Hindawi

Journal of Sensors
Volume 2023, Article ID 8063524, 12 pages
https://doi.org/10.1155/2023/8063524

Research Article
A Design and Development of the Smart Forest Alert Monitoring
System Using IoT

Murugaperumal Krishnamoorthy ,1 Md. Asif,1 Polamarasetty P. Kumar,2

Ramakrishna S. S. Nuvvula ,2 Baseem Khan ,3,4 and Ilhami Colak5
1
Department of EEE, Vardhaman College of Engineering, 501218, Hyderabad, India
2
Department of Electrical and Electronics Engineering, GMR Institute of Technology, Rajam, India
3
Department of Electrical and Computer Engineering, Hawassa University, Hawassa, Ethiopia
4
Department of Project Management, Universidad Internacional Iberoamericana, Campeche 24560, Mexico
5
Department of Electrical and Electronics Engineering, Faculty of Engineering and Architecture, Nisantasi University,
Istanbul, Turkey

Correspondence should be addressed to Baseem Khan; [email protected]

Received 10 September 2022; Revised 12 January 2023; Accepted 18 January 2023; Published 8 February 2023

Academic Editor: Qiang Wu

Copyright © 2023 Murugaperumal Krishnamoorthy et al. This is an open access article distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.

Forest is one of the main sources of living organisms. Its needs start from the human breath to usage of the wood. But due to many
reasons, area occupied by the forest is reducing every year. The reasons behind these environmental impacts are natural disasters
(forest fires), deforestation activities, and unlawful actions. Forest fire could be creating most serious threat to wild animals and
resources of human welfares. The primary phenomenon of the wild fire occurrence is circumstance hotness of the forest. The
dry and hot atmosphere caused the fire in the forest. The deforestation and smuggling activities are also tridents to the
available forest. The main consideration of this article is to detect wildfires in advance and protect forest resources from social
crimes through advanced sensor integration in the IoT (Internet of Things) environment. A smart forest alert monitoring
system has been proposed in this article to avoid forest mishap over by automated self-decision-making protective actions such
as parameter measures and alert and implementing the harm mitigation actions related to the hot temperature, humidity,
smoke, and smuggling of trees. All the sensors work as per the algorithm designed by the specific application of IoT (Internet
of Things). The accurate predictions of the forest fire events and ensuring the forest safety have been tested and verified by a
conducted case study on the real forest zone environment.

1. Introduction The accuracy of the fire prevention and identification

has observed low in satellite measuring system, due to the
Forest fires are a widespread threat in forests, wreaking havoc passing clouds and uncertainty with spot capturing of image
on wildlife and the environment. It may be avoided if a reliable pixel [4]. The lagging sensitivity of fire identification has
technology could be established in forest areas to detect fires been happened sometimes as late sensing after the forest gets
and warning to firefighting authorities, who will be available on fire. It will be mitigating by installation of spot sensors on
to take the rapid action [1]. The goal of this propose system the locations, and transforming the information without
is to create an IOT-based forest fire detection system that will delay has been observed as a field solution [5].
identify the fire and transmit an emergency instantaneous The recent work on this area has been analyzed twofold,
notice to the authorities by the system integrated sensor com- one is in terms of the importance of forest protection, and
ponents. Deploying a wireless sensor network in forest fire another one is a useful technology to make sustainable forest
prevention is identification and execution of the quick solu- activities. Many researchers have done work for fire preven-
tions to mitigate social harm of the forest [2, 3]. tion and detection, but not on the reasonable practical scale
 2

Table 1: State of the art related to wild fire forecasting [14, 16–19].

