Improve Performance of Indoor Positioning System using BLE

Improve Performance of Indoor Positioning System using BLE Speaker : Bo-Ye Chi Advisor : Dr. Kai-Wei Ke

Outline v Introduction v Proposed Solution v Result v Conclusion v Reference

Introduction

Introduction v Indoor positioning is a network of wireless devices that work together to find and locate people or things inside the building. v The choice of BLE is based on a comparison between GPS, WIFI and BLE. v Global Navigation Satellite System (GNSS) including the Global Positioning System (GPS) give good accuracy and performance outside buildings, but unfortunately, they are not working probably inside buildings.

Introduction v WIFI v Supported in most human devices ，higher data transfer，long range v Consumes high power，low accuracy v BLE v Low energy consumption， low hardware cost ，high accuracy v Lower range and transmission power increase the effect of environment

Introduction BLE vs WIFI Characteristic Signal Rate Normal Range Transmission Power Energy Consumption WIFI 54 Mbps 100 m 20 d. Bm 100 - 50 m. A BLE 720 Kbps 10 m 1 d. Bm 15 m. A

Proposed Solution

System Model

System Model v The system consists of four BLE Nordic n. RF 52 modules, one of them acts as BLE central and the others acts as BLE peripheral. v The peripheral modules have to be fixed at specific prior-known locations, each of them is just advertising. Advertising data consists of unique id that identifies each peripheral. v During run time, Central device is moving and scanning for surrounding peripherals. Central device represents the entity that is needed to know its position.

System Model RSSI Measurements RSSI d. Bm ID 1 ID 2 ID 3 RSSI 0 -76 -65 -80 RSSI 1 -70 -78 RSSI 2 -75 -72 -77

RSSI Filtration v RSSI is affected by interference from the environment, which is mainly caused by reflections. v These reflections can lead to multiple paths or fading signals that cause incorrect RSSI measurements.

RSSI Filtration v Kalman filter is a set of mathematical equations that used for estimation of unobserved variable in a noisy environment. v It is a recursive algorithm as it depends on the history of measurements. v Kalman filter is used to increase the accuracy of the measured RSSI to avoid incorrect RSSI measurements leads to incorrect distance calculation.

RSSI Filtration v Kalman filter process consists of two stages: prediction and updating stage. v The predict phase uses the state estimate from the previous one to produce an estimate of the state at the current time. v In the update phase, the current priori prediction is combined with the current observation information to define the state estimate.

RSSI Filtration v the predicted state estimate v the predicted error covariance v Kalman gain calculate v updated state estimate v updated estimate covariance.

RSSI Filtration The RSSI curve with Kalman filter is smoother than the pure RSSI measurements. This will increase the accuracy in distance calculations.

Distance Calculation v The log-normal shadowing is the most popular model used in indoor systems for estimation the distance. This model is used to convert RSSI to distance. v This paper uses log normal model to estimate the distance between central device and three reference peripheral nodes. v RSSI is very affected by the surrounding environment, RSSI value is decreased with distance increasing.

Distance Calculation v equation (6) that shows and simplifies the relation between RSSI, distance and noise. v From equation (6), using reference distance equal to 1 meter, the log normal model can be expressed as in equation (7).

Position Estemation v Trilateration method is used to calculate the position of a target node. v The idea of trilateration depends on three reference nodes which their positions are known. v The intersection of the three circles is the position of the target node.

Position Estemation v In the proposed solution, the target node is the central device and the three reference nodes are the peripheral devices. v there are three equations in two unknown variables, this system is overdetermined. Least square (LSQ) method is used to solve overdetermined systems.

Result

Result

Result By measuring 50 RSSI values at different positions. For each peripheral device, n is calculated and then averaged over them all. Noise Calculations Peripheral Number 1 2 3 4 Average n Noise Factor 2. 82 2. 41 2. 57 2. 74 2. 63

Result

Result Position A is near peripheral 1 but it is far away from the other peripherals, so it has the biggest error value 0. 76 m. Position C has the smallest error 0. 39 value as it is in a suitable location for peripherals 3, 4 and 2. Error Accuracy Position A B C Average Error 0. 76 m 0. 45 m 0. 39 m 0. 53 m

Conclusion

Conclution v The position is detected using RSSI of BLE signals. RSSI values were in need to be smoothed using Kalman filter. Finally, position is calculated statically using log normal, trilateration method and LSQ. v In future, trilateration corner cases like non intersecting circles will be investigated. Also, there will be a research about localization of dynamic devices.

Refrence v E. Essa, B. A. Abdullah and A. Wahba, "Improve Performance of Indoor Positioning System using BLE, " 2019 14 th International Conference on Computer Engineering and Systems (ICCES), Cairo, Egypt, 2019, pp. 234 -237