1 performance analysis of traffic networks based on stochastic timed petri net models jiacun wang,...
Post on 21-Dec-2015
227 views
TRANSCRIPT
1
Performance Analysis of Traffic Networks Based on Stochastic Timed Petri Net Models
Jiacun Wang, Chun Jin and Yi Deng
Center for Advanced Distributed Systems Engineering
School of Computer Science
Florida International University
Miami, FL 33199
2
Contents
1. Introduction
2. Traffic Control of Networks
3. STPN Model of Intersection Traffic Control
4. Modeling and Performance Evaluation of Traffic Networks
5. Conclusion
4
Traffic Control System
Characterized by: shared resources resource conflicts a tendency to deadlock and overflow, requirement of well-planned synchronization, scheduling and control
5
The State of Arts of Performance Analysis of Traffic Control Systems
Queuing Theory Models Simulation Petri Net Models
6
We present:
a compositional method for modeling and analyzing complex traffic control systems.
a typical STPN model of traffic networks using the compositional method.
8
Urban Traffic Control Systems
Isolated Intersection
1 2 3 … … m-2 m-1 m
1
2
3
n-2
n-1
n
Closed Network
11
Urban Traffic Control Systems-- Timing Plan Issues
Cycle length: The time period of a complete sequence of signal indications.
Split: A division of the cycle length allocated to each of the various phases.
Offset: The time relationship determined by the difference between a defined interval portion of the coordinated phase green and a system reference point.
Phase: A portion of signal cycle during which an assignment of right of way is made to a given traffic movements.
13
Petri Nets
p 1
t2
p6
p 5
p 4
p 3
p 2
p7
t1
t 3 t 5
t4
2
2
p 1
t 2
p6
p 5
p4
p 3
p 2
p 7
t1
t3 t 5
t4
2
2t1 fires
14
Stochastic Timed Petri Nets
When "time" is assigned to transitions (or places) of Petri nets, they are called Timed Petri Nets.
If the "time" is random in timed Petri nets, they are called Stochastic Timed Petri Nets.
15
STPN Model of An Intersection -- Control Model (I)
G_A2
ea_A2
GA_A2
eg_A2
G_A1
eg_A1
A_A1
A_A2
ea_A1
GA_A1
The control model of 2-phase intersection
PLACE:G_A1: Green signal for direction EW A_A1: Amber signal for direction EWGA_A1: Green or amber signal for direction
EW
TRANSITION:eg_A1: Green signal endsea_A1: Amber signal ends
16
STPN Model of An Intersection -- Control Model (II)
G_B1
G_A2
ea_A2
GA_A2
G_A1
ea_A1
eg_A1
A_A1
A_A2
GA_A1
eg_B2
eg_B1
G_B2
eg_A2
The control model of 4-phase intersection
17
STPN Model of An Intersection -- Traffic Flow Model (I)
IN
LOUT
ROUT
RDY
arr
rto
lto
ent
INT
OUT
LIN
RIN
Rdep
lti alt
rti
RSL
nrto
nlto
MFControl Part
GA_A1
The traffic flow model of 2-phase intersection
PLACE:RDY_A1 Incoming vehiclesIN_A1 Vehicles ready to enter intersectionRSL_A1 Ready for going straight or turning leftROUT_A1 Right-turn-out vehiclesINT_A1 Vehicles entering intersectionMF_A1 Vehicles moveing forwardOUT_A1 Vehicles outRIN_A1 Right-turn-in vehiclesLOUT_A1 Vehicles turn left outLAR_A2 Left-turn-in incoming vehicles LIN_A2 Left-turn-in incoming vehicles to enter intersection
TRANSITION:arr_A1 Vehicles arriverto_A1 Vehicles right-turn outNrto_A1 Vehicles not right-turn outlto_A1 Vehicles left-turn outNlto_A1 Vehicles not left-turn outent_A1 Vehicles enter intersectionDep_A1 Vehicles depart intersectionlti_A1 Vehicles left-turn inrti_A1 Vehicles right-trun inalt_A1 Left-turn-in vehicles arrive
18
STPN Model of An Intersection -- Traffic Flow Model (II)
ROUT
arr
rto
ent
INT
IN
RDY
RSL
sf
G_B2
OUT
LIN
RIN
G_B1
LARdep
lti alt
rti
lto
tlo
LTO
LOTGA_A1
Control Part
The traffic flow model of 4-phase intersection.