Main objective Measuring parameter Communication used Computing technology Limitation

Wildfire forecasting [14] Humidity and temperature ZigBee module Fog/cloud Limited range of connectivity
Early prediction of Humidity, temperature, and
Digi XBee module Cloud/fog Short range of transmission
forest fire [16] wind speed
Identification of
Smoke and light sensor Wi-Fi Not available Challenges with connectivity
tree cuts and fire [17]
Fire detection [18] Flame and temperature sensor LoRa Not available Lack of computing unit
High power consumed of
Fire detection [19] Not available ESP 8266 Wi-Fi Centralized server
remote unit
Prediction of fire including Temperature, humidity, Combinations of short- and Centralized and cloud-based
Proposed IoT-based SFAMS
deforestation activities smoke, and smuggling long-range communications data computing
Journal of Sensors
Journal of Sensors 3

Satellite

Uplink
frequency Downlink
frequency

4G/LTE link data transfer

Cloud server control

Forest monitoring region with connected
wireless sensor modules

Alert and monitoring team

Figure 1: Architecture of the proposed forest alert monitoring system.

of applications. A high version of the wildfire preventer has professional use of wireless sensor networks as a possible
been proposed to be implemented. The referred articles [6, explanation for the objective of a forest fire. To complete
7] have implemented a node MCU for transferring data the solution process, the suggested system relies on several
thought to use this because it has a lot of advantages. But sensors linked to it as well as data from the wireless trans-
an extra module for Wi-Fi is not preferable to designing a mission. These sensor data are sent to a ground station
power supply circuit. The obtainable projects on this field where they are analyzed by a small satellite in the system.
have more limitations with inbuilt detection and prevention For the previous finding of forest fire, the discourse strategy
techniques of smoke, temperature, and fire sensing and relies on data from wireless sensor reticulation [10]. The
detection. In later years, the satellite has reduced the burden usage of WSN in forest fire detection could be used in vari-
of forest monitoring systems [8]; that system consists of var- ous transmitting modules to a receiving end [11]. So, the fur-
ious detection and monitoring techniques. But it needs more ther implementation of the ZigBee technique provides
investment in the protection system infrastructure develop- additional access benefits to the forest area. The combina-
ment. Here, the main investment issues have been mitigated tions of transmitting the nRF module with the Node.js appli-
by using recent updates of IoT technologies such as con- cation platform have ensured the exact location of the forest
nected optical sensors and digital cameras. Here, it consists hazards to prevent wildfires easily [12]. It has been found in
of some sensors used to detect the fire and temperature rise a survey that 80 percentage losses caused due to fire and
and the digital camera to check whether the fire is present or promptly alert has reduce the mishap in early [13].
not. This method of detection technique provides many The previous work related to the wildfire study is having
advantages such as checking live 24 × 7 monitoring and con- limitations with analog metering and sluggish event capturing
tinuous data logging to the cloud storage for the betterment capabilities [14]. The transformation of digital sensors and
of the prediction process including the human interventions Internet-based integration of different sensors can provide
in the forest area. If anyone enters the forest, it shows an the solution for these issues. The missing of human to forest
alert message at the receiving end due to liked WSN tech- interaction is the main cause of the major wildfire accident
nique with the nRF module for transmitting data [8]. The [13]. The usage of cyberphysical system (CPS) helps to visual-
forest is regarded as one of the most important and neces- ize the events and prediction of the fire and executing the
sary resources [9], and that forest fires are a constant threat harm mitigation actions. The trustful data-based system oper-
to biological systems and environmental factors. Forest fire ations increased the accuracy of the system performance [15].
detection has become a critical issue in the presuppression The big data and cloud computing processes with artificial
phase, necessitating the urgent need to detect forest fires as intelligence help to accurate fire detection and prevention.
quickly as possible. This literature has been emphatic in its The recent field state of the art is shown in Table 1.
4 Journal of Sensors

Temperature
Solar
panel

Battery
MPPT Irradiation
Li-Po LoRaWAN
solar
module
controller

5V I2C
Fire, smoke
DC-DC SDI-12
converter

Power supply to the forest unit Human interaction

Figure 2: The elements of the remote access forest unit.

Send data to
receiving side

Fire extinguisher
Protecting
sensors Wildfire
Wildfire mitigation

Arduino controller

Figure 3: Signal/data controller of the forest unit.

