Wildfire Monitoring and Detection System Using Wireless Sensor Network: A Case Study of Tanzania

This paper proposes a wildfire monitoring and detection system based on wireless sensor network. This system detects fire by monitoring surrounding temperature, humidity and smoke. Once fire is detected, a warning message containing probable location of that fire is immediately sent to the responsible authority over cellular network. In order for the system to be more effective, communities living near forests or national parks can send warning messages through the same system to the responsible authority using their mobile handsets once they witness wildfire or illegal activities. For the system to be fully functional, the only requirement is the availability of cellular network coverage in forests or national parks to enable short message services to take place. The system prototype is developed using Arduino microcontroller, several sensors to detect temperature, relative humidity and smoke as well as wireless network connection modules. At the control center Telerivet messaging platform is used to design the messaging service. The experimental results justify the capability of the proposed system in detecting wildfire in real time.


Introduction
Wildfire is a kind of fire that occurs in the forests or reserved national parks. It starts as small fire but eventually turns into a huge fire due to some favorable con- work deployment for rural and forest fire detection and verification [7]. This system uses IP (Internet Protocol) cameras in a wireless network in order to detect and verify fire in rural and forest areas. The system consists of multiple sensors and access points to detect fire and send a sensor alarm through the wireless network to a central server. The central server selects the closest wireless cameras to the multi-sensor, based on a software application, which are rotated to the sensor that raised the alarm, and sends them a message in order to receive real-time images from the zone. This system requires better internet coverage in rural and forest areas. In Tanzania the forests and rural areas don't have good coverage of internet and thus if this system is deployed it won't function effectively and efficiently as intended. Another downside of this system is the use of IP cameras which consumes more energy and requires large bandwidth to be able to send the images. The overall cost of implementing this system for a developing country like Tanzania is too high.
In 2016, forest monitoring and wildland early fire detection system by a hierarchical wireless sensor network was reported [8]. The research work adopts a standard wireless sensor network approach using two levels for the nodes, central nodes and sensor nodes, and some optimization techniques to minimize power requirements. During network deployment, each sensor would record its delocalization based on GPS (Global Positioning System) information. Sensor nodes capture data from the environment that are uploaded to the central nodes, which transfer all the information to the system middleware. The transfer of sensor data from the central node to the middleware for this system is in terms of packets, so for it to function there has to be internet coverage which is a disadvantage if the system has to be implemented in Tanzania due to unreliable nature of internet services in forests and remote areas. Also to realize the full potential of the system it has to be integrated with all key players in firefighting operations such as fire brigades, communication systems, and aerial, coordination, and land means so the cost of implementation is high. Wang et al. in 2015 suggested forest fire warning system based on Geographical Information Systems (GIS) and WSNs [9]. This system is designed and de- in this research mainly describes the data collection from designed data acquisition system and then its classification. The wireless transmission of sensor node data is done using BTBee module and also artificial neural network approach i.e. support vector machine (SVM) is applied for classification of collected data. For someone to be able to deploy and implement this system has to be equipped with a lot of skills such as knowledge on support vector machine, multilayer perception polynomial and quadratic and radial basis function.
In 2008, Zhang et al. presented a forest fire detection system based on ZigBee wireless sensor network [11]. The topology structure of the system is an adaptation of a cluster-tree in order for the information path to take less memory space.
The hardware circuitry of the network node is based on a CC2430 chip. The environmental parameters detected are temperature and humidity. Communication between gateway and monitoring centre is done through GPRS module.
Processor chip used at the gateway is PXA255 from Intel Company. The downside of this system is requirement of internet within forest areas all the time for it to function.
Hefeeda & Bagheri in 2009 proposed forest fire modeling and early detection using wireless sensor networks [12]. This system is designed based on Fire Weather through giving out information about wildfire to the responsible people/authority as soon as it occurs to give them enough time to control it before it causes great damage. Another importance of this work is the involvement of communities around forests and national parks in conserving the forests by giving out information to the responsible people/authority by using their mobile handsets whenever they witness wildfire or illegal activities (such as charcoal burning, illegal logging, poaching etc.) taking place in the forests.
The rest of this paper is organized as follows: Section 2 presents equipment used and designing procedures of the system prototype, Section 3 provides prototype testing results and corresponding discussions and Section 4 gives out conclusive remarks of this research work.

