Wireless Sensor Networks for Fire Emergency and

Gas Detection
Amro Qandour, Daryoush Habibi and Iftekhar Ahmad
Centre for Communication Engineering Research (CCER)
Edith Cowan University, Perth, Australia
E-mail: aqandour@our.ecu.edu.au

Abstract—Environmental monitoring applications require adept C. Gas Monitoring

networking strategies and reliable communications to ensure the
integrity of their most fundamental operations including the sens- Each node is equipped with a variety of gas sensors including
ing duties, network-node interactions, and system resiliency to Oxygen (O2 ), Methane (CH4 ), Temperature, Carbon Monoxide
unpredictable events. In this paper, we propose a fire emergency (CO), Carbon Dioxide (CO2 ), Nitrogen Dioxide (NO2 ), and
and gas detection system based on WSNs for both indoor and
outdoor environments. The challenges which are addressed in this
flammable alcohols.
paper include: management of nodes, provision of algorithms, risk
modeling and analysis, and Over-The-Air-Programming (OTAP). D. GPRS Communication

I. I NTRODUCTION Mobile telephony is fast becoming the most dominant telecom-

munications platforms in many countries. General Packet Radio
Recent advances in sensor technology and wireless communi- Service (GPRS) is the technology built around mobile phones to
cation have inspired significant research interest in WSNs due to provide many services used in everyday life. The GPRS module
their promising potential to support an extensive range of appli- in the Waspmote is capable of providing the following services:
cations. Nowadays, WSNs have excellent integration capabilities
• SMS messaging or direct calls to specified phone numbers
towards a bigger range of communication protocols, application
used for alarm generation.
services, and software support. The Waspmote platform provides
• FTP server to download/upload files.
comprehensive support to a variety of communication modules,
• Internet access through TCP/UDP ports.
and boasts a sizable range of external modules which are used
in a variety of applications such as environmental, health, smart
metering, safety monitoring, etc.. In this paper, we propose a E. GPS Communication
monitoring system based on WSNs aimed at environmental and
The GPS receiver of the Waspmote provides information
safety applications. The system components include:
regarding the following localization parameters: Universal Time
Coordinated (UTC), latitude, longitude, altitude, ground speed,
A. Sensing Algorithms and direction.
Strategic sensing algorithms are employed to deliver efficient
sensor computations and also to provide organized operations for F. Twitter Integration
all associated devices of the WSN. In addition, they reinforce the
detection process using identification algorithms which determine This functionality is used for transferring received frames
a suitable alarm format linked with a particular event or an from the gateway node to specific twitter account(s). The aim
outcome. is to enhance awareness among community members through the
concept of shared monitoring. That is, people are aware of other
dangers that affect other members of the community.
B. Over the Air Programming (OTAP)
Remote access is adopted in order to enable users to monitor G. E-mail Server
various network activities such as retrieving sensor measurements
(gas concentrations), and network status parameters (power levels, An SMTP server is configured at the gateway to send regular
active nodes, etc.). These capabilities are possible with Over-The- e-mail updates regarding the WSN, and it is also used to send
Air-Programming (OTAP) techniques. alarms in critical situations.

