Transcript I Am the Antenna: Accurate Outdoor AP Location using

I AM THE ANTENNA:
ACCURATE OUTDOOR AP LOCATION
USING SMARTPHONES
ZENGBIN ZHANG, XIA ZHOU, WEILE ZHANG, YUANYANG ZHANG
GANG WANG, BEN Y. ZHAO, HAITAO ZHENG
DEPARTMENT OF COMPUTER SCIENCE, UNIVERSITY OF
CALIFORNIA, SANTA BARBARA, USA
OUTLINE
 Introduction
 Accurate AP Location
 Directional analysis
 User navigation
 Implementation
 Evaluation
 Conclusion
INTRODUCTION
UBIQUITOUS
BROADBAND ACCESS
 Cisco: WiFi traffic will surpass wired IP traffic in 2015
 Governments and companies are deploying public WiFi networks
 High Density

We need well tuned and managed WiFi networks
CONVENTIONAL AP
LOCATION METHODS
INSIGHT: THE BODY
BLOCKING EFFECT
 Can we use this to detect AP location?
 No…Effect is not strong
 Observation
ROTATION BASED
MEASUREMENT
 The difference is significant, because
 User’s body is much larger than the phone
 User is close to the phone
 We can emulate a directional antenna just by rotating with
Smartphones
GENERALITY OF THE
EFFECT
 Devices
 Motorola Droid, HTC G1 (Android)
 LG Fathom (WM 6.5)
 iPhone4 (iOS)
 Protocols
 802.11 b/g
 802.11n (MIMO)
 Different users
 7 users in our lab
 Environments
 Outdoor LOS/Non-LOS
 Different distances to AP
GENERALITY OF THE
EFFECT
ACCURATE AP LOCATION
USER ROTATION
BASED AP LOCATION
MODELING THE BODY AS
A SIGNAL OBSTACLE
 Develop an accurate AP direction prediction mechanism, along with a
way to estimate the confidence of each estimate.

Modeling the Blocking Sector
2𝑝
𝑏𝑝
𝛽 = 180° − 2(𝑎𝑟𝑐𝑡𝑎𝑛
− 𝑎𝑟𝑐𝑠𝑖𝑛
)
𝑏
𝑑 4𝑝2 + 𝑏2
𝑑
,
𝑠𝑖𝑛𝛾
2𝑝
𝑎𝑟𝑐𝑡𝑎𝑛 , we
𝑏
By the sine law,
Given that 𝛾 =
𝑏
2𝑠𝑖𝑛𝜃
=
𝛽 ≈ 180° − 2𝑎𝑟𝑐𝑡𝑎𝑛
𝛾=𝜃+
𝛼
2
𝛾
arrive at 𝛼 = 2(𝑎𝑟𝑐𝑡𝑎𝑛
𝜃
2𝑝
𝑏
− 𝑎𝑟𝑐𝑠𝑖𝑛
𝑏𝑝
𝑑 4𝑝2 +𝑏2
2𝑝
= 90° (𝑤ℎ𝑒𝑛 𝑑 ≫ 𝑏, 𝑑 ≫ 𝑝, 𝑏 ≈ 2𝑝)
𝑏
)
OUR DIRECTIONAL
ANALYSIS MODEL
 Signal degradation occurs at a range of directions
DIRECTIONAL
ANALYSIS
 They represent the measured RSS profile during a user rotation
by
𝑅 = 𝑅𝑆𝑆 𝜃𝑖 , 𝜃𝑖 |𝑖 ∈ 1, 𝑁 }
N: number of measurement points
𝑅𝑆𝑆 𝜃𝑖 : RSS reading (dBm) at point 𝑖
𝜃𝑖 : user orientation at point 𝑖
 Locating the Blocking Sector
Diff(𝑗) =
𝜃∈𝑆𝑗 𝑅𝑆𝑆(𝜃)
𝑁 − |𝑆𝑗 |
−
𝜃∈𝑆𝑗 𝑅𝑆𝑆(𝜃)
|𝑆𝑗 |
𝑆𝑗 : candidate sector
|𝑆𝑗 |: the number of measurement points within the sector
𝑁 : the total measurement points in the signal
The blocking sector 𝑺∗ = 𝒂𝒓𝒈𝒎𝒂𝒙 𝐃𝐢𝐟𝐟(𝒋)
DIRECTIONAL ANALYSIS
IS NON-TRIVIAL
 Minimum RSS direction?
 Using Minimum RSS
direction would cause
large errors
LOCATING THE
BLOCKING SECTOR
 Find the sector with the largest RSS degradation
 Sliding window
 𝑆𝑖𝑛: average RSS inside the
sliding sector
 𝑆𝑜𝑢𝑡: average RSS outside
the sliding sector
 Degradation = 𝑆𝑖𝑛 − 𝑆𝑜𝑢𝑡
NAVIGATION
 How does a user navigate using directional hints?
 Strawman design: periodic
•
Refine AP direction every 20m
 However, nothing is perfect
•
Temporal/spatial variation
 Adaptive method
 Measurement confidence
•
The similarity of measured RSS profile and ideal RSS profile
 If confidence is high
•
Walk further between measurements
CONFIDENCE OF
DIRECTIONAL ANALYSIS
 Borealis uses confidence value of each estimate to control
how often a user needs to repeat the direction estimate.
1
𝜌= ∙
𝑚
𝑖∈[0,𝑚−1](𝑡𝑖
− 𝑡)(𝑟𝑖 − 𝑟)
𝜎𝑇 ∙ 𝜎𝑅
T = 𝑡0 , ⋯ , 𝑡𝑚−1 , R = 𝑟0 , ⋯ , 𝑟𝑚−1
where 𝑡, 𝑟 are the mean values of T, R
𝜎𝑇 , 𝜎𝑅 are the standard deviations of T, R
A BOREALIS
PROTOTYPE
 Application layer
 Leveraging WiFi scan to read RSS
•
•
Default scan is very slow
Scanning all channels each time
 OS layer
 Modified WiFi driver
•
Only scan the channel the AP is on
 Two benefits
•
•
Accelerate the process: 10 seconds per rotation (10 times faster)
Save power: WiFi’s energy consumption is 14 times less
EVALUATION
TESTING SCENARIOS
ACCURACY OF
DIRECTIONAL ANALYSIS
 Rotation at 400 + locations for each scenario
 Absolute angular error = |detected direction – actual direction|
 We compared Borealis to
 Offline analysis
•
•
Clustering-based learning algorithm
Optimized by a large training set
 GUIDE
•
RSS gradient based
 Min RSS
•
Minimum RSS direction based
ACCURACY OF
DIRECTIONAL ANALYSIS
 Rotation at 400 + locations for each scenario
 Absolute angular error = |detected direction – actual direction|
 We compared Borealis to
 Offline analysis
•
•
Clustering-based learning algorithm
Optimized by a large training set
 GUIDE
•
RSS gradient based
 Min RSS
•
Minimum RSS direction based
NAVIGATION
EFFICIENCY
LOCATION INDOOR
APS?
 Most APs are mounted inside buildings
 They mounted the AP on a table in a office trailer
 Try to locate it outside in Complex LOS/NLOS environment
 More complex concrete buildings would be much harder
CONCLUSION
 AP location requires cheap, online and accurate solutions
 Very beneficial in network management
 Borealis: an efficient and accurate solution for WiFi AP location
 Leveraging the body blocking effect on smartphones
 General to a wide rage of devices and systems
 Feasible even in complex environments
THANK YOU