20
Compositional Modeling
Traffic and traffic control of an intersection: STPN model Traffic of an road segment: Random motion model Compositional model of a traffic system: STPN + random
motion model Interactions between different directions are partially
approximately by statistical models.
21
Compositional Analysis
Based on individual intersection model One direction of traffic along a two-way street is considered
separately from the other, Incrementally evaluate system’s performance by analyzing
intersections one by one according to a carefully selected order
22
STPN Model of 2-phase Intersection
IN_A2
LOUT_A2
ROUT_A2
G_A2
ea_A2
GA_B
RDY_A2
arr_A2
rto_A2
lto_A2
ent_A2
INT_A2
OUT_A2
LIN_A2
ROUT_A1
G_B2
LAR_A2
dep_A2lti_A1
alt_A2
rti_A2
eg_A2
RSL_A2
nrto_A2
nlto_A2
MF_A2G_A1
ea_A1
eg_A1
A_A1
A_A2
GA_A1
IN_A1
ent_A1
INT_A1
OUT_A1
dep_A1
RSL_A1
nrto_A1
nlto_A1
MF_A1
rto_A1
ROUT_A1
RIN_A1
LOUT_A2
rti_A1
GA_A2lti_A
arr_A1
RDY_A1
LOUT_A1
lto_A1
West East North South
23
STPN Model of 4-phase Intersection
IN_A2 RIN_A2
G_A2
ea_A2
GA_A2
RDY_A2
arr_A2
rto_A2
lto_A2
ent_A2
OUT_A2
LIN_A2
RIN_A1
G_B2
LAR_A2dep_A2
lti_A1
alt_A2
rti_A2
RSL_A2
sf_A2
MF_A2
G_A1
ea_A1
eg_A1
A_A1
A_A2
GA_A1
IN_A1
ent_A1
OUT_A1
dep_A1
RSL_A1
sf_A1
MF_A11
rto_A1RIN_A1
RIN_A1
LOUT_A2
rti_A1
GA_B1 lti_A2
arr_A1
RDY_A1
lto_A1
eg_B2
eg_B1
G_B1
G_B2
eg_A2
G_B1
OUT_A1
G_B2
LOUT_A1
LTO_A1
tlo_A1tlo_A2
LTO_A1
West East North South
24
Compositional Analysis
1 2 3 … … m-2 m-1 m
1
2
3
n-2
n-1
n
j
ii+j=2 i+j=3 i+j=4
Analyze intersections along the increasing i+j line
25
Example
Intersection Offset AQL ATD Offset AQL ATD(0,0)(4p) 0 10.5575 35.433 0 10.5575 35.433(0,1)(2p) 60 10.808 37.9069 30 9.80259 34.217(1,0)(4p) 60 13.7656 47.0608 30 10.7584 36.273(0,2)(2p) 0 9.87869 33.9058 60 3.95742 13.4698(1,1)(4p) 0 7.93539 26.9096 60 10.7768 35.9995(2,0)(2p) 0 5.54243 19.0341 60 8.39731 29.021(0,3)(4p) 60 7.33246 24.0968 90 9.24824 30.4988(1,2)(2p) 60 4.95944 17.8978 90 8.93898 31.7011(2,1)(4p) 60 18.0724 57.2082 90 13.8584 43.9798(3,0)(2p) 60 5.03002 16.7118 90 9.83201 32.6142(1,3)(4p) 0 14.7134 53.6546 0 10.2563 37.0925(2,2)(2p) 0 15.1647 48.6528 0 12.8156 41.1378(3,1)(4p) 0 13.8706 44.8415 0 12.4933 39.9578(2,3)(2p) 60 8.97933 30.5812 30 9.95824 34.2142(3,2)(4p) 60 5.99334 18.6261 30 16.4617 50.5948(3,3)(2p) 0 7.10573 25.482 60 12.1873 43.395
26
Discussion
Dimension of State Space (number of places used in PN model)
Our model: 27;
Global models: 3316 = 528. Number of Reachable States (suppose that there are M reachable
states for each intersection in average)
Our model: 16M;
Global models: M16.