The above state of the art clearly expressed the scope extinguisher to prevent the fire. The possible improved
of the effective wildfire monitoring system and smart alert solutions towards the identified field research gaps by this
with automated harm mitigating action process executions. study are listed as follows:
In this study, the proposed system is ensuring the high
performance in monitoring, quickness in the information (i) The summative attainment of the IoT aided com-
transferring, and execution of the social harm mitigation munication protocols, its technical requirements
actions. The temperature sensor starts its alert when the were designated, and performance accuracies were
temperature rises more than its threshold value. And also, verified with conducted wildfire case study
smoke sensor detects when smoke helps to spread the fire
like petrol and LPG. A fire sensor is the main one that (ii) The case study outcomes have expressed the signif-
detects the fire instantly when a fire occurs. These sensors icance of the Internet sensor modules with real-time
when any sensor gets alerted the whole system got alert data logging and tracking and monitoring to fulfil
and transfer an alert message to receiving end. The pro- the task allocated
posed system has the proficiency to activate the self-
actuators over by spot sensors; when cutting of trees, (iii) The implementation steps for a smart forest alert
switch on the water storage tank motors and CO2 fire monitoring system with self-decision-making
Journal of Sensors 5

ESP 32 CAM
Transmitting side module

LCD Temperature sensor

PIR sensor
Arduino UNO

Gas sensor

Vibration sensor

Submergible pump

Figure 4: RAU (remote access unit) for the forest alert monitoring system.

flame sensors, humidity sensors, and temperature measuring

elements including CCTV cameras. The collected data from
the various sensors have been transferred through a 4G/LTE
link to the cloud server. The real-time forest observations
have been monitored by the authorities in the control room.
The automated fire extinguishment and deforestation pre-
vention activity has been initiated by the smart alert system’s
activator operations based on the harm severity and priority
values generated by the field sensors. Figure 1 illustrates the
proposed architecture of the forest alert monitoring system
[14, 16–18].

2.1. The Elements of Forest Remote Access Unit (RAU). The

forest remote access unit’s power supply has been taken from
Figure 5: Smart alert RTU signal gateway for LoRa. the solar PV arrangement. The complete regain of the protected
forest is equipped with advanced wireless sensors such as flame
identifier, humidity, temperature analyzer, and vibration sen-
protective actions related to parameter measures
sors. These sensors observed the forest live and transferred the
and alert and implementing the harm mitigation
data to the available cloud platform through by 4G/LTE mod-
actions were established with installed model
ule. The uninterrupt of the system has been ensured by the
prototype
Li-Po type of battery energy storage setup. The liveness of the
In this article, Section 2 illustrates the proposed design forest has been monitored by the authority in the control
methodology. The complete system configuration and com- room through the LoRaWAN networking module [19]. This
ponent detail have elaborated in its subsection. The con- automated alert system has the additional proficiency of fire
ducted case study and observed outcomes and system’s extinguishing and prevention of deforesting action done by
performance justifications are presented in the Sections 3 preprogrammed Adriano controller operations. Each sensor’s
and 4. The further conclusion of the proposed work has been unit outcome signals have been classified as per the severity
expressed in the final section. and action priority levels. The sensor observed range and
controlled accuracy could be calibrated by machine
learning-based training and execution practice. Figure 2
2. Proposed Design Methodology shows the needful components regarding constant forest
remote access unit.
The advancement of the Internet of Things (IoT) has been The main objective of this work is to implement the fast
implemented for the application of forest alert monitoring and quick technique which helps to detect and prevent wild-
systems. The natural and manipulated hazards in the forest fires using a wireless sensor network. When wildfires occur,
need continuous monitoring to avoid fire and deforestation the temperature sensor detects the fire and it will alert the
activities. The fast and quick complaint responses are the system. When the system got alerted by the nRF module
additional provision here. The defined forest area regions and sends the data to the receiver side in parallel, it helps
were concentrated with the various wireless sensors such as to mitigate the wildfire by using a submerged pump water
6 Journal of Sensors

Initiation of all the sensors:

Temperature, Humidity, Gas/
Smoke from the environment,
Human availability

Threshold values
setting for different
sensors

Checking the sensed

NO values with threshold YES
values

Data transfer and

Sensor keeps
Buzzer alarm at control
sensing
room

Alerting the supportive

department over by hot
line communications

Updates: includes
Mitigating the
spot severity and
deforestation
action priority of
events
the event

