RSSI and Feed Forward Neural Network (FFNN) Based Indoor Localization in WSN

In the advent of Internet of Things (IoT), Wireless Sensor Network (WSN) technologies play an important role in acquisition of different physical quantities for different applications. The Received Signal Strength Indicator (RSSI) based indoor localization is a well-known localization method used in WSN technologies due to its low complexity, availability and low energy consumption. In this research we explore the possibility of applying RSSI value based Feed Forward Neural Network (FFNN) jointly to identify the correct location of a moving object or a person, which is an important requirement of IoT-based Ambient Assisted Living (AAL) applications. We setup an experimental test bed for the acquisition of RSSI data remotely, which contained two types of nodes called beacon node and the mobile node. The ESP 8266 is used as the controller for nodes, which is based on IEEE 802.11 standard. The RSSI values from the beacon nodes will be sent to a remote server via Mosquitto Message Queuing Telemetry Transport (MQTT) broker, and then the RSSI values will be secondhand utilized by the FFNN supervised learning model that we developed at the remote server. Output of FFNN model gives the location of the object or person in two dimensional (2D) space. In the end of the research, the validity is checked by using the statistical assessment models and the results substantiate the significance of using supervised learning method in RSSI based indoor positioning.