INDOOR POSITIONING USING MEMS SENSOR

Abstract

In the outdoors having open air, a GNSS (GPS, Glonass) based system is able to estimate the position with a very high precision. However, in the inside of the building where the GNSS signals are very bad or almost not available, the accuracy from GNSS receivers are fallacious and are of no use in reality. The main motive of this project is the determine the position of the person using only the embedded mems sensors present in android mobile phones. Examples of such navigation are large hospitals, large office campus, big college campus, big shopping malls, big museums, large exhibition grounds, big train stations, etc.

The device used for this project is Samsung Galaxy S7 edge smartphone. This mobile has all the required sensor present so this device was chosen for this project. The Micro Electro Mechanical System (MEMS) based sensors present in these current mobile phones such as accelerometer sensor, gyroscope sensor, magnetic field sensor and barometer sensor allow navigation in GNSS shadowed areas. This project uses accelerometer sensor for calculating steps by an individual and distance travelled is determined using step counts, a magnetic sensor is used for detecting turns (90, 180, 270, 360 degrees) an individual makes using azimuth value derived from magnetic field generated by earth and a barometer sensor is used for detecting height changes for floor detection. Further, a tabular based map is used for pointing individual location which will also avoid false locations. The advantages of using sensors for estimating location are that they are very low cost, there is no installation time required and there is no need to make any major change with change in infrastructure of the building.

As this is a very big project so it was divided into sub projects. Floor Detection is already done in minor project 1. In major project I, position estimation is done using built-in step counter sensor and heading estimation is done. In this major project, user location is determined using an own-designed step counter algorithm based on accelerometer sensor as built-in step counter sensor was found to be not suitable and a full-fledged map of 10th floor and ground floor of Samsung R&D Institute is created using a grid based approach. A person is placed initially at fixed location and then as per person’s direction and number of steps, position is changed accordingly. Considering correct motion occurs with steps of the person, it turns out that with this approach we are able to achieve 98 % accuracy.