Using Wireless Multimedia Sensor Networks to Enhance Early Forest Fire Detection

Introduction

International congresses that focused on the environment, climate and air pollution; have confirmed unanimously that forest fires release fifth one (1/5) of CO₂ into the atmosphere, and burn a total of 48,341 hectares of green space in Oran City (Algeria) mainly in the M’sila Forest (see Figure 1 and Figure 2) during the summer season of the current year (C.E., 1987). According to FAO (“Planetoscope – Statistiques,” n.d.), forest fires destroy up to 700,000 hectares each year in the Mediterranean region, and more than 100,000 fires break out during the hot season. In some countries, experts record more than 20,000 wildfires per year. Globally, 130,000 to 150,000 km² (about 15 million hectares) of forest are lost each year. This means that every year, the world destroys forests equivalent to the surface of Belgium representing $4.1 trillion of free eco-systemic services, or $594 per person (“Planetoscope – Statistiques,” n.d.). From 1990 to 2010, the deforestation has represented four times the surface of a country like Italy. Every minute, 2400 trees are cut and in 2015, 18 million hectares of forest were lost according to (“Planetoscope – Statistiques,: n.d.).

Facing these horrific figures, the fight against the possible ignition of these fires becomes a vital interest of all concerned services. The study of various mechanisms and their improvement is a priority. Some of the traditional methods used by authorities to deal with this problem have a negative impact on society and the environment, such as watchtowers (“Forest Service Watchtowers,” n.d.), satellite imagery (“Forest Land Mapping, Forestry Mapping,” n.d.).

Recently, with the proliferation of the new technology of wireless sensor networks (WSNs) (Akyildiz, Su, Sankarasubramaniam, & Cayirci, 2002)(Sohrabi, Gao, Ailawadhi, & Pottie, 2000), several solutions have emerged (Alkhatib, 2014; Al-Dhief, Sabri, Fouad, Latiff, & Albader, 2017; Zaheer, Riaz, & Ahmad, 2016). The majority of these solutions have adopted a mono-modal approach in which scalar sensors such as temperature, humidity, wind speed, precipitation, etc. are used through the use of standard calculation formulas to determine the presence or absence of fire. The previous contribution (Bouabdellah et al., 2013), in which the authors have developed a fire detection system, is an example of these solutions. This system is based on the use of WSN by measuring periodically physical phenomena such as temperature, humidity which allowed us to compute a decision factor called FWI or Y depending on whether the detection method used is Canadian or Korean, respectively. Unfortunately, this solution did not help to reduce or eliminate false alarms and led to major practical drawbacks related mainly to the economical aspect.

By combining the scalar sensing technology with a richer acquisition technique using wireless multimedia sensor networks (WMSNs), especially video sensory data, the authors can achieve a multimodal innovative solution based on the cooperation paradigm to mitigate this problematic of false alarms. Accordingly, this would allow to a scalar sensor, which had just detected in its vicinity a sudden increase in temperature for example, to cooperate with the camera sensor through message exchanges to make sure visually that it is a real wildfire and not a false alarm. This is exactly the basic idea on which the authors build up this contribution; and for which the relevance of expected findings would be the consequence of the fact that the latter was developed, validated and evaluated experimentally.