Fog computing based information classification in sensor cloud-agent approach

•Raw information is saved directly in Sensor Cloud that impacts latency and accuracy.•Fog computing analyzes and classifies information before saving it in sensor cloud.•Agents are triggered to optimize the energy in prolonging the network lifetime.•Random forest and genetic algorithm used to classify information with low variance.•Optimized information with great accuracy is saved into sensor cloud. Sensor Cloud has emerged as a rising technology in removing the barriers of Wireless Sensor Network. It increases the lifetime of the sensor network by storing all the sensed information into cloud server instead of saving in node’s memory. Yet sensor cloud has some important issues like latency, information classification and accuracy which still need an attention to improve overall performance of the sensor cloud system. Fog computing is a problem solver for sensor cloud in removing latency and carrying out computational tasks in faster way. Fog computing is acting as a middleware between physical network and cloud, located at the edge of the network for fast computation and quick response. The proposed work utilizes the functionalities of sensor cloud and fog computing to classify and save the information in better way along with minimizing the latency issue. The Random forest classifier along with Genetic Algorithm is utilized for classifying the information and saving only required amount of information into cloud server with priority. Agent paradigm is included to save the energy of the physical sensor nodes and also to perform quick analyzing and classification of information at the fog server. The result shows that proposed work is working far better than conventional methods in terms of classification accuracy, latency, packet delivery ratio, energy consumption and network lifetime.