udel mobility model & simulator jonghyun kim advisor : dr. bohacek email : [email protected]

50
UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : [email protected]

Upload: loreen-shields

Post on 18-Dec-2015

250 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

UDel Mobility Model &Simulator

Jonghyun Kim

Advisor : Dr. Bohacek

Email : [email protected]

Page 2: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Contents

- Objectives

- Simulator Design & Modeling Node Mobility

- Demo Simulation

- Simple Simulation Set

- Simulation Results

- Summary

- Future work

Page 3: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Objectives

1. Implement realistic mobility simulator for pedestrian and car by applying UDel mobility model

2. Generate realistic mobility data from our simulator

3. Compare our simulated mobility data to actual measured data from pedestrian literature

4. Based on our mobility data, we will investigate impact on performance of MANET routing protocols (future work)

Page 4: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simulator Design & Modeling Node Mobility

1. Project overview

Map builder

UDel Mobility Simulator

QualNet

1. Generate map data

2. Generate realistic mobility data

4. Generate any statistical data ex ) data about routing protocols

Raytracing

3. Generate Pathloss data

5. Analyze the data and get the result

Page 5: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simulator Design & Modeling Node Mobility

Map Builder

Page 6: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simulator Design & Modeling Node Mobility

UDel Mobility Simulator

Page 7: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

2. Implementation method of UDel Mobility Simulator

* Discrete event method - Some different events are specified - Whenever an event occurs, the function for the event is executed

* Events - REACH_END_OF_SEGMENT - CATCH_UP - EXIT_FIFO - START_UP - MEET_IN_OPPOSITION - SEND_NEXT_CAR

Simulator Design & Modeling Node Mobility

Page 8: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - REACH_END_OF_SEGMENT

Segment can be one of sidewalk, roadway, hallway, or walkwayMobile node can be one of pedestrian or car

Segment

Mobile nodeLane

Simulator Design & Modeling Node Mobility

Page 9: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - REACH_END_OF_SEGMENT

Segment

Mobile nodeLane

Simulator Design & Modeling Node Mobility

Page 10: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - CATCH_UP

Let’s assume that node A’ speed is faster than Node B’s speed

Node A Node B

Lane

Simulator Design & Modeling Node Mobility

Page 11: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - CATCH_UP

• Node A should decide whether or not it changes current lane to overtake front node B• If there is enough space in the changing lane, node A has chance to change lane

• Probability of changing lane :

-V1 : the average speed of all nodes on current lane -V2 : min (the average speed of all nodes on changing lane, A’s desired speed) -For pedestrian, A = −0.225 B = 1.7 -For car, A = −0.225 B = 0.1

A B

)1/(1 ))21(( VVBAe

Reference : K. I. Ahmed, “Modeling drivers’ acceleration and lane changing behavior,” Ph.D. dissertation, MIT, 1999

Lane

A’s changing lane

A’s current lane

Simulator Design & Modeling Node Mobility

Page 12: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - EXIT_FIFO

• Let’s assume that all nodes’ speed is the same

Simulator Design & Modeling Node Mobility

Page 13: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - EXIT_FIFO

FIFO

• Since there is no enough distance to enter next segment, green node goes into FIFO• How much distance is needed for green node to enter the next segment ?

Simulator Design & Modeling Node Mobility

Page 14: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simulator Design & Modeling Node Mobility

Distance-Speed Relationship for Pedestrians

Reference : S. J. Older, “Movement of pedestrian on footways in shopping street,” traffic engineering and control, pp. 160–163, 1968. F. P. D. Navin and R. J. Wheeler, “Pedestrian flow characteristics,” traffic engineering, pp. 30–36, 1969.

Page 15: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

• We derived the equation below based on distance-speed relationship

For pedestrian,

Distance (S*, S) =

S* : Desired speed, S : Current speed, Dmin : minimum distance between people (at least 0.35m)

For car,

Distance (S) = A + B * S In dry conditions, (A, B) = (1.78, 10) and (1.45, 7.8) In wet conditions, (A, B) = (0.415, 8.3) and (0.230, 6.0)

)08.1/( *min SSDS

Simulator Design & Modeling Node Mobility

S. Shekleton, “A GPS study of car following theory,” in Conference of Australian Institutes of Transport Research (CAITR), 2002.T. Dijker, P. H. L. Bovy, and R. G. M. M. Vermijs, “Car following behavior in different flow regimes,” in Motorway Traffic Flow Analysis pp. 49–70.J. Piao and M. McDonald “Analysis of stop and go driving behavior through a floating vehicle approach,” in Proc. Of the IEEE Intelligent Vehicles Symposium, 2003