Authority action
taken report
generation

Stop

Figure 6: Proposed deforestation-mitigation action flow.

sprinkler. The high range of fire could be controlled by a within minutes. So, the forest department has started trying
servo motor-operated fireball extinguisher. The future to detect things with the help of satellite pictures, but they
extension of these combinations might be used in various also have a delay in it. The protection system needs to detect
fire accidents and possible places such as industries, parks, the fire using electronic components like sensors, and it
and shopping malls. The auto mode operation of the alert could be placed in a tower in the forest. The human interfer-
system gives the alarm alert to the fire safety department ence in the forest is also monitored to avoid deforestation
and ensured to reach the fire location via the shortest route activity, so these sensors could be fixed at ground levels.
map. It helps with fire prevention in the early stages. As The CCTV and vibration sensors help for this protection.
per the forest protection code of behavior, the collected data This module is used to find the human images and infrared
has been received by the forest department. From the forest range-based detection [13, 15, 20–24].
department, additional supportive disaster management
actions could be initiated. The fire sensor protection range 2.2. Protection against Deforestation. When humans start to
starts from 180 degrees to 360 degrees to detect wildfires. cut the trees using tree cutters as an unethical based, the
In the summer seasons, due to high temperatures, the forest vibration sensors start to send the signals to the control
soil gets heated and the water evaporates from the soil. Trees room, and the control authority can initiate the steps
get dried and caught the heat, but we cannot instantly pre- towards the deforestation activity. After identifying the wild-
vent it because huge areas and fires spread over the forest fire, the prevention process has initiated through water
Journal of Sensors 7

1100
1000 1000 mm 900 mm 800 mm 700 mm 600 mm 500 mm 400 mm 300 mm 200 mm 100 mm
900
800
10-bit ADC counts

700
600
500
400
300
200
100
0
300

600

900

1200

1500

1800

2100

2400

2700

3000
Time (s)

Sensor 1 Sensor 4
Sensor 2 Sensor 5
Sensor 3

Figure 7: Typical fire sensor’s simulation performance.

Fire sensor
ESP-32 Pixel calibration calibration

Gas sensor
calibration
Vibration sensor Experiment setup for
calibration sensor calibration

Figure 8: Calibrations of sensor’s threshold values.

Table 2: Calibration of sensors as per event severity.

Severity levels
Sensor type
Level 1 (low) Level 2 (medium) Level 3 (high)
Atmospheric temperature ≥30°C ≥38°C ≥42°C
Air humidity ≤30% ≤20% ≤10%
Level of CO2 ≥36 ppm ≥2000 ppm ≥5000 ppm
Level of CO ≥10 ppm ≥30 ppm ≥50 ppm
8 Journal of Sensors

Table 3: Count of sensor used vs. data size and computational immediate alert messages and related descriptions of the fire’s
time. nature could be sent to the control centre through by RTU sig-
nal gateway. The nRF module helps to receive the data from
Data size Computational time RAU on time. The received data has to be quickly analyzed
Used sensor count
(in MB) (in sec)
by the trained algorithm and classified as the nature of events
10 3.5 24 as per the severity and step priority.
20 4.9 32 The action-taken steps and follow-up procedures will be
30 7.9 44 communicated with the related management authorities.
40 11.9 53 The continuous live location observation details have to be
50 14.9 61 updated every 5 minutes to quick up the action procedures.
Human detection and any vibration occur in the forest like
cutting trees; the unwanted intervention of humans is also
20 900 captured by the control room monitoring system. The min-
18 800 imal viable product of control room receiving assembling is

Solar radiation (W/m2)

16 700 illustrated in Figure 5.
Temperature (°C)

600
14
12
500 3. Case Study
400
10 300 In this section, the conducted case study in the Shamshabad
8 200 region, Hyderabad, India, has been discussed. The portion
6 100 (3000 square feet) of the forest area has selected and installed
4 0 5 numbers of needful sensors, such as temperature, humid-
ity, gas, vibration, and PIR sensors, that have been fixed with
12-04

