Fire Detection Using Support Vector Machine in Wireless Sensor Network and Rescue Using Pervasive Devices

In the recent days, environment is polluted day by day due to various factors. One of the causes is smokes during the massive fires. Using Wireless Sensor Networks (WSN) fire can be detected earlier and also initiate the rescue operation before it becomes fire. In this paper, we will examine the possibility Support Vector Machine (SVM) for detecting the Fire which has large data sets. It is achieved by using the sketch of classes distribution which is obtained by using Minimum Enclosing Ball (MEB). This approach has many distinctive advantages on dealing with large data sets particularly forest fire data sets. Also, the Support Vector Machine has gained profound interest among the researchers because of its accuracy and the same is extremely important in this Forest Fire context as the cost of misclassification using a classifier is very high. Hence, this approach using multi class Support Vector Machine shows a higher accuracy in detecting the Forest Fire. The experimental result also shows a better accuracy in predicting the Forest Fire. Further, the rescue process will be initiated through the pervasive devices which are placed around the fire sensational area. This process will suppress the fire sensation and protect the field from the fire. We are including an architectural level procedure for implementing the rescue process.