978-1-4673-0390-3/12/$31.00 ©2012 IEEE 250

 II. BACKGROUND 6) Reliable data delivery: the DM protocol helps overturn
communication problems that exist in difficult terrains such
Research challenges [1], [2], [3], [4] which dominate WSN as dense vegetation, building walls, etc..
implementations in environmental systems are: 7) Node synchronization: low power sleep modes and syn-
• Real time self-organization. chronized wake up are supported by this protocol.
• Fault tolerant routing.
• Node localization. B. Ad-Hoc On-demand Distance Vector (AODV) Routing
• Sizable coverage demands (different terrains and topologies). This form of routing is used in the DM protocol to relay
• Long distance communications. data from the transmitting nodes to the base-stations or end-
• Self-healing requirements. devices [7]. In a mesh-network topology, only a single node is in
Different techniques are currently used to address some of the communication distance with the base-station, AODV routing is
challenges of WSNs. The emergence of modern protocols and used to map the transmission path using multihop algorithms, and
radio communication technologies have increased the number also maintains a list of previously used paths in routing-tables.
of outdoor WSN deployments. The sensor networks are used Path discovery is achieved by means of flooding the network
in a variety of applications such as detecting fire accidents [5], with broadcast messages until the final destination is discovered.
studying the air quality levels, and monitoring vegetation growth The path is determined for the delivery of data from the source
in forests. In this paper, we deal with the several issues that exist node to the base-station where it could take an N-amount of hops
in outdoor WSN applications including: before it is received. In the DM protocol, it is possible to specify
the maximum amount of hops allowed between routing nodes
• Reducing energy consumption in sensor nodes.
until the final destination, and once this value is exceeded the
• Delivering reliable environmental data from the source to
data get dropped. In addition, DM enables multi-route requests
the destination.
in order to select between various paths which are discovered; the
• Flexible network/node interactions using multi-protocol
source node broadcasts a Route-Request (RREQ) message which
communications.
is received by any node within communication range. Various
paths are determined based on the routes found by each node, then
A. DigiMesh (DM) Protocol a Route-Reply (RREP) gets transmitted back by the destination
The Digi-Mesh (DM) protocol is a proprietary owned stack node which contains a list of all the paths discovered. The source
which provides management and networking functions in the node receives the RREP and the path with the shortest Round-
WSNs. The main feature in this protocol is it allows mesh net- Trip-Time (RTT) gets selected to deliver the data.
work configuration and topologies. The mesh network topology
III. S YSTEM A NALYSIS
is extremely useful in WSN architecture as it enables longer
communication distances to exist between the nodes and the base- A. Outdoor WSN for Fire Detection
station through a routing technique known as Ad-Hoc On-demand
Distance Vector (AODV). Furthermore, DigiMesh incorporates
additional features including [6]:
1) Self-healing: node association or dissociation does not
affect the network performance, and therefore does not
cause a failure.
2) Peer-to-peer architecture: nodes are not subjected to obliga-
tory relationships as in parent-child constraints, and there is
no hierarchical based network architecture which is present
in the Zigbee protocol.
3) Routing capabilities: AODV routing is used in the DM
protocol to improve the distances between nodes and base-
stations.
4) Route discovery: this technique uses broadcast messages
to determine the route for transmitting data from the
source node to the final destination. Therefore there is no
intermediate requirement to maintain a routing map.
5) Selective acknowledgments: only destination nodes are able
to reply to route requests. Figure 1. Outdoor fire detection architecture.

251
 The system proposed in Figure 1 is an outdoor WSN which is The AODV algorithm uses routing techniques which enable the
used to detect fires. The sensor nodes are capable of detecting the sensor nodes to transmit data using hops as illustrated by Figure
toxic gases released in the smoke. In addition, they also monitor 3. The benefits associated with multihop networks include:
the Oxygen concentrations, temperature, and humidity levels in 1) Allows long distance communications.
the surrounding environment. The main features exhibited by the 2) Decreases the power consumption in the WSN.
outdoor fire monitoring system: -
• Power efficiency applications:
– The detection frequency is variable according to time
and date.
– Manageable tasks executions using dynamic battery
threshold monitor.
• Integrated GPRS module used to generate the alarms (SMS,
voice calls).
• Integrated GPS module for locating the exact coordinates of
fires. Figure 3. Multi-hop data network
• Suitable for deployment in various environments such as:
farms, rural areas (countryside), national parks, natural re- D. Over The Air Programming (OTAP) Architecture
sorts, etc.. We propose the use of OTAP functions for the following
B. Indoor WSN for Flammable-Gas/Fire Detection reasons:
The system proposed in Figure 2 is an indoor WSN which is • Upgrade nodes over the air, that is, wired cables are not

used to detect gas leaks, pollution levels, and fire. The main tasks needed.
performed by this applications include: • Resolve network issues which may rise from communication

• Centralized networking architecture.

channel selection, and modification of encryption keys (data-
• Instantaneous alarm is generated, and transmitted to the
payload).
• Organize programs which are stored inside the μSD module.
affected resident and the fire-emergency department.
• Flash the desired program onto microcontroller to provide
• Preventative sensing algorithms.
• Integration with Internet services.
different operations as needed by the application(s).
The OTAP shell is executed inside command-line environments
such as the Ubuntu terminal. The functions that are available
include scan network, retrieve node bootlist, send new program
files, and start new program. For example, by executing the
scan command we obtain a list of nodes which are operating
in a specific communication channel. The output response of the
“scan_nodes” includes the MAC address, Node Identifier (NI),
and the program name that each node is running. This can be
observed in Figure 4.

