IoT-based Carbon Monoxide (CO) Real-Time Warning System Application in Vehicles

. The project is about develop a system and application for detect the presence of Carbon Monoxide(CO) in car, since recently there are many cases of drowning while sleeping in car due to inhaling CO. The build system are able to detect the presence of CO and provide warning about level of CO to the users. It uses Blynk application to monitors level of CO inside the vehicle, MQ-9 gas sensor as the input sensor, ESP 8266 as medium to send data to the application via IoT-based and the level concentration of CO is displayed on the LCD in real-time displayed. For the output, it has 3 different condition based on the level concentration of CO. This project has been testing in six different situation. Based on the result, ambience air and in car with open window situation have lowest of CO level


Introduction
There were many cases report of teenagers drowning while sleeping due to inhaling CO in the car whose engine was turned on. For examples, one of the teenagers found dead in car at Kampung Jaya Baru, Kinabatangan, Sabah on 11 October 2018 [1]. There are also cases reported from the abroad on 19 March 2020 and 5th October 2020 which also involves teenagers dying in cars due to CO poisoning [2,3]. While the latest case happens in Butterworth, Penang involving, four students taking a nap inside the car resulting three of them have die due to CO poisoning [4].
CO is of one element carbon and one element oxygen which is a colourless, odourless and tasteless toxic gas [5]. CO is toxic to humans when found in high concentrations as it causes disorders in the blood. This case occurs because there is no specific device to assist in monitoring the level of CO and warn from time to time (real-time) in vehicles especially for cars. By respect to all this problem occurred, a system or device that are able to monitoring the presence of hazardous gasses are much appreciated. Hence, this research is aimed in developing a system that can give information as well as monitoring level of CO inside the vehicles such as a car and sending a notification if there have any dangerous situation occurred. This device or system is believed can give benefits for users, mostly for teenagers as well as for their parent for monitoring purpose since it can be access through smart phones.
From the Table 1 below, there are few symptoms that can detect based on CO level of concentration and length of exposure. This also used to consider limit of the sensor and for making decision of the output. The risk to health of CO poisoning also can be measured by time weighted permissible exposure limit. For example, Occupational Safety and Health Administration (OSHA) have recommended time weighted permissible exposure limit (PEL-TWA) of 50 PPM. Besides that, National Institute for Occupational Safety and Health (NIOSH) have recommended exposure limit (REL-TWA) of 35 PPM. Furthermore, Iran Health Ministry have recommended occupational exposure limit (OEL-TWA) of 25 PPM. This also need to be considered in making decision for the output for this project [9], [10].

Flowchart of the project
The process of this project is shown below in Figure 1 while Figure 2 shows the LCD shows the CO level and it condition which is medium. The project is starting by collecting data via sensor that have been used in the system. After that, the data is sends to the Blynk application for monitoring purpose and the system will make future action based on the level CO that have been collected. Next, the data stores in Google Sheets every 5 seconds. The data that have been stores can be used to analyse the project functionality. Lastly, the data will be clustering or grouping by using PCA.  Figure 1. Flowchart of the project. Figure 2. The LCD shows the CO level in medium condition.

Flowchart of the system
Overall, the system consists of three operations which are the input sensor, processing unit, and output. The system uses MQ-9 sensor as the input sensor in which to detect the level of CO gas and passes the output voltage to the Arduino Uno for process. The output voltage is converted to concentration of CO in PPM by using program code that has written. The LCD acts as a monitor to display the concentration of CO level in real-time which is in every 5 seconds. As referring to Figure 3, the system starts by reading the CO level. Next, the microcontroller will do a calculation of CO level. After that, the reading of CO level will be display using LCD in real-time and all the data will be sent to the application. The concentration of CO will be determine using the application. As a result, it will trigger the certain output such as trigger the alarm, light the bulb, lower the car window(presented by servo motor) and sends notification based on the condition that has been decided. More details, the output consists of three conditions based on the concentration of CO as per described below: -•If the PPM of CO lowers than 25 PPM, then there is no action from the system. •If PPM of CO is between 25 PPM until 199 PPM, then it will trigger alarm only.
•If the PPM is 200 PPM and above, then it will trigger alarm, light the bulb, lower the car window that presented by the servo motor and notify the users using application. The alarm will keep warn continuously and the notification will send to the mobile phone via Blynk application. Figure 4 shows the connection of the components that consists of NodeMCU, MQ-9 gas sensor, buzzer, LED, LCD, servo motor and resistor.

