ParkSense: A Smartphone Based Sensing System For On-Street Parking -- mobicom 2013

Motivation 30%

Traditional methods SFPark: wireless sensor nodes embedded into asphalt(沥青公路) at each parking location Google OpenSpot: allow drivers to mark a parking spot.

Primary Problem. How to detect the evacuation of parking space?

A sequence of specific actions

Typical approach GPS based location Network based location significant amount of time to acquire an initial fix power hungry Network based location perform a scan for surrounding wifi offers relatively coarse grained location

Performing passive wireless scans is the lowest energy operation compared to other location sensors. This motivates us to explore the wireless interface as a sensor to sense driver mobility.

Sensing that the user has returned to the parked vehicle location. Detecting the user is in a moving vehicle after returning to the parking location.

How to sense the user has returned to the parked vehicle location?

User returning to vehicle wireless signature(why?) SSIDs of the Wifi access points. the number of times a beacon frame was received from each AP. capture a wireless signature during payment phase.

User returning to vehicle periodically perform new wireless scans, create new signature.(St and Wt) Compare the signature St, Wt with Sp, Wp Optimize define the signature over a set of scans rather than a single scan. AP was visible in all the scans gets a higher weight Matching function

How to detect the user is in a moving vehicle after returning to the parking location?

User driving away Walking OR Driving ?

User driving away vehicle traveling follow a certain pattern what effect do these have on the wireless beacons observed by the mobile device inside the vehicle? compute the Jaccard similarity.

Special Cases After parking, the driver drives away in a friend’s car,two cars are closed to each other. A driver parks a vehicle and give the key to someone else. The driver returns and unparks as expected, but immediately gets stuck in the traffic.

-environment is dynamic 34/38 Sensing Accuracy 38/41 -environment is dynamic 34/38 -drive slowly

Evaluation Signature Matching(accuracy of signature matching functions)

User Driving

Energy Consumption

Thank You!