13-04

14-04

15-04

16-04

17-04

18-04

19-04

the necessary level of heights in the selected locations. The

Date actions required for deforestation mitigation are defined as
Air temperature an action flow chart (Figure 6). All the sensors are energized
Solar radiation with a standalone solar PV power supply with a battery
backup system. Sensors have initiated and enabled the data
Figure 9: Typical weekly profile of air temperature and solar collections by Arduino and node MCU configuration. The
radiation as per cloud storage. different sensor data capturing capability concerning
reached period at the control circuit has been illustrated in
3 83.7 Figure 7. Most sensors can generate the signals with high
83.6 range 10 bit ADC counts at starting of periods. And its
Barometric pressure (bar)

2.5 83.5 values going to reach minimum count at the maximum

Precipitation (mm/h)

83.4
2 allowed range of periods. The further operation setup is also
83.3
83.2
ensured by driven codes of the Arduino controller when a
1.5
83.1
sensor gets alerted by fire; then, it follows that the flow chart
1 83 alerts the system and shows the alert message in the LCD
82.9 and then automatically started to water sprinkling which is
0.5
82.8 stored in the water tank. Based on fire severity, the con-
0 82.7 nected servo motors release the fire extinguisher balls into
the fire site. The PIR sensor detects the object based on infra-
12-04

13-04

14-04

15-04

16-04

17-04

18-04

19-04

red radiation ranges, and an extended version of an ESP-32

Date type of camera was used to capture the human’s entries into
Precipitation the forest. The complete gathered information has been
Barometric pressure transferred to the central server through 4G/LTE links and
available cloud platforms. The combination of hard and soft
Figure 10: Typical weekly profile of air precipitation and tool mix helps to develop the forest monitoring system. With
barometric pressure as per cloud storage. help of the headers, plug and functioning of Arduino have
been initiated. The PINs are written in white adjacent to
sprinklers and automated target fixed fire extinguisher CO2 the headers on the board. Pin 13 controls the onboard
ball throwers. The minimal viable forest remote access unit LED to control. The development of Arduino coding has
is illustrated in Figure 3. been done in the C++ programming language. The setup ()
and loop () void methods are required in every sketch ().
2.3. RTU Signal Gateway for LoRa Network Module. The for- No value is returned by a void type function. When Arduino
est monitoring control room consists of the data received from is turned on, the setup () method is called once, and then
the forest RAU (remote access unit as shown in Figure 4) and after, loop method () is called repeatedly. The initialization
emergency disaster management tools and corresponding steps are to be written in setup () and then want to run the
action procedure follow-up setups. When a wildfire is found, code in loop () sketch.
Journal of Sensors 9

17 90

16.5 85

Temperature (°C)
16

Humidity (%)
80
15.5
75
15

14.5 70

14 65
00:00

02:00

04:00

06:00

08:00

10:00

12:00

14:00

16:00

18:00

20:00

22:00

00:00
Time (hours)

Air temperature
Humidity

Figure 11: Day temperature and humidity data collected by RAU sensor at alert monitoring screen.

3540 1030

3520 1020

10-bit ADC counts

1010
3500
Voltage (mV)

1000
3480
990
3460
980
3440 970

3420 960
00:00

02:00

04:00

06:00

08:00

10:00

12:00

14:00

16:00

18:00

20:00

22:00
00:00

Time (hours)

Battery voltage
Flame sensor 5

Figure 12: Real-time RSU’s battery voltage and flame sensor data at alert monitoring screen.

Table 4: Event predictions based on various sensing data set 4.1. Calibrations of Sensor’s Threshold Values and Its
provided by the site sensors. Performance. Mainly, the measuring indexes of all the
Forest event Precision MSE RMSE MAPE Accuracy components have been defined, and their corresponding
values are to be embedded with Arduino control coding
No fire 1.000 0.002 0.044 0.2 0.992
for real site observations and monitoring. In the second
In fire 1.000 0.001 0.033 0.1 0.992 fold, the component’s performance and accuracy have
Post fire 0.990 0.005 0.071 0.5 0.992 been verified with the help of created trail events. The dif-
Unsafe forest 1.000 0.0017 0.042 0.01 0.992 ferent possible deforestation events have been created
manually, and corresponding data sets are to be stored
for training the system. The threshold values for the differ-
ent sensors and the nature of severity of events have fixed
4. Result and Discussion as shown in Table 2.
Generally, the complexity of system performance has
The evaluation of this proposed work has been carried out increased, when no sensors came to the role for data captur-
twofold. A first fold calibration of sensors has done with ing. These issues could be clear here instead of individual
standard test conditions at the laboratory test bench setup sensor data collection and transfer; the combined data hand-
(shown in Figure 8). In this stage, the calibration of the dif- ing process could be followed here. So-called data trans-
ferent sensors and their expected function has been defined ferred time duration has become narrow. Due to the
with its threshold value setting. compact form of the data set structure, it can be
10 Journal of Sensors