Prototype Development
This system consists of three main parts: The Sensing Node, The Gateway Node and The Control Center. The sensing node is for detecting the presence of fire by monitoring surrounding temperature, humidity and presence of thick smoke.
These three parameters are chosen because wildfire is always accompanied with thick smoke, increase in surrounding temperature and decrease in relative humidity due to dryness of air. The gateway node acts as the coordinator, it receives warning notifications in terms of text messages from various sensing nodes and generates a wildfire notification message which contains probable location of where that fire is and sends it to the control center. The responsibilities of the control center is to receive wildfire notification from various gateway nodes as well as wildfire and illegal activities notification from villagers living around forest areas and automatically constructs and send warning messages to responsible people/authority. The system can have any number of gateway nodes depending on the size of forests or national parks that need to be covered. Each gateway node acts as a coordinator node to up to 10 sensing nodes. The distance between sensing node and gateway node depends on propagation power of the XBee module used; in this work the distance is between 100 -120 m. Figure 1 below shows system conceptual diagram.

Sensing Node Designing
The sensing node consists of sensors to sense fire parameters, ZigBee module for wireless transmission of data, microcontroller for controlling logic flow of events and the power supply to power up the system as shown in Figure 2 below.     Figure 3 below shows the flowchart at the sensing node.

Gateway Node Designing
The gateway node consists of the microcontroller for controlling logic flow of events, ZigBee module for receiving wildfire notification messages from the Microcontroller and ZigBee module used are the same as the ones used in the sensing node. Elecrow's GSM/GPRS shield was chosen because it is configured using SIM800C module from SIMCOM which is a Quad-band module supporting 850/ 900/1800/1900 MHz frequencies [16]. In Tanzania mobile cellular communication uses 900 and 1800 MHz frequency bands. The power supply to power the gateway node is the same as the one discussed in Section 2.1. Figure 5 below is the flowchart diagram which shows flow of events at the gateway node.

The Control Center
In this research work, the control center is designed using Telerivet Messaging Platform [17]. The main reason for choosing this messaging platform to design a control center is that Telerivet service provides an application that turns an Android phone into a low cost SMS (Short Message Service) gateway and by using this platform this work has avoided expenses associated with traditional SMS gateways like a need to purchase short code and negotiating with telecommunication companies [18].
A phone running android operating system is needed to act as a gateway for receiving and uploading received text messages into the Telerivet cloud servers.
A laptop to act as a Telerivet dashboard will also be needed where the rules to control the incoming messages will be written.
In this research, the gateway nodes and communities living around forests and national parks communicates with control center through short message services (SMS), so the main requirement is the presence of cellular network coverage in those areas that will enable communications through SMS. There is no need of having reliable internet coverage in forests for the system to function because sensing nodes, gateway nodes as well as communities around it do not send notifications over the internet.
An android phone which acts as a gateway can be placed anywhere far away from the forests or national parks in order to have access to reliable internet    If the message is not from a valid gateway node, the messaging system will check whether it is from a villager living around the forest by comparing the phone number of the sender with the contact numbers stored at the "Villagers" contact group on the messaging service. If the phone number of the sender is not available at the "Villagers" group the message will be discarded and the messaging service will wait for a new message to arrive. If the phone number of the sender matches one of the phone numbers stored at the "Villagers" group it means the message is indeed from a villager living nearby that forest. The messaging service will then scan the message contents to see whether it contain a word "wildfire or words "illegal activities". If it contains the word "wildfire" the system will extract location information from the message received and then it will construct a notification message with words: "THERE IS WILDFIRE AT MOSHONO WEST, PLEASE TAKE APPROPRIATE ACTION NOW BEFORE IT IS TOO LATE!" The words in italics being the location information extracted from the message received. The message will then be sent to the responsible authorities.
Otherwise, if after scanning it is found out that the message does contain the words "illegal activities", the system will extract location information from the message received and construct a notification message with words: "THERE IS ILLEGAL ACTIVITY TAKING PLACE AT MOSHONO WEST, PLEASE TAKE APPROPRIATE ACTION NOW BEFORE IT IS TOO LATE!" The notification message will then be sent to the responsible authorities.
If the messages received by the messaging service do contain words "wildfire" or "illegal activities" but the sender's phone number does not match any of the phone numbers stored at either "Gateway" contact group or "Villagers" contact group a reply message will be sent to that sender with the following words: "PLEASE REACH ME VIA MY OTHER MOBILE PHONE NUMBERS". All the events at the control center (the receiving of messages from either the gateway or a villager, the construction of notification messages for wildfire incidence or illegal activities incidences and the sending of those notification messages to the responsible authorities) will take place automatically without human intervention. The main reason is to speed up the notification process to the responsible authorities for them to take controlling measures as quickly as possible.

