Towards mobile phone localization

Is Skyhook the solution?Middle GroundLower Accuracy than GPS, Longer Battery lifetime Better Accuracy than GSM, Shorter Battery lifetime but …... Is Skyhook the solution?Middle GroundLow

Towards Mobile Phone Localization

without War-Driving

Location is an IP address

Location is an IP addressfor content delivery

What about WiFi Localization?

E.g., SkyHook:

Basic Idea:

1 Several trucks war-drive a place

2 Create Radio map = <Location: WiFi IDs>

3 Distribute map to phones

4 Phone user goes to war-driven region, overhears WiFi IDs

5 Reverse Look Up IDs against radio map

Basic Idea:

1 Several trucks war-drive a place

2 Create Radio map = <Location: WiFi IDs>

3 Distribute map to phones

4 Phone user goes to war-driven region, overhears WiFi

IDs

5 Reverse Look Up IDs against radio map

Is Skyhook the solution?

Middle GroundLower Accuracy than GPS, Longer Battery lifetime Better Accuracy than GSM, Shorter Battery lifetime

but …

Is Skyhook the solution?

Middle GroundLower Accuracy than GPS, Longer Battery lifetime Better Accuracy than GSM, Shorter Battery lifetime

but …

At the cost of:

 Degraded location accuracy: walking paths ~ 60m

 Reliance on infrastructure (APs)

 War-driving ($$ + carbon footprint)

Visual Representation

Of the State of the Art

(Duke Campus)

 No War-Driving

 Cannot drive walking paths (campus, parks, …)

 Expensive / Environment unfriendly

 No reliance on WiFi infrastructure

 Rural regions / developing countries

 Good accuracy (~GPS)

 Improve energy-efficiency

 Better than Skyhook, GPS

CompAcc: Basic Idea

 Direction(compass) + Displacement(accelerometer) = User’s directional trail

Directional Trail

CompAcc: Basic Idea

 Direction(compass) + Displacement(accelerometer) = User’s directional trail

 Compute path signatures

Path Sig

nature

…

CompAcc: Basic Idea

 Direction(compass) + Displacement(accelerometer) = User’s directional trail

 Compute path signatures

 Compare directional trail with path signatures

Directional Trail

Path Sig

nature

…

 No war-driving

 No reliance on WiFi infrastructure

 Maps available ubiquitously

 Improves battery lifetime

 GPS ~10h

 Skyhook ~16h

 Accelerometer ~ 39h

 Compass ~48h

Tile Database

Tile Database

6 Current location (lat A, long B)

2 Report initial location

(lat X, long Y)

Tile

4 Direction (Compass)

5 Displacement

1 Initial location GPS:

(lat X, long Y)

Initial location Directional trail Current location

3 Obtain paths in the user vicinity

Directional trail: displacement

 Accelerometer based step count

 displacement = step_count * step_size

Directional trail: direction

Directional trail: direction

Path Signature

 Extract from Google Maps

 Geodesic formulas

Matching Directional Trail with Path

Signatures

Dissimilarity Metric:

ci = compass readings

p = path computed direction

Directional Trail

Path Signature

Instantaneous Error = distance(estimated, real)

Average Localization Error (ALE) = Average Instantaneous Error

CompAcc Instantaneous Error

CompAcc Instantaneous Error

Results

Results

Average ALEGPS: 10m

CompAcc: 11m

WiFi-War-Walk: 30m

Skyhook: 70m

Energy GPS: 10h

CompAcc: 23h

WiFi-War-Walk:16h

Skyhook:16h

Skyhook CompAcc

Limitations and Future Work

 Map Generation

 Manually mark footpaths

 User Position Estimated along the Path

 Apply particle filters to accommodate wide roads

 Multiplexing between Localization Methods

 Hand-off to Skyhook/GPS when driving

 Extend to vehicular movement

 Today’s localization technologies limited

 Energy- Efficiency

 Coverage/Accuracy

 Rely on simple localization mechanism

 Need: Compass, Accelerometer and Maps

 Evaluation results:

Thank You!

Visit the SyNRG research group @

http://synrg.ee.duke.edu/

Additional Slides

Skyhook on walking paths

g – GPS position

S – Skyhook position

Assume GPS

is absolute truth

Skyhook Coverage

Directional trail: displacement

 Step count/displacement accuracy

Trail/Path Size

Fallback Mechanism: A-GPS

 What if the dissimilarity metric is large?

 Trigger A-GPS

Fallback Mechanism

Estimated Location

Fallback Mechanism

Estimated Location

Pervasive wireless connectivity

+Localization technology

=

Location-based applications (LBAs)

Pervasive wireless connectivity

+Localization technology

Location-Based Applications (LBAs)

 Examples:

Location-based recommendations, geo-tagging

GeoLife: shopping list when near a grocery store

TrafficSense: real-time traffic conditions

 Location expresses context of user

 Facilitates content delivery

200-400m GSM ~days