The increasing trend in the number of fires in cities today is due to the significant increase in industries, population boom and overcrowding of habitats. There is an urgent need to propose intelligent intervention methods in case of fire in cities (Yusuf, Navastara, & al, 2020) (Ceballos, 2020). However, Cameroonian cities, despite their urbanization plans, are not marginalized from these problems (Njume & Krah, 2020). Urban fires are more and more recurrent.
The use of classical methods such as satellite images, ranger posts and the establishment of fire safety organizations are expensive and inefficient fire detection methods for others (Vijayalakshmi & Muruganand, 2017). Solutions based on the Internet of Things have been the subject of several beneficial studies (Dinesh & Anette, 2019) (Fute & Tonye, 2013). They have indicatively focused on the prediction of natural hazards using supervised learning and multilinear regression (Mirza & Sheikh, 2020) (Dinesh & Anette, 2019). ). Disaster management in several scientific works has a limit, which is the non-confirmation of alerts in real time (Fernandino, George, Villanueva, & al, 2020) (Kavita, Amruta, Shruti, & al) (Wei-Ling, Ji-Yun, Chien-Shiun, & al, 2019).
Several works have used approaches such as the detection of the flame from a combination of motion analysis, the detection of flame colors and video images, a system with GPS for the prevention of alerts by GSM, a solutions smart home automation management of the connected kitchen for fire prevention through digital sensor values as well as video confirmation of the flame (Fernandino, George, Villanueva, & al, 2020) (Wei-Ling, Ji-Yun, Chien-Shiun, & al, 2019). Although these solutions are similar, it is important to note that there is too much dependence on the functionality of these applications which is often impacted by the deployment environment, the internal or external factors considered as well as the hardware used. In Cameroon, the main source of urban fires remains electric current, which is not addressed in previous work. However, monitoring electrical power continuously is a high-risk task for a human. Of our time, today there are several cloud services such as IBM bluemix, Carriots, Xively, Cayenne and ThingSpeak, which have a fundamental role for the creation and management of IoT applications (Cloud, 2021) (Gravelle, 2021) (Operator) (Cayenne) (Network, 2021). Thus, the results of this work highlight the efficient services provided by Cloud Computing which allow users to meet the challenges of data flexibility, scalability and viability, energy efficiency and optimization of the use of hardware and software resources (Faming, Chuantao, Wenjuan, & al, 2019) (Wan-Soo, Won-Suk, & al, 2020) (Aisha, Aisha, & Kashim, 2021).
Table 1
Comparative study of the functionalities of existing systems
System and Features | (Ravi & Subbachary, 2017) | (Fernandino, Thelma, & Napoleon, 2018) | (Wei-Ling, Ji-Yun, Chien-Shiun, & al, 2019) | (Hamood, Amgad, & Suliman, 2021) | (Haryanto, Lilik, Diana, Sahputra, & Laksono, 2021) | Our System |
Gas | Mq2, Mq5 | Mq2 | Mq4 | Mq2 | Mq2 | Mq2 |
Temperature | no | DHT11 | DHT11 | DHT11 | LM35 | DS18B20 |
Humidity | no | DHT11 | DHT11 | | no | DHT11 |
Flame | Yes | Yes | CSE0005-1 | Yes | no | ky-026 |
Local access | no | No | no | no | no | WampServer |
Remote access | Yes | Yes | IFTT | no | no | ThingSpeak |
Data analysis | no | No | no | no | no | Correlation study |
Prediction | no | Yes | no | no | no | Yes |
Presence detection | no | PIR motion sensor | no | no | Yes | no |
electric power | no | no | no | no | no | Yes |
The Table1 above presents all the experimental results. The particularity of our work is noticeable in terms of accessibility, efficiency and reactivity because our solution uses the ThingSpeak platform, which makes the data exchange faster and more reliable. Some works do not report sensor data values and focus only on the characteristics of the sensor and its calibration to assess the level of accuracy of the system, which does not inform on the functional capacity of these solutions (Ravi & Subbachary, 2017) (Fernandino, Thelma, & Napoleon, 2018). However, the values taken by the system in some works with a similar collection environment show alert values identical to our (Wei-Ling, Ji-Yun, Chien-Shiun, & al, 2019) (Hamood, Amgad, & Suliman, 2021). The sensors used for some solutions are insufficient for the idea presented (Haryanto, Lilik, Diana, Sahputra, & Laksono, 2021). (Haryanto, Lilik, Diana, Sahputra, & Laksono, 2021).