Arizona State University Department of Computer Science and Engineering CSE 535 : Mobile ComputinG Indoor localization in kids

1
Arizona State UniversityDepartment of
Computer Science and EngineeringCSE 535
Mobile ComputinGIndoor localization in kids
network

• Project Report prepared by
• Shayok Chakraborty
• Weijia Che

2
Agenda

• Introduction
• Data Collection
• Proposed Approach
• Results
• Comparison
• References
• Questions

3
Introduction

• Goal To localize kids in a room using a fixed
  set of beacon nodes installed in the walls
• Challenges
• - Random movements of the kids
• - Presence of obstacles in the room

4
Data Collection
5
Data Collection
6
Proposed Approach - Weijia

• FingerPrinting
• Points near each other have Similar RSSI
• Manhattan distance
• a, basic algorithmassign mote to the anchor
  with the smallest distance
• b, improved sampling --drop samples exceeds
  mean(RSSI)-1
• c, improved calculationstatic/dynamic
  threshold algorithm
• d, improved calculation k (k4) nearest
  algorithm
• e, weighted calculation static/dynamic
  threshold/Knearest (did not improve accuracy)

7
Proposed Approach - Weijia

• f, proposed calculation check and feed
  backKnearest (improves accuracy for points find
  the right anchor point and degrades accuracy for
  points did not)
• Results
• a. avg. ERROR1.4342m
• b. avg. ERROR1.1229m
• c. avg. ERROR0.62435m/0.96065m
• d. avg. ERROR0.91275m
• e. avg. ERROR0.86755m/0.8475m/1.03447m
• f. avg. ERROR 0.99745m

8
Results

• Defense
• The algorithm is based on the assumption
  that the mobile host already find the right
  nearest anchor point and based on this to improve
  accuracy.
• (It did not generate better result in our
  experiments coz we only have two motes and failed
  to find the correct nearest point most of time)
• Further Improvement
• combine this algorithm with static threshold
  algorithm instead of Knearest algorithm.
• introduce power levels into the calculation
  to improve accuracy

9
Proposed Approach - Shayok

• Triangulation
• Power(P) varies with distance(d) according to
• Pj kdj-a
• or, dj (Pj / k)-1/a
• j beacon number ( 1,2)
• a attenuation exponent
• k a constant
• a and k are calculated from the reference
  points

10
Triangulation - Illustration

• Find distances using
• power equation. Then
• use Euclidean distance
• measure to find the
• position of the node

Beacon 2
Node to localize
Beacon 1
11
Results
Average error 0.8864 m
12
Explanation

• Radio propagation indoors is very chaotic due to
  the presence of obstacles, reflection etc.
• More beacons can increase the accuracy of the
  method

13
Comparison
14
References

• 1 P. Bergamo, G. Mazzini . Localization in
  Sensor Networks with Fading and Mobility
• 2 Zang Li, Wade Trappe, Yanyong Zhang, Badri
  Nath. Robust Statistical Methods for Securing
  Wireless Localization in Sensor Networks
• 3 Yi Shang, Wheeler Ruml, Ying Zhang, Markus
  P. J. Fromherz. Localization From Mere
  Connectivity
• 4 Paramvir Bahl, Venkata N.Padmanabhan, RADAR
  An In-Building RF-based User Location and
  Tracking System
• 5 K Lorincz, M Welsh, MoteTrack a robust,
  decentralized approach to RF-based location
  tracking
• 6 Eiman Elnahrawy, xiaoyan Li, Richard P.
  Martin, The limits of localization using signal
  strength a comparative study

15

• Questions ???

16

• Thank You !!!