Page 16: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - EXIT_FIFO

FIFO

Simulator Design & Modeling Node Mobility

Page 17: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - EXIT_FIFO

FIFO

Simulator Design & Modeling Node Mobility

Page 18: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - EXIT_FIFO

FIFO

FIFO

Simulator Design & Modeling Node Mobility

Enough distance is now available

Page 19: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - EXIT_FIFO

FIFO

Simulator Design & Modeling Node Mobility

Page 20: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - EXIT_FIFO

FIFO

Simulator Design & Modeling Node Mobility

Page 21: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - EXIT_FIFO

FIFO

Simulator Design & Modeling Node Mobility

Page 22: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - START_UP

FIFO

…FIFO

…Yellow node starts up

• When red node exits FIFO, it checks to see if the following node stopped If the following node stopped, red node makes it start to move

Simulator Design & Modeling Node Mobility

Enough distance is now available

Page 23: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - START_UP

White node starts up

Simulator Design & Modeling Node Mobility

Page 24: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - START_UP

Show mobility simulator version 1.0

Simulator Design & Modeling Node Mobility

Page 25: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - MEET_IN_OPPOSITION

Right-hand side

Right-hand sideLeft-hand side

Left-hand side

• Each lane is bi-directional

Simulator Design & Modeling Node Mobility

Page 26: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - MEET_IN_OPPOSITION

Right-hand side

Right-hand sideLeft-hand side

Left-hand side

• When two nodes meet in opposition, left-hand side node gives a way to right-hand side node

Simulator Design & Modeling Node Mobility

Injected

Page 27: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - MEET_IN_OPPOSITION

Right-hand side

Right-hand sideLeft-hand side

Left-hand side

Simulator Design & Modeling Node Mobility

Page 28: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - MEET_IN_OPPOSITION

Right-hand side

Right-hand sideLeft-hand side

Left-hand side

Simulator Design & Modeling Node Mobility

Injected

Page 29: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - MEET_IN_OPPOSITION

Right-hand side

Right-hand sideLeft-hand side

Left-hand side

Simulator Design & Modeling Node Mobility

Page 30: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - MEET_IN_OPPOSITION

Right-hand side

Right-hand sideLeft-hand side

Left-hand side

Simulator Design & Modeling Node Mobility

Page 31: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - SEND_NEXT_CAR

Simulator Design & Modeling Node Mobility

Urban street

• During simulation, cars will exit or enter the city• When it’s time for some cars to enter the city, this event occurs• The number of vehicles that enter the city per traffic signal period is Poisson distributed

mean : oadsEntering RNumber of

ing)-prob_turn(iod Signal Perλ rtRateVehicleSta 1

Reference : A. Kamarajugadda and B. Park, “Stochastic traffic signal timing optimization,” Center for transportation studies at the university of Virginia, Tech. Rep. UVACTS-15-0-44, 2003.

Page 32: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - SEND_NEXT_CAR

Simulator Design & Modeling Node Mobility

Urban street

Page 33: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - SEND_NEXT_CAR

Simulator Design & Modeling Node Mobility

Urban street

Page 34: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Events - SEND_NEXT_CAR

Simulator Design & Modeling Node Mobility

Urban street

Page 35: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

3. Trip generation1) Pedestrian case - Pedestrian has a home office - Pedestrian initiates trips from its office at random times - Pedestrian chooses a destination - Destination can be an office, group meeting location or class room - Pedestrian goes to the destination with desired speed through shortest path

* Desired speed of pedestrian Pedestrian desired speeds are approximately Gaussian distributed Mean speed = 1.34 m/s Standard deviation = 0.26 Minimum speed = 0.7 m/s Maximum speed = 1.86 m/s

Simulator Design & Modeling Node Mobility

Reference : D. Helbing, “Sexual differences in human crowd motion,” Nature, vol. 240, p. 252, 1972 “The statistics of crowd fluids,” Nature, vol. 229, p. 381, 1971 G. K. Still, “Crowd dynamics,” Ph.D. dissertation, university of warwick, 2000.

