Resource Monitoring for Wireless Sensor Networks using ANFIS

Abstract

Abstract Wireless sensor networks (WSNs) are usually a resource constrained networks which have limited energy, bandwidth, processing power, memory etc. These networks are now part of Internet by the name Internet of Things (IoT). To get many services from WSNs, we may need to run many applications in the sensor nodes which consumes resources. Ideally, the resources availability of all sensor nodes should be known to the sink before it requests for any further service(s) from the sensor node(s). Hence, continuous monitoring of the resources of the sensor nodes by the sink is essential. The proposed work is a framework for monitoring certain important resources of sensor network using Adaptive-Neuro Fuzzy Inference System (ANFIS) and Constrained Application Protocol (CoAP). The ANFIS is trained with these resources consumption patterns. The input to ANFIS is the resources consumption levels and the output is the resources consumed levels that needs to be sent to the sink which may be individual or combinations of resources. The trained ANFIS generates the output periodically which determines resources consumption levels that needs to be sent to the sink. Also, ANFIS continuously learns using hybrid learning algorithm (which is basically a combination of back propagation and least squares method) and updates its parameters for better results. The CoAP protocol with its observe option is used to transport the resource monitoring data from the sensor nodes to the cluster head, then from the cluster head to the sink. The sensor nodes runs coap server, the cluster head runs both coap client and server and the sink runs coap client. The performance of the proposed work is compared with LoWPAN network management protocol (LNMP) and EmNets Network Management Protocol (EMP) in terms of bandwidth and energy overheads. It is observed that proposed work performs better when compared to the existing works.