Figure 13: The summary of the sensor decisions with conditions of the system activators.

transferrable with a low range of network bandwidth. The nature to the sensors. The biometric pressure monitoring
computational time and expected data transfer packet size helps the system for event prediction and the follow-up mea-
concerning different count of alert sensors has been illus- sures toward deforestation. This observation has been illus-
trated in Table 3. trated in Figure 10.
With data monitoring and data acquisition, the pro-
4.2. Data Set Management and Event Prediction. The pri- posed forest alert system has the additional proficiency of
mary objective of the proposed work is a fast alert monitor- real-time monitoring. The continuous battery voltage and
ing system that includes event prediction and avoidance of flame sensor data set help the endurance of the system sta-
deforestation issues. Attainment of this task could be done bility towards uncertainty issue avoidance with the RAU’s
by the previous method of data mining and handling power supply. Figure 11 expresses the real-time typical day
towards the quick decision hopes including avoidance of site temperature and air humidity data at the control room
the false fire and unsafe conations of the forest. The con- monitor. Figure 12 illustrates the real-time battery voltage
nected temperature sensor and solar irradiation measure- and flame sensor 5 data at the forest alert and control room
ment units have been updating the site measured values to screen.
the cloud’s data storage platform for every minute interval.
The typical day observations of site temperature and solar 4.3. Proposed System Performance on Various Sensing
irradiation values are illustrated in Figure 9. This data obser- Parameters. The proposed forest alert monitoring system
vation gave hints regarding seasonal annual variations of the has the trust to predict the forest event exactly as it hap-
weather and event predictions as per post-event-registered pened without false alarm to the control authority. Due to
data sets. The solar irradiation value observation helps the the high level of accuracy predictions, the rescue team can
energy availability to operate the RAU independently with- do their service without any panic.
out conventional energy supply. There are many previous solutions available in this field,
The precipitation and atmospheric pressure are also but due to single sensor dependency, many possibilities of
important indexes to design the forest alert monitoring sys- fake calls and fault alerts happened. In this work, every event
tem in terms of avoidance of the false alarm and the present has conformed to multisensors and its functions are also
working status of the installed sensors and their perfor- trained with all site possible events. The cloud registry data
mance levels. Less precipitation is providing better working set-based prediction technology leads the system’s accuracy
Journal of Sensors 11

Table 5: Performance comparisons among the different wildfire prediction techniques.

Notification False Fire localization Information on fire Detection device

Wildfire prediction methodology Cost
delay alarm error behavior mobility
Animals as mobile sensor
Long High High Yes Mobile High
techniques [25]
Radio acoustic-based techniques [26] Small High High No Fixed Medium
Image processing techniques
Long Medium High No Fixed or mobile High
(camera based) [27]
Neural and fuzzy logic-based
Small Medium Low Yes Fixed Low
techniques [28, 29]
Proposed sensor-based techniques
Small Low Low Yes Fixed Low
(wireless sensor network)

