BLE for Indoor Positioning

Bluetooth Low Energy uses 40 channels on the 2.4 GHz band. Three of those channels are used for advertising messages. The received signal strength (RSS) of these channels from multiple BLE devices are used to determine relative distance and location. Transmission power for BLE devices can be set from 0 dBm to -40 dBm and the advertising rate can be configured up to 50 Hz. Typically, transmission power is set to less than -16 dBm and the advertising rate to less than 10 Hz, for the purpose of energy conservation. Zhuang 2016

Several proposed and/or existing solutions for indoor positioning require costly infrastructure and deliver a wide range of accuracy. WiFi is restricted by environment, power supply and mobile device support, while RFID is requires specialized equipment. Many technologies have limited accuracy due to fluctuating signal strength Wang 2016. Bluetooth Low Energy (BLE), also known as Bluetooth Smart, provides a low-cost, reasonably accurate, and zero infrastructure solution to this problem with the following advantages.

BLE is a low energy solution for indoor positioning. Depending on frequency and signal power, Bluetooth Low Energy devices can run on a coin cell battery for years.Reduced power consumption is achieved by very short duration messages, which can be either a data message or an advertising message. Advertising messages, which are broadcast at periodic intervals, are used for positioning.
Unlike other wireless technologies, BLE can maintain relatively stable signal strength and has a higher sample rate, which is important for accurate distance determination.
Bluetooth Low Energy beacons can be obtained for less than ten dollars apiece and can provide coverage from 25 to 100 square meters, allowing for relatively low cost implementations for large areas Wang 2016.
BLE is supported by most mobile devices. WiFi RSS is not supported by Apple devices, which severely limits widespread adoption Zhuang 2016.

Several algorithms have been proposed for calculating location with varying degrees of accuracy. One such algorithm uses Euclidean distance correction to achieve an approximate 1.5 meters of positioning error. This can be found in paper 3 of our literary review. Another algorithm that has been studied using Kalman Filtering and triangulation to reduce the positioning error to approximately .2-.4 meters. This is studied in paper 6 of our literary review.

These algorithms encompass several methods and produce a wide range of accuracy. In the following section we look at each algorithm in more detail, do a qualitative comparison, and recommend the best method for producing the most accurate results