−− OTAP S h e l l O u t p u t−−

. / o t a p −s c a n _ n o d e s −−mode BROADCAST −−t i m e 10

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−
T o t a l Nodes : 2 − Time e l a p s e d : 15 s
0 − Node 0013 a 2 0 0 4 0 6 9 1 8 c 9 − WASPMOTE1 − p r o g 0 0 0 − READY
1 − Node 0013 a 2 0 0 4 0 6 9 1 8 b c − WASPMOTE2 − u p d a t e r − READY
−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−

Figure 2. Indoor gas/fire detection architecture. Figure 4. OTA scan network output

C. Routing Data with AODV E. Sensing Algorithms and Risk Modeling

We implemented a multihop network for its abilities to transmit We propose that algorithms should be used to direct the sensing
data between nodes in a Non-Line-Of-Sight (NLOS) environment. operations which are carried out by the nodes. The algorithms

252
are based on risk models, these are used to distinguish between The calibration process provides accurate interpretation of the
different emergencies, then call the appropriate handler function gas sensor measurements found in the environment. This is an
for each type that is found. The utilization of algorithms provide example how we achieved calibration for some sensors:
many advantages such as:
• Reduction in computational power in the nodes.
void get_Oxygen ( ) {
• Reduction in transmission power used by the radio modules. S e n s o r G a s . s e t B o a r d M o d e ( SENS_ON ) ;
• Event management based on highest/lowest priority tasks, S e n s o r G a s . c o n f i g u r e S e n s o r ( SENS_O2 , 1 0 0 ) ;
S e n s o r G a s . s e t S e n s o r M o d e ( SENS_ON , SENS_O2 ) ;
and adequate allocation of node resources depending on the O2 = S e n s o r G a s . r e a d V a l u e ( SENS_O2 ) ;
type of event. s p r i n t f ( Oxygen , " Oxygen=%d " , O2 ) ;
}

Figure 6. Oxygen sensor API implementation

1) SK-25 (O2 ) sensor: is used to measure the concentrations of

molecular oxygen in air between the 0% ∼ 30% range [8]. Oxy-
gen concentration under normal air varies between 21% ∼ 30%
depending on the geographical location. At maximum sensor
resolution, voltage output is approximately 0.6mV , this value
corresponds to 21% concentration of O2 in air. In addition, a
decrease of output voltage implies a decrease in O2 concentra-
tions. A linear relationship is formulated (Eq-1) that represents
the percentage of O2 with respect to voltage output of the sensor.

5 ∗ VOU T
Oxygen(%) = (1)
2

Figure 5. Fire risk levels

The risk model in Figure 5 is based on three different condi-

tions: Safe, Medium, and High risk. The risks are influenced by
environmental factors such as the surrounding temperature and
gas concentrations in air. The algorithms use the data from the
nodes to determine the type of event and its corresponding handler
function. Events are classified according to lookup tables which
affect the outcome of the handling-functions used by the node,
for example, the functions in Figure 5 are classified according to
the following events:
1) Condition-1: This event implies the area is clear of any
pollutants or fire risk, therefore there is no urgency to
transmit the measurements. However the node will retain
the data in the SD memory.
2) Conditions-2,3: In these events, there is a high risk of
pollution, or fire. The node powers the GPRS module and
transmits a warning signal (SMS, voice) to raise awareness
Figure 7. Sensors Architecture
of the situation.

