Device-based Real-time Single User Indoor Localization Using Internet of Things

Abstract

Background: Localization is an important area of implementation of internet of things based Wireless Sensor networks. Outdoor user tracking is possible using the global positioning system; however, the global positioning system accuracy decreases in indoor environments. The wireless sensor network is used on the internet of things based technology for localization to overcome this problem. Objective: The wireless sensor network-based indoor localization is categorized into two categories; range-based and range-free localization. In range-based localization, first, it computes the relative distance, and then it calculates relative coordinates. In range-based techniques, distance and position are calculated. The internet of things-based localization used the range-based and range-free techniques of a wireless sensor network to localize any object. The Light-Dependent Resistor-based localization work has been proposed in previous research. In this research, the light-dependent register traces a person’s entry /exit event as the person switches ON/OFF lights of the building. It was not a sufficient effort to localize a person using a light-dependent register. The two PIR have been used in each room, along with one RFID-based approach proposed in this study to overcome light-dependent records. Methods: An Indoor scenario has been considered in this study. The hardware setup has been configured to trace the user. When a user enters inside a building, he will switch on the lights, and the light sensor records the light intensity and gives some reading. The difference in light sensor reading (before switching ON the light and after switching OFF the light) gives some clue about a user in an indoor scenario. Nevertheless, if the lights of many rooms remain switched ON, the user cannot be localized using the above method. Two passive infrared sensors in each room and one radio frequency identification-based model have been proposed in the present study to sort out this ambiguity of light sensors. Implementing the single-user localization using light dependent resistor sensor becomes erroneous if a person moves from one room to another and the lights are turned ON/OFF. Moreover, the LDR-based model is affected by sunlight during the daytime. Results: As the implementation of passive infrared sensors and one RFID-based localization technique gives efficiency 93 %, the light-dependent register-based localization system provides an efficiency of 35%. Conclusion: The light-dependent register-based approach is prone to more errors because the user may enter multiple rooms while the lights of each room remain ON. A passive infrared sensor and radio frequency identification-based approach for a single user indoor localization has been proposed to overcome this problem. The proposed techniques are easy and cost-effective for implementation. The results show that the proposed method provides better localization accuracy than the light-dependent sensor-based technique for single indoor localization.

Plant Health Prediction and Monitoring Based on convolution Neural Network in North-East India

“…The user/device localization and home automation have been done using IoT devices such as passive infrared sensors, radio frequency identification, light-dependent register, Bluetooth low energy, and smartphone. The result shows that the passive infrared sensor along with radio frequency identification-based IoT localization gives better accuracy of 93% than light dependent register [2]. Smartphone-based localization gives better accuracy than Bluetooth low energy-based localization [4].…”

“…IoT devices may be used for various applications such as user localization [1,2] home automation and to control of electric appliances in the office/home as per user movement [3,4,5]. The user/device localization and home automation have been done using IoT devices such as passive infrared sensors, radio frequency identification, light-dependent register, Bluetooth low energy, and smartphone.…”

“…The proposed architecture for CNN is given below in table 1. As shown in table 1, conv2d is performed in the input image with 32 filters having input shape of (224,224,3) with the activation function as relu, then max pooling operation is performed with pool size (2,2) and Finally flattening into a 1dimension array has been done. In the first and second hidden layers, 224 neurons are used with relu activation function.…”