to a high level of 0.992. The mean average percentage error puters and Electrical Engineering, vol. 101, article 108046,
(MAPE) of the predicted data set is between 0.01 and 2022.
0.5%. The precision level of forest events is between 0.99 [2] S. Lahlou, Y. Moukafih, A. Sebbar, K. Zkik, M. Boulmalf, and
and 1.0 values. Table 4 provides the forest event predictions M. Ghogho, “TD-RA policy-enforcement framework for an
based on on-site sensor data sets. Figure 13 illustrates the SDN-based IoT architecture,” Journal of Network and Com-
relations among the different sensors’ decisions with the puter Applications, vol. 204, article 103390, 2022.
operating states of the activators, which were used to avoid- [3] M. E. Karar, A.-H. Abdel-Aty, F. Algarni, M. F. Hassan, M. A.
ance of deforestation activities. The performance compari- Abdou, and O. Reyad, “Smart IoT-based system for detecting
sons among the different wildfire prediction techniques are RPW larvae in date palms using mixed depthwise convolu-
illustrated in Table 5. tional networks,” Alexandria Engineering Journal, vol. 61,
no. 7, pp. 5309–5319, 2022.
5. Conclusion [4] S. Sperling, M. J. Wooster, and B. D. Malamud, “Influence of
satellite sensor pixel size and overpass time on undercounting
The deep observation of the forest is the way to save that of cerrado/savannah landscape-scale fire radiative power
from natural and manmade calamities. The present sustain- (FRP): an assessment using the MODIS airborne simulator,”
ability goals and policies highly depend on green forest Fire, vol. 3, no. 2, p. 11, 2020.
development. That mission could be simply attained by this [5] D. M. Szpakowski and J. L. R. Jensen, “A review of the applica-
proposed design and development of a smart forest alert tions of remote sensing in fire ecology,” Remote Sensing,
mentoring system. Due to the IoT advancements and data vol. 11, no. 22, p. 2638, 2019.
analytic techniques, the efficient forest fire fighting infra- [6] J. Á. Martín-Baos, L. Rodriguez-Benitez, R. García-Ródenas,
structure could be proposed here. Its effectiveness in the fast and J. Liu, “IoT based monitoring of air quality and traffic
and quick alert of forest fire and avoidance of unsafe forest using regression analysis,” Applied Soft Computing, vol. 115,
article 108282, 2022.
events has proved over by conducting a case study on a real
forest zone. The forest unsafe events are also predicted here [7] J. V. Sorribes, J. Lloret, and L. Peñalver, “Analytical models for
with a high level of accuracy rate of 0.992, and the minimum randomized neighbor discovery protocols based on collision
detection in wireless ad hoc networks,” Ad Hoc Networks,
range of MAPE values is in between 0.01 and 0.5 without
vol. 126, article 102739, 2022.
false alerts to the control authorities. This multi-sensor-
based forest safe assurance system framework will be used [8] J. Li, G. Zhou, A. Chen et al., “Adaptive linear feature-reuse
network for rapid forest fire smoke detection model,” Ecologi-
in other application sectors such as industries, parks, and
cal Informatics, vol. 68, article 101584, 2022.
shopping malls for their safety from fire hazards.
[9] Y. Tarazona, V. M. Mantas, and A. J. S. C. Pereira, “Improving
tropical deforestation detection through using photosynthetic
Data Availability vegetation time series - (PVts-β),” Ecological Indicators,
vol. 94, pp. 367–379, 2018.
Data will be available on request. For the data-related queries,
kindly contact Baseem Khan ([email protected]). [10] T. M. Basu, R. Karthik, J. Mahitha, and V. L. Reddy, “IoT
based forest fire detection system,” International Journal of
Engineering & Technology, vol. 7, no. 2.7, pp. 124–126,
Conflicts of Interest 2018.
The authors declare that they have no conflicts of interest. [11] A. A. A. Alkhatib, “A review on forest fire detection tech-
niques,” International Journal of Distributed Sensor Networks,
References vol. 10, no. 3, Article ID 597368, 2014.
[12] P. Rai and M. Rehman, “ESP32 based smart surveillance sys-
[1] S. Chen, J. Ren, Y. Yan, M. Sun, H. Fuyuan, and H. Zhao, tem,” in 2019 2nd International Conference on Computing,
“Multi-sourced sensing and support vector machine classifica- Mathematics and Engineering Technologies (iCoMET), pp. 1–
tion for effective detection of fire hazard in early stage,” Com- 3, Sukkur, Pakistan, 2019.
12 Journal of Sensors

