Location detection of the mobile sensor with fingerprinting‐based cascade artificial neural network model using received signal strength indicator in 3D indoor environment. (29th September 2021)

• Record Type: Journal Article
• Authors: Tuncer, Taner; Erdem, Ebubekir
• Is Part Of: International journal of communication systems. Volume 34:Number 18(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-09-29
• Subjects: 3D -- cascade ANN -- fingerprint -- mobile node -- RSSI; Telecommunication systems -- Periodicals; 621.382
• DOI: 10.1002/dac.5000
• Languages: English
• ISSNs: 1074-5351

Abstract

Summary: The problem of localization of mobile sensor nodes has been extensively studied in the literature in recent years. In any localization technique, the aim is to determine the location of the sensor node of unknown location with low error. In this article, a cascading artificial neural network (ANN)‐based location detection algorithm is proposed, which detects the location of a mobile sensor node in 3D indoor environment. In a 3D indoor environment of 6 × 20 × 3 m 3, received signal strength indicator (RSSI) signals were collected using a mobile node with XBee sensor, and a fingerprint database was created. Cascade ANN system was trained using this database. Then, while a mobile node is in any location, RSSIs measured by anchor nodes are given as an input to the cascade ANN system, and the location of the mobile node is determined. Fingerprint steps 1 and 0.5 m were taken, and two applications were carried out in the article. According to the RSSI values taken from 100 different coordinates for the test, the total error was 3216 and 2838 cm, respectively. The average error is 32.16 and 28.38 cm. Abstract : Location determination was performed with fingerprinting‐based cascade ANN model in 3D indoor environment of 6 × 20 × 3 m 3 . The fingerprint steps 1 and 0.5 m were taken, and the RSSI values measured between the mobile node and anchor nodes were recorded in each database. According to the RSSI values taken from 100 different coordinates for the test, the total error was 3216 and 2838 cm, respectively. The average error is 32.16 and 28.38 cm.