In this article, we will be detecting a forest fire using an IoT sensor network and making a portable device with battery operation. Getting information about a fire hazard in a forest in time can prevent the forest fire from spreading. The sensor network can give the particular location of the fire as we already know where the sensors are installed.
Let’s discuss how we will detect the exact location of forest fire by IoT sensor network.
Purpose
1. Detect- Sensing the forest fire
2. Report – Report to the server
3. Locate – Locate the location of fire occurrence
4. React – Alerting the forest fire team
Architecture
Figure 1.
Figure 2.
Strategy
1. We can divide the forest in ‘N’ no. of areas as seen in Fig 1.
2. Each area is again divided into four cells- A, B, C, and D. Each cell will have several sensing units, which is shown in fig 2. The sensors are not evenly distributed; some sensors are in more quantity than another sensor.
The purpose of using the uneven amount of sensors:
Some environmental parameters do not change rapidly and can be sensed by fewer sensors in a large area.
3. The sensor unit consists of mainly three parts:
– Sensor unit
– RF (Radio frequency) Unit with (Inbuilt controller)
– Battery management unit
The sensor unit consists of several sensor nodes, as described in fig.2, and each node will have an RF module and a battery for a power source.
RF unit collects the data from the sensor nodes and transmits it to the Gateway. The Gateway listens for the incoming data from pre-registered nodes and forwards those data to the SIGFOX base stations. From those base stations, the data gets transferred to the SIGFOX backend, and in this above architecture, we are using our customized cloud. All the data will be shifted to the cloud via the SIGFOX callback function. Our customize cloud is supposed to do the major processing at the cloud end only; the cloud will make the decisions and provide the alerts to the team in case of fire detection.
Battery unit– It includes the battery, which we have described further in this report.
Some more details regarding the local data processing
We will use short-range data communication RF transreceiver nodes. The nodes will be communicating with the local Gateway, which will be present in each cell. Now, the Gateway at one end will collect the data from the sensor nodes and transmit the data to the SIGFOX module (The module details are described in the SIGFOX module section). This module will communicate over long-range to the base stations.
The reason to choose the short-range RF trans-receiver:
1. The power consumption in transmission and reception mode will be low, which will help in overall low power consumption.
2. Cost reduction.
SIGFOX module selection
The module is selected for the RCZ1 region, which includes European countries and covers the Middle East. The operating frequency band for these countries is 868MHz.
We can take AX-SIGFOX-API trans-receiver (up-link and downlink both) SoC which will fulfill our application requirement with the additional functionality.
Different sensor units with causes of false alarm and their prevention technique
Sensing Unit
We can detect the fire by using the following sensing unit.
1) Smoke and gas detector
Smoke detector – For early detection of fire, we can use two types of smoke detectors
-Ionization smoke detector – Fast response for open flame fire (high energy) as this type of fire produce small smoke particles.
-Photoelectric smoke detector – Fast response to a smoldering fire.
Gas detector The sensors have to be sensitive enough to detect even very low smoke concentrations. For this reason, gas sensors or a combination of gas sensors together with an aspiration system should be used. It consists of majorly CO2, CO, NO2, CH2, H2 gas sensors.
Causes of false alarm by smoke detector
Fog and cloud, Non-smoke objects like plants and animals.
Solution
We can also reduce the false detection rate by considering the result of the gas sensor and other sensors (heat and humidity).
2) Thermal detector
The smoldering and open flame raises the temperature of the surrounding. By using a temperature sensor, we can detect the change in the environmental condition.
We have two modes of temperature sensing
-Rate of rise – This will respond quickly to the high flame fire
-Fixed- temperature – This will respond to the slowly increasing smoldering or ground fire when the temperature reaches a pre-defined threshold level.
Cause of false alarm- Sunlight
Solution
By analyzing the temperature of the whole forest- On a sunny day, the temperature will rise uniformly all over the forest region, and all the sensors will give approximately the same reading.
By considering the season of that area.
By implementing the thermal detector in a shady area
3) Flame detector
Flame detectors optically sense either the ultraviolet (UV) and infrared (IR) radiation flames give off. Flame detectors are “line of sight” devices, and they are subject to being blocked by objects placed in front of them.
Mode of flame detector
We can use our detector in scanning mode to prevent the flame detector from being covered by any obstacle. In this mode, the device will rotate by 360°, and it stops when the signal is received. The detector will only alert when the signal persists for a specific set of times.
Causes of a false alarm: Sunlight and non-smoke object
Solution
By using such filters, which block solar radiation.
In the case of non–smoke objects, we can consider the result of other sensors also.
4) Wind speed detector and Humidity detector
Temperature affects the humidity as well as wind speed. By measuring both parameters, we can detect fire in that area.
Detection algorithm
The algorithm plays a crucial role in minimizing the false alarm rate in our system. Instead of relying on one or two sensor data, we will analyze the data from all the sensors, resulting in fewer chances of false alarms.
Gas and temperature detector data
S gas, Temp is equal to
[ (NH2,filtered KH2 + NCO,filtered KCO + NCH,filtered KCH + NCO,filtered KCO + NCO2,filtered KCO2)(1+NTemp,filtered KTemp) ]
S gas, Temp >= Threshold value => Activate temperature and gas sensor boolean
Humidity and smoke detector data
S Humidity, Smoke = (Humidity >= TH) (Smoke >= Ts) => Activate Humidity and smoke sensor boolean
TH = Threshold level of humidity
Ts = Threshold level of smoke
Flame detector data
S flame = (Flame data >=Tt) => Activate flame sensor boolean
Tt =Threshold time for flame detector
The measured sensor data for smoke, gas, temperature, humidity, and flame will be transmitted to our Gateway and fed into the detection algorithm/customized cloud.