Results and discussions
The CO level that sense by the MQ-9 sensor has been displayed on the LCD. On the LCD, it not only shows the CO level but it also shows the condition for situation based on CO level whether it is low (0PPM until 24 PPM), medium (25PPM until 200 PPM) or high (more than 200 PPM). The condition whether it is low, medium or high will affect the output part which is alarm (buzzer), bulb of inside car (led), lower window (servo motor) and notify using application as shown in Table 2. • Notify using application

Blynk Application
For the monitoring, it can be access by using application which is Blynk application that supported by Android mobile phone. In this application, it will show CO level in real-time same as it shows on the LCD. Besides that, this application also shows the graph of CO level in real time and also for a certain period that can choose by the user. This monitoring system that can access through this application also can be monitor by third-party users such as parent or other family member. Furthermore, data collected is stored in Spreadsheet which can be access by using mobile phone or using other devices such as tablet or laptop. This stored data can be used for references or evidence if there are any undesirable incident cause by the high CO level in the car. The monitoring system by Blynk application and store data by Spreadsheet are based on IoT mechanism that very useful for humans especially for the users.

Analysis of the result
The result of this project has been collected in different situation such as in ambience air, in car with open window, in car with close window, in car while the air conditioner is turn on and smoke that are produce from fuel combustion of the car exhaust. From the Table 3 below, there shown all result that have collected in different situation that been stated.  Figure 5 and Figure 6, in ambience air and in car with open window situation it has same lowest reading of CO level. However, for the highest reading of CO level, in ambience air has higher CO level compared to in car with open window situation. This is because it is affected by the environment of the places. It is possibly when data collected in ambience air, there are several vehicles passing by in the area.
For in car with close window and in car while the air conditioner is turn on, both situations give same result for lowest and highest of CO level. Both situations have reading of CO level a little bit higher compared to in car with open window situation. This is because when in confined space, the CO gas that enter before the window close will be trapped inside car. But this does not endanger the health of the users. For the four condition that has been interpret before this, it can be classified as low condition of CO level.
On the other hand, smoke that are produce from fuel combustion of the car exhaust at distance 30 cm, have mid-range CO level reading. This can be categorized as medium condition of CO level. In this condition, health of users may be risk when long exposed in that situation based on World Health Organization (WHO) CO level exposure limits [8]. Last but not least, smoke that are produce from fuel combustion of the car exhaust at distance 5 cm have the highest reading CO level compared to all situation. This situation can be classified as high condition CO level. If this smoke enters inside the car because of leaking from the exhaust or the engine, this can be harmful for the drivers and passengers. By referring to data collected in Table 2 also, it can be related with maximum exposure allowed by OSHA for CO level. The maximum CO level that allowed by OSHA is 35 PPM as in Table 1. According to this maximum CO level that allowed by OSHA, the smoke that are produce from fuel combustion of the car exhaust in distance 5 cm is dangerous. The CO level in that situation is higher than 200 PPM for both lowest and highest reading of CO level from data collected in field test. This can cause fatigue, mild headache, nausea and dizziness as stated in Table 1 if the smoke enters inside the car. It will also be worsened if the driver or users is not aware of this so it can cause the users drown and die due CO poisoning.  Figure 7 shows graph about time sampling against PPM for every situation except for smoke of fuel combustion of the car exhaust at distance 5 cm and at distance 30 cm. Then, Figure 8 show graph about time sampling against PPM for every situation. Both figures illustrate the collected data in sampling data of every 5 seconds. In these two figures, clearly show that the smoke from fuel combustion at 5 cm and 30 cm have higher reading compared to other situations.
Therefore, this project is very useful to overcome this dangerous situation as well as can reduce the number of humans died causes of CO. This project has a system that can detect and give warning about the level of CO to the users. In this situation, the system will send signal to alarm (buzzer) to trigger it. Besides that, lamp (led) inside the car will automatically turn on and the window of the car will also automatically lower. This can notice and attracting the attention of those in surrounding about this situation. This will make it easier for people to help if the situation is detrimental such as the driver fainting, sleeping or drowning in the car. In addition, Blynk application that has been used in this project will also send warning to their parents or other family members. So, their parent or family members can monitor the CO level which can threaten life or not. This is one of advantages for this project that used IoT-based system which is have integration between the devices and the system by using the Internet.

PCA
PCA were applied in order to verified the ability of this CO system in differentiate or clustering these 6 situations. Figure 9 shows the scatter plot of PCA of every situation that involved in this project. The first PC is yield 71.82% accounted for the biggest variance while the second PC is 28.18%. It clearly shows the situation of smoke fuel combustion of 5cm and smoke fuel combustion is clearly separated with each other as well as separated with ambience, air condition turns on, car close window and car open window. Ambience, air condition turns on, car close window and car open window look like overlapping each other since it can be considered very low in CO presence.
However, the PCA plot were tested by eliminating the data of smoke fuel combustion of 5cm and 30cm. Next, Figure 10 shows the scatter plot of PCA of every situation except for smoke of fuel combustion of the car exhaust at distance 5 cm and at distance 30 cm where the first PC is 86.64% and the second PC is 13.16%. By comparing the result, it clearly shown that each situation is clearly grouped. This result indicates and proved that the CO system are able to differentiate every situation by respect to the value of CO. This is meant that the MQ-9 gas sensor with the build system are able to be applied in detecting the presence of CO in different situation.