Page 36: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* First step to choose a destination Fraction of trips that leave a building : U/M M = mean time between trips that leave a building (M depends on the characteristic of the building and time of the day) U = mean time between trips (i.e pause time) as exponentially distributed

* Second step to choose a destination Probability of selecting a range of distance to travel :

Simulator Design & Modeling Node Mobility

])1[400/1(])[400/1( iDiD ee

D[1]

D[2]

D[3]

Page 37: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simulator Design & Modeling Node Mobility

• CCDF of Distance Traveled During Outdoor Walking Trips

Reference : B. Pushkarev and J. M. Zupan, Urban Space for Pedestrians. MIT press, 1975

Page 38: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

* Simple group mobility - Some nodes will join group - All nodes in group move together - Speed of all nodes is the average over desired speed - All nodes in group occupy the whole lanes - A node behind group just follow group even if the node is faster than group - Group just follow a node ahead even if group speed is faster than node - Groups of pedestrians play an important role in platooning

* Group trip - There exist lots of trip cases ex ) case1 : office class a walkway disperse case2 : office class hallway disperse - What is the probability for each case ?

* Traffic light - If there is traffic light on segment and red light is on, pedestrian stops until green light is on

Simulator Design & Modeling Node Mobility

Page 39: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simulator Design & Modeling Node Mobility

2) Car case - Car initiates trips from a certain location - At each intersection, cars turn or go straight according to the turning probabilities

*Turning probability : 0.2

* Desired speed of car Car’s speed/ speed limit is approximately Gaussian distributed Mean = 0.78 m/s Standard deviation = 0.26 Speed limit = 13.4 m/s Minimum speed = 13.40.5 m/s Maximum speed = 13.41.4 m/s

Desired speed = Speed limit Random number

J. E. Hummer, “Unconventional left-turn alternatives for urban and suburban arterials,” ITE Journal, vol. 68, 1998 M. J. Bayarri, J. O. Berger, G. Molina, N. M. Rouphail, and J. Sacks, “Assessing uncertainties in traffic simulation: A key component in model calibration and validation,” National Institute of Statistical Sciences, Tech. Rep. 137, 2003.

Page 40: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simulator Design & Modeling Node Mobility

CDF of the ratio of observed speeds to speed limit andCDF of a fitted Gaussian distribution

Reference : Jianhe Du and Lisa Aultman-Hall, An Investigation of the Distribution of Driving Speeds Using in-Vehicle GPS Data, Vermont Institute of Transportation Engineers Annual Meeting, 2004, available at {http://www.neite.org/vt/dist1_2004/

Page 41: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Demo Simulation

Show mobility simulator version 1.05 with UDEL map

*Parameters - Number of pedestrian nodes : 3,000 - Number of car nodes : 150 - Simulation time : 3,000 seconds - Map : UD map - Number of lanes on walkway: 4 - Traffic light period : 90s - U : 300 seconds

Page 42: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simple Simulation Set

• Map1 with building on measuring walkway

• Map2 without building on measuring walkway

measuringpoint150 meter

measuringpoint

Page 43: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simple Simulation Set

*Parameters - Number of nodes : 10,000 - Simulation time : 1,800 seconds - Map : Map1, Map2 - Number of lanes on walkway: 4, 8, 16 - Traffic light period : 90s, 120s, 150s, 180s, 210s - Passing rule : easy to pass, hard to pass, our mobility passing rule - U : 18000, 12000, 6000, 4800, 3000, 1200, 600, 300 seconds - Mobility model: constrained random way point, our mobility model

*Measurement data - Passing time, current speed, desired speed, node ID, direction

Page 44: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simple Simulation Set

Choose next destination randomly

Random way point model

Initial point based on random seed

Step 1

Step 2

Page 45: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simple Simulation Set

Pause for some random time

Choose next destination randomly

Step 3

Step 4

Step 3 Step 4 repeat

Show random way point model movie

Page 46: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simulation Results

Reference : B. Pushkarev and J. M. Zupan, Urban Space for Pedestrians. MIT press, 1975 PP. 94

Page 47: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simulation Results

UDel mobility model conforms to the actual measurement

Page 48: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Simulation Results

Since there is no interaction among pedestrians, flow rate is so highSo, constrained random way point model is not realistic

Page 49: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Summary

1. MANET protocol performance evaluation varies by mobile node mobility 2. Researchers may use random way point mobility model as mobile node mobility3. As we saw, constrained random way point is not

realistic4. UDel mobility model approaches realistic model5. MANET protocol performance needs to be re- evaluated based on UDel mobility model

Page 50: UDel Mobility Model & Simulator Jonghyun Kim Advisor : Dr. Bohacek Email : kim@eecis.udel.edu

Future work

1. Investigate impact of UDel mobility model on performance of MANET routing protocols