[13] S. Tozlu, M. Senel, W. Mao, and A. Keshavarzian, “Wi-Fi [28] B. C. Ko, J.-H. Jung, and J.-Y. Nam, “Fire detection and 3D
enabled sensors for Internet of Things: a practical approach,” surface reconstruction based on stereoscopic pictures and
IEEE Communications Magazine, vol. 50, no. 6, pp. 134–143, probabilistic fuzzy logic,” Fire Safety Journal, vol. 68, pp. 61–
2012. 70, 2014.
[14] A. Tsipis, A. Papamichail, I. Angelis, G. Koufoudakis, [29] O. Satir, S. Berberoglu, and C. Donmez, “Mapping regional
G. Tsoumanis, and K. Oikonomou, “An alertness-adjustable forest fire probability using artificial neural network model in
cloud/fog IoT solution for timely environmental monitoring a Mediterranean forest ecosystem,” Geomatics, Natural Haz-
based on wildfire risk forecasting,” Energies, vol. 13, no. 14, ards and Risk, vol. 7, no. 5, pp. 1645–1658, 2016.
p. 3693, 2020.
[15] J. Yun and S.-S. Lee, “Human movement detection and identi-
fication using pyroelectric infrared sensors,” Sensors, vol. 14,
no. 5, pp. 8057–8081, 2014.
[16] H. Kaur, S. K. Sood, and M. Bhatia, “Cloud-assisted green IoT-
enabled comprehensive framework for wildfire monitoring,”
Cluster Computing, vol. 23, no. 2, pp. 1149–1162, 2020.
[17] R. Niranjana and T. HemaLatha, “An autonomous IoT infra-
structure for forest fire detection and alerting system,” Interna-
tional Journal of Pure and Applied Mathematics, vol. 119,
no. 12, pp. 16295–16302, 2018.
[18] N. C. Gaitan and P. Hojbota, “Forest fire detection system
using LoRa technology,” International Journal of Advanced
Computer Science and Applications, vol. 11, no. 5, 2020.
[19] H. D. Ahlawat and R. P. Chauhan, “Detection and monitoring
of forest fire using serial communication and Wi-Fi wireless
sensor network,” in Handbook of Wireless Sensor Networks:
Issues and Challenges in Current Scenario's, pp. 464–492,
Springer, Cham, 2020.
[20] T. Brito, A. I. Pereira, J. Lima, and A. Valente, “Wireless sensor
network for ignitions detection: an IoT approach,” Electronics,
vol. 9, no. 6, p. 893, 2020.
[21] A. Divya, T. Kavithanjali, and P. Dharshini, “IoT enabled for-
est fire detection and early warning system,” in 2019 IEEE
International Conference on System, Computation, Automa-
tion and Networking (ICSCAN), pp. 1–5, Pondicherry, India,
2019.
[22] B. Son, Y.-s. Her, and J.-G. Kim, “A design and implementa-
tion of forest-fires surveillance system based on wireless sensor
networks for South Korea mountains,” International Journal
of Computer Science and Network Security (IJCSNS), vol. 6,
no. 9, pp. 124–130, 2006.
[23] G. S. Amacher, A. S. Malik, and R. G. Haight, “Not getting
burned: the importance of fire prevention in forest manage-
ment,” Land Economics, vol. 81, no. 2, pp. 284–302, 2005.
[24] Y. E. Aslan, I. Korpeoglu, and Ö. Ulusoy, “A framework for use
of wireless sensor networks in forest fire detection and moni-
toring,” Computers, Environment and Urban Systems, vol. 36,
no. 6, pp. 614–625, 2012.
[25] Y. G. Sahin, “Animals as mobile biological sensors for forest
fire detection,” Sensors, vol. 7, no. 12, pp. 3084–3099, 2007.
[26] Y. G. Sahin and T. Ince, “Early forest fire detection using radio-
acoustic sounding system,” Sensors, vol. 9, no. 3, pp. 1485–
1498, 2009.
[27] Q.-x. Zhang, G.-h. Lin, Y.-m. Zhang, X. Gao, and J.-j. Wang,
“Wildland forest fire smoke detection based on faster R-
CNN using synthetic smoke images,” Procedia Engineering,
vol. 211, pp. 441–446, 2018.