for time-critical data sensing applications. Again data transmission consumes
much more energy than data sensing. So even though the nodes are sensing
continuously,energy consumption in this scheme can be potentially much more less. But
the main drawback of this protocol is that if the threshold are not reached,the nodes will
never communicate. So the user will not get any data from the network at all, and even
not come to know if the nodes die. Thus this protocol is not well suited for our
application.

APTEEN((Adaptive Periodic Threshold Sensitive Energy Efficient Sensor


Network):
In APTEEN cluster-heads are decided in each cluster period and the cluster-heads have
to broadcast the following parameters:Attributes,Thresholds(Hard Threshold and Soft
threshold),Schedule and Count-Time(TC).Nodes transmit in time slot only if the sensed
value is greater than the Hard Threshold(HT), or value differ from the last transmitted
value(SV) by more than the Soft Threshold. If a node transmit for a maximum time TC ,
or if required by some sink,it transmits and after transmission SV is set to the current
transmitted value. By sending periodic data it gives user a complete picture of the
network. It can also respond immediately to drastic change,thus making it responsive for
time-critical situations. It also affords flexibility of allowing user to set the Count-
Time(TC) and the Threshold values for the attributes. Energy consumption can be
controlled by the Count-Time and Thresholds. The main drawback of this protocol is
that additional complexity required to implement the Threshold functions and Count-
Time. However,it is a reasonable trade-off and provides additional flexibility and
responsibility. So this protocol is perfect for our application.

SPIN(Sensor Protocol for Information via negotiation):


In this method network wide broadcast is limited by negotiation and using local
communication. There have two kinds of SPIN protocols:SPIN-PP and SPIN-BC
SPIN-PP(Sensor Protocol for Information via negotiation-Piconet-to-piconet
Communication):Data is described by a mete-data ADV(advertisement) message. A
node having the data sends this ADV message to the neighbors. If the neighbors do not
have the data,they can send a REQ(request) message. The node then respond by sending
the “DATA” packet. This process continues around the network. Nodes may aggregate
their data to ADV. The drawback of this protocol is that in a lossy network ADV may be
repeated periodically and REQ may also be repeated if not answered.
SPIN-BC(Sensor Protocol for Information via negotiation-Local Broadcast
Communication):Like SPIN-PP here also nodes having data send ADV and DATA,but
through broadcast. Broadcast makes it limited scaling as every node handles O(n)
messages. Data is updated throughout the network which is unnecessary in many cases.
Higher the number of nodes higher will be the power needed.

DD(Directed Diffusion)
A query or interest is broadcasted by a node or a sink. Query reaches the relevant sensor
sources. This sets-up an Exploratory Gradients. Once the data is available with a
source,it is sent back via a Reinforced path. Failing links or nodes are being bypassed. It
is robust and fault tolerant as the faults are bypassed. But this protocol has some
drawbacks like it will face “Hot Spot” problem near the sink. The periodic broadcast of
interest reduces the network lifetime. There is a trade-off between energy efficiency vs
robustness and scalability. Complex data aggregation may lead to an expensive node.
MCF(Minimum Cost Forwarding):
Data is sent to a sink from a node through the optimal path where the cost is minimum.
The data sending node first broadcast an ADV(advertisement) message to all the
sinks,and get answers from all the sinks and create cost-fields. The Back-off timer is
calculated proportional to the cost per sink. Needed informations are then sent through
slop. If no ACK(acknowledgement) get until timer expire,it resend an ADV. This
protocol seems suitable for our application,but it has some drawbacks. The time
complexity will be very high due to back-off. If there have many number of sinks,cost
table will be very high. So the cost-field set-up time will also be very high. No load
balancing is there. So it is not applicable to our system.

RW(Random Walk):
Here the nodes are assumed to be located at Cubic Grid Junctions. There have three kind
of RW topology.
RW-RSG(Random Walk-Regular Static Graph):Its a local communication. The co-
ordinate differences (dx.dy) are found using Distributed Bellman Ford. For every
node,the probability of moving on X and Y are computed by the diagonal to the
destination. On each node move to a adjuscent one on X or Y using that probability
adjust near end. All paths together draws a straight “Banana”.
RW-ISG(Random Walk-irregular Static Graph):It is irregular due to some dead nodes. It
is similar as RW-RSG. If one node is missing it will go the other with a probability P=1.
If both are missing,go to the neighbor whose B-F distance to the destination is strictly
smaller than the current node. This will create a detour,but can be optimized by not
getting to the node.
RW-DG(Random Walk-Dynamic Graph):Nodes may sleep or wake,so graph will be
dynamically adjusted. It is same as RW-ISG,but when a node changes state,the one hop
neighbor change B-F levels and possibly trigger further level,so the distance will
change. It is concerned with the delay in propagation updates and it is sensitive to
inaccuracies in levels.
The major drawback of this topology is that it may not be practical as nodes are assumed
to be located at Cubic Grid junctions

RR(Rumor Routing):
For many applications,no need for a shortest path;any arbitary path will work. Here the
nodes are densely distributed with bidirectional links. The nodes are considered to be
stationary and the communication is localized. When a node get some event or
query,movement on the net is done by several agents,trying to walk straight randomly.
Every node maintains a list of neighbors and events to know how to get to the reporting
node. An agent coming from an event updates all the nodes it visits and an agent coming
from a query,searches for the way to the reporting node. There is a high probability that
the lines will intersect. It would be applicable to our system only when the ratio between
the events and queries is inside a threshold where it is not attractive to a flood neither.
Optimal parameters depend heavily on topology,but can be adaptively tuned. It does not
gaurentee delivery. So we can not depend on this protocol.