The decision of “active alarm/no alarm” results from data from all sensors compared to the threshold value.
False alarms caused by Human activity
– Camping and stubble fire
Solution
Permission from the forest department authority
Other techniques to reduce false alarm
I. FWI system (Fire Weather Index system)
This system will measure the risk of a forest fire or determine the possibility of a forest fire. It is composed of six components that individually and collectively account for the occurrence of fire.
If FWI is measuring the high risk of forest fire, then there are fewer chances of false alarms, and in the case of low fire risk, we can double-check the trustworthiness of the alarm.
II. Animal as Biological Sensor – Two different detection methods can be implemented. These methods are the Thermal Detection (TD) method to measure instant temperature changes and the Animal Behavior Classification (ABC) method to classify sudden animal changes.
Decision-making unit
Customized Cloud
The personal cloud will accept the data from the SIGFOX backend through the callback APIs. The cloud will majorly perform these operations:
1. The cloud will collect the sensor data and log the data with the device ID, time, location, and RSSI strength.
2. The acceptable sensing range will be stored in the cloud, and the real-time sensing value will be compared and processed onto the cloud only.
3. Based on the backend calculations in the cloud, the cloud will alert the team in case of fire detection.
4. The cloud will be highly secured with TLS security enabled.
Maximizing Long life
Enclosure
We will use the ‘Ingress protection (IP) rating system to protect environmental effects like dust, dirt, wind, etc. Our enclosure is made of polymer, which can sustain high temperatures and other environmental extremes.
-Use of Faraday cage for blocking electromagnetic fields.
-The detector units are prone to dust, corrosion, and environmental extremes; for them, we can make a protective cap made of sintered metal which prevents them from soiling with dust and humidity
Sleep time
As the SIGFOX network can accept only 140 packets per day, we don’t need to sample the sensor data every second or minute. We will be enabling cyclic sleep or pin sleep on the sensor nodes. This will significantly reduce the power consumption of the device; this will maximize the device’s lifecycle.
Customized Cloud
We are doing all the processing at our cloud end only; the cloud will decide the case of fire occurrence and send an alert. This will significantly reduce the power consumption of our device.
Data transmission and battery life
As our SIGFOX device can only receive 140 messages per day, we have divided the total forest area in the ‘N’ section in our proposed design. Each section is then further divided into nine cells; each cell consists of 9 nodes. If we take an average, we can send 14 messages or alerts from each node per day.
-Per day, each node will only get active about 14 times and consume more power; the rest of the time, it will remain in sleep mode.
-If the total time for transmission of data from one node to the RF module is approx. 200ms. Then each node will remain active for about 3s (14*200ms) per day.
Active time of one node
-Per day-active time – 3s
-10 years active time – 3h approx.
-As per theoretical calculation, the average power consumption of each node is approx. 1.6W per day.
Calculation of battery capacity and discharge time.
For instance, Rated voltage of battery = 3.6V
The rated capacity of battery = 2000mAh
Discharge rate = 0.25C or 500mA (discharge rate as per each node current consumption)
Total discharge time in hrs. = Battery capacity(mAh)/Discharge current(mA).
T = 2000/500mA = 4h
So we can use a battery of 2000mAh or above, which suits best to our design. Now we will discuss the chemistry/type of the battery, which should have a long shelf life.
Battery selection
The suitable battery for our system is primary batteries as they have a long shelf life, extensive temperature range, and low self-discharge rate and cost. As our device’s maximum operating voltage is not more than 4v – 5V, we can use lithium-ion-based primary batteries as the Li-ion battery has a high energy density and long life.
Battery type
I. LiSOCl2
II. LiMnO2
III. XOL series batteries
All the above batteries have a shelf life of up to 10 years with a low self-discharge current and extensive temperature range.
Other technique
Energy harvesting system
To increase battery life, we can also adopt the energy harvesting system as a primary battery operating our system, so we cannot charge them by energy harvesting system. But we can operate our device directly through an energy harvesting system.
Below is the list of a few available systems that can be installed as per the available source of energy
i. Solar system – This system needs solar light to generate sufficient power.
ii. Wind system – Can be installed where frequent wind flow is there
iii. Electromagnetic energy system – Can be installed near transformer system which produces EM wave.
In brief description
How it works
For instance, if we take an electromagnetic (EM) energy system, some transducers convert the EM wave into an electrical current. We need to use the copper panel with the transducer for sensing the EM wave from the surroundings. The converted energy from the transducer is then used to power our device.
For another energy source like wind or solar energy, we need a different sensing system that will capture the solar energy/wind energy.
EM wave source
The source of energy for EM waves can be a transformer which emits EM waves. If we place our device near the transformer line, the energy harvesting system can capture the emitted EM wave from the transformer.
Switching unit
For powering our device with an energy harvesting system, we need a switching circuit. We can use an electronics switch (transistor) or electromechanical switch (relay), which suits our design best. This will switch our device from the battery to the energy harvesting system. The switching will take place only when our energy harvesting system stores sufficient energy to power our device.
Combination of energy harvesting system
For more power storage, we can use two or three energy harvesting systems simultaneously, like a combination of solar and wind systems or two EM systems.
This is how we can detect fire in forests using an IoT sensor network.
You may also like:
Filed Under: Applications, IoT applications, Sensors, Tutorials