IV. S YSTEM P ERFORMANCE 2) MiCS-2710 (NO2 ) sensor: is sensitive to NO2 presence in

air between for a given range of 0.05 ∼ 5 ppm. A buildup in
A. Sensor Calibration Process the source resistance is evident when the concentrations of NO2
The proposed sensing architecture is illustrated by Figure 7, increase. In [9], it is suggested that air resistance under normal
which includes processes such as calibration, API implementa- air condition is approximately 2200 Ω. The air resistance (Rs )
tion, and obtaining conversion formulas for numerous gas sensors. can be calculated from the output voltage (Vout ) from Eq-2. We

253
conducted our own tests in order to verify the air resistance, The results were calculated based on the number of packets
and also to obtain a match between theoretical and real life received or dropped at the receiver end. The nodes were setup as
measurements. follows:
  • Maximum number of allowed hops in all nodes is 7-hops.
VIN P U T ∗ RLOAD
RS = − RLOAD (2) • Maximum number of network route requests is 3-RREQ.
VOU T • Maximum number of packet delivery attempts is 2-attempts.

The output voltage fluctuated between 0.4 ∼ 0.5 V olts for a load The results indicate that 86% of all packets were delivered
resistance of 2.2 kΩ, and supply voltage of 1.8 V olts. This yields successfully to the destination. A total of 200 packets were
to the measured air resistance of 2828 Ω. It is concluded that this transmitted with maximum 4 hops per route were used.
resistance is within the accepted range of the theoretical values
C. Locating Nodes using GPS
of this sensor.
Global Positioning System (GPS) relies on satellite communi-
B. AODV Performance in Stationary Sensor Networks cation to provide location and time information for users. NMEA
(National Marine Electronics Association) is a communication
In this experiment, we investigated the reliability of AODV
standard used to specify the electrical requirements for GPS
protocol in routing sensor information using multihop transmis-
navigational equipment. Serial communication in the NMEA pro-
sion architecture to overcome the environmental barriers which
tocol allows data rates between 4800 ∼ 115200 bps. The NMEA
initially limited our communication range, and prevented peer-to-
standard specifies techniques for interpreting frame identifiers and
peer association between nodes and base-station. The layout of
their respective data type. There are several important definitions
the environment is illustrated in Figure 8.
in the NMEA standard [10] which include:
• NMEA-GGA: Providing fix data to provide 3D location and
accuracy data.
• NMEA-RMC: Provides location, speed, time, and date in-
formation.
• NMEA-GSA: Provides details on the nature of the fix such
as the number of satellites which are being used in the
current deployment.

Time : 1 1 4 0 5 6 . 9 0 0 ( hhmmss . s s s )
Date :180811 ( Day : Month : Year )
Latitude :3145.0915 ( ddmm .mmm)
Longitude :11546.3228 ( dddmm .mmm)
Altitude :64.7 ( meters )
Speed : 0 . 2 ( knots )
Course : 2 9 1 . 0 1 ( degrees )

Figure 8. AODV implementation in stationary WSN Figure 9. NMEA GPS data frame

We placed the sensor nodes at random locations and distances, The GPS data in Figure 9 conforms with the NMEA decimal
except for the transmitter node which had to be out of the com- representation. For example, a latitude of 3135.0915 (decimal) is
munication range of the base-station. Table I lists the obstacles equivalent to -31° 27’ 3.2934" (dd:mm:ss).
which exist around the testing environment.
D. Node Functions
Barrier Characteristic Value
Cement Walls Total Cement Walls = 5
Wooden Doors Total Doors = 1
In this experiment, we setup a WSN to detect the fumes exiting
Distance to nearest routing node Dmin =8 meters
Distance to furthest routing node Dmax =25 meters from the car exhaust system. The sensor nodes are equipped
with gas sensors such as CO, CO2 , NH3 (Ammonia), and H2 S
Table I
P HYSICAL BARRIERS IN AODV IMPLEMENTATIONS (Hydrogen Sulphide). The functions of the sensor nodes include:
1) Collect gas measurements.