6. Security Issues:
The nature of environmental application itself solves the problem of security.
The data obtained from these applications is generally freely available and is of concern
to everyone one way or another. The use of security protocols would only lead to energy
consumption thereby reducing network life time. Therefore, if using a technology
standard DASH7, the encryption modules should not be linked in the protocol stack if
unnecessary.

7. Fault-Tolerance:
Faults in Wireless Sensor networks can be due to a variety of reasons. These
include :
1. Node failure
2. Communication errors
The sensor network needs to be able to perform its tasks correctly throughout the
network-lifetime. In case of node failure this can be achieved by recognition of dead
nodes and changing transmission paths accordingly. APTEEN routing protocol is
suitable for this purpose.
For Communication failure, there has to retransmission capabilities built in to the
communication protocol through intelligent design of the MAC and Routing Protocols.
For Event-monitoring there need to be a synergy between the MAC and the routing
protocol. This means that there is a need to build a custom protocol stack for these
applications. For this a number of cross-layer issues need to be considered such as
MAC-LLC-Network Layers.
Among existing protocols, directed diffusion serves the purpose of dynamically
selecting the best path for data transfer. However, for time critical applications the time
to set up the gradients and reinforcement may be too much.

8. Software Design for the Application:

TinyOS:
This is an operating system specially used in Wireless sensor network. The application
being built is tightly integrated with the operating system during compilation. This
makes the whole software application specific. At the same time it minimizes the
software footprint by integrating only those components necessary for the application
while discarding the rest. It provides basic features like hardware interrupt handling,
libraries, compiler ,assembler. It executes only one program at atime
Use of a programming platform:
1. NesC : As a variation of the C programming language. It is a procedural
programming language where application is developed using interfaces and
components to represent real objects and data to be handled.
It uses the ncc compiler.
2. TinyGALS(Tiny Globally Asynchronous Locally Synchronous):This is a data-
flow style programming platform.

DATABASE: TinyDB:Only basic querying features are provided.

Software Development:
The code written on a suitable programming platform is the compiled ,then assembled
and linked. The Binary image is then downloaded into the device as firmware using the
tools provided by the manufacturer of the mote. This is generally a serial or parallel port
ISP programmer. A USB serial or parallel converter along with drivers may be
required. The Routing protocols can be tested on network simulators like NS3 and
OMNET++.
6. System Architecture:
Hardware and software Components:
Mote:

Application

TinyOS
DC-DC Radio
Converter MCU
Chipset
Memor
y
Battery
ADC

Sensors

Figure 1: The general Architecture of the mote

APPLICATION

ROUTING PROTCOL

MAC LAYER PROTCOL

PHYSICAL LAYER

Figure 2: Communication Protocol Stack


7. Conclusion:
The Design of a wireless Sensor network is highly application specific. For
Environmental non-time-critical applications the PEGASIS Data Gathering MAC
Protocol on top of a TinyOS on a Berkeley Mote

8. References:
1. Lakshmi Venkatraman,”Design Trade-offs in Wireless Sensor Network
System Development”
2. Zhongmin Pei , Zhidong Deng, Bo Yang, Xiaoliang Cheng “Application-
Oriented Wireless Sensor Network Communication protocols and Hardware
Platforms: a Survey”
3. Vamsi Paruchuri, Arjan Durresi, Leonard Barolli ,Energy Aware Routing
Protocol for HeterogeneousWireless Sensor Networks
4. A.Kruger, ”Wireless Sensor Networks ”,Lecture Series, University of Iowa
5. DASH 7, Rick , Wikipedia
6. Rajesh Yadav , Shirshu Varma , N. Malaviya ,“A survey of MAC Protocols
for wireless Sensor Networks”
9. Glossary:

TI: Texas Instruments


MCU: Micro-Controller Units
PEGASIS: Power Efficient Gathering in Sensor Information systems
LEACH: Low Energy Adaptive Clustering Hierarchy
TEEN: Threshold-sensitive Energy Efficient sensor Network
APTEEN: Adaptive Periodic Threshold-sensitive sensor Energy efficient networks
S-MAC: Sensor Medium Access Control
B-MAC: Broadcast Medium Access Control
WSN : Wireless Sensor NEtwork

S-ar putea să vă placă și