Experimental Set-Up and Results
During testing of this system, 4 sensing nodes, 1 gateway node, 2 mobile phone handsets (one to act as a villager, and the other to act as responsible authority), 1 android phone with Telerivet software installed and 1 laptop (where rules to control messages uploaded to the Telerivet cloud were written) where used. Sensing nodes and gateway node prototypes where deployed at a remote place in Arusha region. The location where the gateway node was deployed was named MOSHONO_WEST. Two people, one with a mobile phone acting as a villager and the other with a mobile phone acting as a responsible authority stood few kilometers apart from each other about 2 to 3 kilometers away from where the gateway node was deployed. The laptop and the android phone were placed in another place where there was access to internet. Figure 7 and Figure 8 below shows the system prototypes and the experimental setup respectively.
The sensing nodes where exposed to controlled actual fire and smoke made by burning some dry grasses and pieces of fire woods. The system was able to detect fire and immediately send warning messages to the mobile phone of a person acting as responsible authority. The person acting as a villager was asked to send notifications messages about fire or illegal activities to the control center and the system was also able to send respective warning messages to the mobile phone of a person with a mobile phone acting as responsible authority. (a) (b) Figure 9. (a) Text message sent to the responsible person after a sensing node has detected fire due to high temperature; (b) Text message sent to the responsible person after a sensing node has detected fire due to presence of smoke. Figure 9(a) and Figure 9(b) show screenshots of messages sent to responsible people/authority after the sensing node has detected fire incidence. The responsible people can use any kind of mobile phone; it can be a feature phone or a smart phone. They can also use any SIM (Subscriber Identity Module) card, from any telecommunication company of interest to be able to receive warning notification about wildfire. The messages received contain probable location of where fire is, in order to enable responsible people to locate it as quickly as possible, to give them enough time to control it. Figure 10(a) and Figure 10(b) show screenshots of messages sent to responsible people/authority after a villager has sent warning messages to the control  center after witnessing wildfire or illegal activities incidences in forests. The villagers can use any kind of mobile phone to send warning messages to the control center; it can be a normal feature phone or smart phone. In order to be able to send warning messages to the control center, they are free to use any type of SIM card, from any telecommunication company available in Tanzania. The messages sent to the responsible people/authority contain the probable location of where wildfire or illegal activity has been witnessed by the villager as well as the type of event witnessed (whether it is wildfire or illegal activity). This will enable the responsible people/authority to locate the location of wildfire or illegal activity witnessed as quickly as possible to give them enough time to take controlling measures.

Conclusions
In order to be able to effectively control wildfire, there should be mechanisms to detect it immediately as soon as it occurs and inform responsible people/authority so that they can take appropriate measures as quickly as possible. Any failure or delay to inform responsible people about wildfire on time will result in the fire becoming very huge to the extent that controlling becomes almost impossible leading to irreversible damages, distraction of infrastructures and other valuable properties, deaths of people as well as wild animals.
There are a number of systems around the world for monitoring and detection of wildfire but most of them can't be deployed and used in developing countries like Tanzania due to several issues such as absence of reliable internet coverage within forests, high power consumption as well as lack of technological infrastructures to support smooth operation of the systems. The contributions of the work proposed in this paper to already existing systems for monitoring and detection of wildfire based on wireless sensor network can be seen in three aspects; the sensing and gateway nodes can be deployed in areas where there is no reliable internet coverage, a scenario common in forests and national parks in developing countries like Tanzania. This is achievable through the use of short message services available in mobile cellular network. The second contribution is the involvement of communities living around forests or national parks in conserving the environment by giving them a means of reporting wildfire or illegal activities taking place in forests or national parks to responsible people/ authority by using their normal feature phones. This is achievable because almost every village household has got at least one feature phone due to having cheaper prices. The last main contribution is its ability to send warning messages about wildfire or illegal activities to responsible people/authority automatically without human intervention. There is no need to have people staying all the time at the control center to monitor wildfire or illegal activities taking place in forests and national parks.
This work can be further extended by designing an efficient solar or wind energy harvesting system to provide reliable power to the system.