254
 functions are used in emergency events such as fire accidents,
or increasing pollution levels. For example, let us assume that
a sensor node has detected a decrease in O2 levels below 18%.
Firstly, the GPRS module gets switched on and then waits for the
signal which connects it with the mobile network. Secondly, it
calls a specified mobile telephone number in which the recipient
is expected to become informed of the emergency. And lastly, the
information regarding this particular event are appended to a file
located in the μSD memory and then uploaded to an FTP server.
V. C ONCLUSION
Figure 10. Sensor measurements
In this paper, we presented two environmental monitoring
systems based on WSNs which are used for detecting fire and
2) Analyze toxicity. gas pollution in indoor/outdoor environments. We demonstrated
3) Apply risk analysis. the use of over the air programming (OTAP) in order to up-
4) Transmit alarm signals. grade/update the sensor nodes in a WSN. Also, we highlighted
5) Save data into memory. the procedures involved in the calibration of different gas sen-
Under normal conditions (no pollution risk), the nodes save sors. The experimental analysis cited in this paper included an
energy by switching off the radio, GPS, and GPRS modules. The investigation on Ad-Hoc On-demand Distance Vector (AODV)
non-transmitted information are retained on the memory card on- and Digimesh (DM) communication protocol in stationary WSNs.
board the sensor node. This procedure is highlighted in Figure 11. In addition, we provided a real-life experiment based on toxic
Furthermore, let us assume that a node has detected abnormalities gas monitoring in which we have outlined a number of features
in CO2 , O2 , and other fuel by-products as illustrated in Figure 10. including emergency response functions, alarm classification, and
In this event (Flag=15), an emergency task handler is triggered, interpretation of gas-measurements.
then an alarm is generated based on the type of task and level of
the risk involved (Alarm Level=3). R EFERENCES
[1] W. O. Kewei Sha, Weisong Shi, “Using wireless sensor networks for fire
rescue applications: Requirements and challenges,” in IEEE International
/ *−−−−−W r i t i n g t o uSD Memory−−−−−* / Conference on Electro/information Technology, 2006.
char *CO2 ; [2] S. A.-S. Ying Sun, Quan Yuan, “A bushfire monitoring and detection
char * H2S ; system for smart homes using zigbee technology,” International Journal of
char *NH3 ; Computer Aided Engineering and Technology, vol. 2, pp. 234–249, 2010.
char S e n s o r s [ 1 0 0 ] ;
[3] A. F. Chaczko, Z., “Wireless sensor network based system for fire endan-
v o i d Write_To_SD ( ) {
s p r i n t f ( S e n s o r s , "CO2:% s | Temp:% s | Power :% s \ r \ n " , gered areas,” in Third International Conference on Information Technology
CO2 , Temp , Power ) ; and Applications, 2005.
SD .ON ( ) ; / / Powers t h e SD module [4] R. Mustafa, J. Kim, J. Kelly, R. Le, J.-H. Kim, and W. Alexander, “Wireless
SD . m k d i r ( " S e n s o r s " ) ; / / Creates folder " Sensors " sensor network application for cost effective environmental monitoring,” in
SD . cd ( " S e n s o r s " ) ; / / Go t o f o l d e r S e n s o r s Southeastcon, 2009. SOUTHEASTCON ’09. IEEE, 2009.
SD . c r e a t e ( " G a s _ S e n s o r s . t x t " ) ; / / Creates the f i l e [5] J. Solobera, “Detecting forest fires using wireless sensor networks with
SD . a p p e n d ( " G a s _ S e n s o r s . t x t " , S e n s o r s ) ; waspmote.”
/ / Writes data i n t o f i l e
}
[6] Digi International, 11001 Bren Road East, Minnetonka, MN 55243,
XBee/XBee-PRO RF Modules, September 2009.
[7] L. Klein-Berndt, “A quick guide to aodv routing,” tech. rep., National
Figure 11. Saving gas measurements in external memory Institute of Standards and Technology, 2008.
[8] Figaro, “Figaro oxygen sensor sk-25f,” October 2009. Figaro SK-25 Oxygen
sensor datasheet.
E. Remote Connectivity [9] e2v Technologies, “Mics-2710 no2 sensor,” July 2008. MiCS-2710 NO2
Sensor Datasheet.
We have configured the GPRS module to send e-mails, make [10] Libelium, “Waspmote: Technical guide,” tech. rep., Libelium Comunica-
missed-calls, transmit SMS, and upload data to FTP server. These ciones Distribuidas S.L., 2011.

255