factors effecting positional accuracy of ibeacons

Factors effecting positional

accuracy of iBeacons

Chris Thomson (chris@codepilots.com)

Background

Indoor localization

?Where am I

iBeaconsSmall (50x30mm)Cheap (£8-30 per beacon)Range of up to 75mBattery poweredSupported by Apple (unlike RFID tags)

Distance to a iBeacon

Anon (2014) 'Adjust beacon range with Estimote’s new App and change UUID using Estimote’s new SDK', Reality matters [Online]. 28/1/2014. Available at http://blog.estimote.com/post/74816977799/estimote-app-v1-2 (Accessed 19/3/2014).

iBeacon localization

Estimote iBeacons, from http://estimote.com

The only maths: distance

Qiu, T, Zhou, Y, Xia, F, Jin, N, & Feng, L 2012, 'A localization strategy based on n-times trilateral centroid with weight', International Journal Of Communication Systems, 25, 9, pp. 1160-1177, Academic Search Complete, EBSCOhost, viewed 18 March 2014.Bulusu, N, Heidemann, J, & Estrin, D n.d., 'GPS-less low-cost outdoor localization for very small devices', Ieee Personal Communications, 7, 5, pp. 28-34, Science Citation Index, EBSCOhost, viewed 18 March 2014.

Overcoming Environmental factors

Bayesian learning (Öktem & Aydin, 2010)Real-time reference measurements (Hyo-Sung & Wonpil 2009)Clever uses of the radio (Wu et al, 2013)

Öktem R, & Aydin, E 2010, 'An RFID based indoor tracking method for navigating visually impaired people', Turkish Journal Of Electrical Engineering & Computer Sciences, 18, 2, pp. 185-196, Academic Search Complete, EBSCOhost, viewed 18 March 2014.

Hyo-Sung, A, & Wonpil, Y 2009, 'Environmental-Adaptive RSSI-Based Indoor Localization', IEEE Transactions On Automation Science & Engineering, 6, 4, pp. 626-633, Business Source Complete, EBSCOhost, viewed 18 March 2014.

Wu, K, Xiao, J, Yi, Y, Chen, D, Luo, X, & Ni, L 2013, 'CSI-Based Indoor Localization', IEEE Transactions On Parallel & Distributed Systems, 24, 7, pp. 1300-1309, Business Source Complete, EBSCOhost, viewed 18 March 2014.

Practical issues to consider

Differences in devices: Tablets, Smart phones and iBeacons

Noisy environmentsObjects in the environmentCost of deployment and maintenance Limited access to raw hardware

Solutions?

Experimental setup4mx3m Test ‘Lab’IPad Air (2013 wi-fi only model)3 Estimote pre-production iBeacons (early 2014)Core Location SDK and Estimote SDKTrilateration algorithm (wwnick &Austin, 2010)wwnick, Austin, J. (2010) 'Trilateration using 3 latitude and longitude points, and 3 distances', Geographic

Information Systems Stack Exchange [Online]. 26/6/2012. Available at http://gis.stackexchange.com/a/415 (Accessed 18/3/2014).

Technical stuffEstimotes x 3

Hardware: D3.2, Software: A1.9 Power: 4 dBm, Interval 200msEstimote API dated 19/02/2014Elevation: 210cm (wall) or 240cm (celling)

iPad AirElevation: 130cm (held) or 70cm (on stool)Held flat, home button to left of operator, hands under the device.

Estimote UniformityMeasured power levels (typical) at 1 meter.Receiver iPad Air, home button towards Estimote spot.

Typical RSSI values: -65 dBm, -71 dBm, -66 dBmLevels fluctuated approximately ±5 dBM

Stayed within this range with other WiFi and Bluetooth devices disabled.

Test ‘lab’

Operator and facing direction

Trilateration location

Beacon and reported distance

Test ‘lab’ obstructions

WiFi access point

Lamp shade

Mac with WIFI

Beacon on bookshelf,obstructed to left

Shelf belowbeacon

Window

Door

Door

Operator Rotation

iPad rotation

No Operator

Rotation of iBeacon

Bottom right beacon rotated only, stuck to wall, vertical orientation facing iPad, operator stood in front of iPad, iPad at 70cm elevation on stool.

Rotation of iBeacon

Bottom right beacon rotated only, stuck to wall, Horizontal orientation, base facing iPad

Rotation of iBeacon

Bottom right beacon rotated only, stuck to celling, horizontal orientation facing iPad, operator stood in front of iPad

RecommendationsCeiling mounting of Estimotes reduces variance and may also help with blocking by the operatorSignal transmission and reception are dependent on the orientation of both the transmitter and receiver. So this should be built into positioning models.Radio absorption by the operator is significant in determining distance. Again models should take account of the direction of the operator.Other radio interference does not seem significant in practice.

Future researchDoes other hardware suffer from the same directionality issues?Improve experimental protocol to eliminate operator issues, and take more accurate measurements.Collect further data on the variation in signal strength, to check for statistical significance of effects observed.Experiment with a predictive model to take observed effects into account.