Agent based Load Balancing in Sensor Cloud

Sensor Cloud is an emerging technology which combines Wireless Sensor Network and Cloud computing for providing seamless services to wide variety of applications. In sensor cloud, sensor network senses the environmental parameters, collects them and finally through gateway the information is stored into cloud servers. Each time unprocessed raw information coming from physical network is stored into cloud server resulting unbalancing in the overall network. Due to dealing of unprocessed information, node’s battery energy drains soon and even cloud server stores most of the unwanted information degrading its accuracy. Proposed work exploits agent based load balancing in sensor cloud using Neuro-fuzzy approach to make the propose load balance between physical nodes and also improve the network lifetime of the whole system. The redundant information is removed using Neuro-fuzzy approach and necessary information with greater accuracy is saved into cloud server. Simultaneously agents are triggered at the physical sensor network to collect the sensed information and submit to Cluster Head saving node’s energy. The Neuro-fuzzy combination balances the overall network load by rejecting similar information coming from multiple sensor nodes and take decision on the received information. The accuracy of the decision is improved by adding weights to the decision output which further improves system accuracy at great extent. Result shows that, there is an unpredictable improvement in network lifetime as well as information accuracy and small information with great accuracy is saved into cloud server thereby improving overall system stability through load balancing approach.