1

Phone-Radar： Infrastructure-free Device-to-deveice Localization

班級：碩研資工一甲

姓名：高逸軒

學號：MA4G0110

Author:Zheng Song, STATE KEY LAB. OF NETWORKING & SWITCHING TECHNOL., BEIJING UNIV. OF POSTS & TELECOMMUN., BEIJING, CHINA

Source:VEHICULAR TECHNOLOGY CONFERENCE (VTC SPRING), 2014 IEEE 79TH

2

Outline1) INTRODUCTION

2) THE PROPOSED APPROACH

A. System Overview

B. Relative Movement Calculation

C. Distance Measurement using RSSI

D. Delative Location Calculation

3) EXPERIMENTS

A. Impact of moving distance to localization accuracy

B. Localization accuracy in indoor environment and outdoor environment

C. Comparison between PhoneRadar;Wifi localization,cell-id localization and GPS

4) CONCLUSION

3

1) INTRODUCTION• Supporting applications such as the ones above requires providing accurate

distance and direction to both smartphone users. An easy-to-find solution is share locations obtained from existing localization method, e.g.,GPS or WiFi based localization methods.

• However, existing commonly-used pedestrian localization approaches are constrained either by limited coverage or by low accuracy and is not suitable for these scenarios.

• GPS always fails to function in indoor environments due where the GPS signals are too weak.

4

1) INTRODUCTION• Cell-id based localization method is considered to be an important aid to

GPS, the basic idea of which is to use the received signal strength of GSM cell towers combined with the localizations of cell towers to calculate localization. It functions well in any area that is covered by GSM network, including indoor environments, but the hundred meter accuracy of cell-id localization method constrains its usage in device-to-device localization scenario.

• Another widely used localization technique is the fingerprinting localizationbas-ed on the received-signalstrength(RSS) observations of mobile devices to fixed WiFi access points. It match all measured RSS values to pre-trained RSSI finger-prints and determine the position that gives the best match.

5

1) INTRODUCTION• Dead-reckoning method is another recent approach, which use the low-cost

inertial sensors equipped in most smartphones to provide continuous position, velocity, and also orientation estimations.

• Phone-Radar is a hybrid approach taking advantages of both WiFi RSSI-based positioning and dead-reckoning techniques which improves its accuracy.

• Phone-Radar estimates the relative location between two smartphones by taking two measurement on the movement of the two smartphones separately and the change of the RSSI strength during the moving procedure.

6

2) THE PROPOSED APPROACHA. System Overview

• The scenario comprises two mobile devices held by pedestrians, denoted as A and B. As localization requirements mostly exist in scenarios that two devices can not see each other, a wall is added as a obstruction between two mobile devices to make the proposed method more adaptive.

• By the measured parameters, the relative location of A towards B can be calculated. The details of the implement are given in the coming part of this section. It is worth noting that, during time period and , at least one of the two devices should be moving.

�⃗�2𝐴 �⃗�1

𝐴

�⃗�2𝐵 �⃗�1

𝐴

𝑟1𝑟2𝑟3

7

B. Relative Movement Calculation

a. Step detection:Nonlinear model is selected to estimate the step length. The step length L of

pedestrians is determined by his accelerator readings:

Based on the counted number of steps, the estimated step length and the direction of each step, the movement vector of device A and B can be calculated separately by:

b. Relative movement:

a. The relative movement of A towards B, denoted by mi, ∀i ∈ T1 T2, T2 T3, can be calculated by:

2) THE PROPOSED APPROACH

8

C. Distance Measurement using RSSILet denote the distance between A and B on time i, ∀i ∈ {T1, T2, T3}. According to

the free space radio propagation model, the relationship between and is denoted as:

2) THE PROPOSED APPROACH

9

D. Relative Location Calculation

• B is set to be the origin of the coordinates, the direction of axis y is north, and the direction of axis x is east. A’s location at is set to be . On , A’s relevant location according to B is set to be , which can be calculated by = + . Accordingly, is calculated by = + . To know the original relative location of A towards B, is to know the distance between and B, denoted as and the direction of B, denoted as α.

2) THE PROPOSED APPROACH

Targeted Parmeter:

10

D. Relative Location Calculation

1. Calculation of : Accoording to Cosine theorem, .

2) THE PROPOSED APPROACH

Calculation:

11

2. Calculation of α: To calculate α, the angle

2) THE PROPOSED APPROACH

12

3) EXPERIMENTS• Through simple setup procedures, one device serves as Wi-Fi access

point(AP) and the other device is connected to the AP. They communicate by socket connection and the AP gathers both movement information and the received signal strength from the other device. When the relative location is calculated, the client’s location is shown in the screen of the AP.

• Before the experiments, two parameters, K and n are trained in advance. According to the test data, K is set to 0.304 and n is set to 2.3.

• First, we verify the impact of the distance of relative movement to localization accuracy and we compare localization accuracy while the initial distance between the two devices is different. Second, we verify localization accuracy in both indoor and outdoor environments.

13

3) EXPERIMENTSA. Impact of moving distance to localization accuracy

Initial locations of volunteers A and B

Impact of walking distance to localization error

14

3) EXPERIMENTSB. Localiztion accuracy in indoor environment and outdoor environment

Localization error of different initial distance in indoor environment and outdoor environment

15

3) EXPERIMENTSC. Comparison between PhoneRadar,Wifi localization,cell-id localization and GPS

16

4) CONCLUSION• we present Phone-Radar, which is an infrastructure-free device-to-device

localization system. We further study the relationship among the initial relative locations between the two devices, their relative movements and the change of received signal strength measurements.

• we implement the proposed method and measure its performance under real world conditions. As far as we know, the proposed method is the first to solve device-to-device localization on mobile devices without any fixed infrastructure or add-on module.

17

Thank you

• The ending......