A Framework for Intelligent Decision Support System for Traffic Congestion Management System

doi:10.4236/eng.2010.24037

Paper Menu >>

Journal Menu >>

Engineering, 2010, 2, 270-289

doi:10.4236/eng.2010.24037 Published Online April 2010 (http://www. SciRP.org/journal/eng)

A Framework for Intelligent Decision Support System for

Traffic Congestion Management System

Mohamad K. Hasan

Department of Quantitative Methods and Information Systems

College of Business Administration, Kuwait University, Kuwait city, Kuwait

E-mail: mkamal@cba.edu.kw

Received November 18, 2009; revised January 22, 2010; accepted February 4, 2010

Abstract

Traffic congestion problem is one of the major problems that face many transportation decision makers for

urban areas. The problem has many impacts on social, economical and development aspects of urban areas.

Hence the solution to this problem is not straight forward. It requires a lot of effort, expertise, time and cost

that sometime are not available. Most of the existing transportation planning software, specially the most

advanced ones, requires personnel with lots practical transportation planning experience and with high level

of education and training. In this paper we propose a comprehensive framework for an Intelligent Decision

Support System (IDSS) for Traffic Congestion Management System that utilizes a state of the art transporta-

tion network equilibrium modeling and providing an easy to use GIS-based interaction environment. The

developed IDSS reduces the dependability on the expertise and level of education of the transportation plan-

ners, transportation engineers, or any transportation decision makers.

Keywords: Traffic Congestion Management System; Transportation System Management; Intelligent

Decision Support System; Urban Transportation Systems Analysis; Multiclass Simultaneous Transportation

Equilibrium Models; Intelligent Scenario Creation Assistance Agent

1. Introduction

Over the last few decades the dimensions of the trans-

portation system and the interaction between these di-

mensions and the socioeconomic system, have increased

by many folds. And despite the fact that significant ad-

vancements have occurred in Information and Commu-

nication Technologies (ICT)-based transportation sys-

tems and many network equilibrium models have been

developed, comprehending this dynamic environment,

identifying its critical issues, and responding to them by

selecting the right alternative at the right time is still an

immensely demanding and difficult task to accomplish

even for well trained and experienced transportation ex-

pertise.

Urban transport problem can be understood in relation

to land use and other socio-economic activities which the

transportation network serves. A relationship exists be-

tween trip frequencies and socioeconomic characterris-

tics of trip makers and the purpose for which the trip is

made (such as journey to work, to school, to shopping, to

social activities, etc. Each of these types of trips is spa-

tially distributed over the city between the residential

areas and places of interest and activities. Trip purposes,

also influence the volume of trip generation in various

land use in the city [1]. This is because the volume of

journeys made from origin to destination is functionally

related to the ability of respective areas to generate or

attract movement. Reasons such as inadequate transport

infrastructure, rapid increase in urban population, income

and automobile ownership etc., had been noted to con-

tribute to traffic congestion in many of the urban areas

[2]. Traffic congestion is a critical problem in urban ar-

eas. It is not always fruitful to expand the transportation

systems, if the demand is not satisfactorily met by the

existing capacity. As such it is prudent to optimize the

use of existing transportation system by certain man-

agement techniques.

Gazis [3] proposed a way to control closely spaced

and oversaturated intersections. Gordon [4] suggested

that an intersection control scheme should be aimed at

controlling the relative queue lengths on the phases.

Longley [5] identified two types of congestion and pro-

posed a signal control and queue management procedure

that aims at reducing secondary congestion. Pignataro et

al. [6] described a queue management procedure. Gartner

M. K. HASAN.271

et al. [7] presented a procedure with multi-level design

for real time, traffic-adaptive control. Other work on

queue management appeared in the literature [8-10].

Reddy et al. [11], Reddy [12], described the Transporta-

tion System Management (TSM) actions to improve ve-

hicular movements with the help of a case study. Thom-

son [13] described the traffic management problem faced

by highly urbanized cities of developing world due to the

rapid urbanization relating to traffic congestion. This

effort at reducing traffic congestion without providing

for more road capacity is expected to shift the trip pattern

of motorists in various ways, notably to redistribute trips

spatially, temporarily, and modally. Other studies on

urban traffic management emanating from non commis-

sioned studies include that of Adefolalu [14], on traffic

congestion in the city of Lagos; Oyefesobi [15] on acci-

dent reduction: Orioke [16], on traffic education: Ogun-

bondede [1], on the contribution of land use to traffic

congestion.

These above studies show that traffic congestion

problem is a common phenomenon in urban and part of

TSM actions to improve the existing transportation sys-

tem for the better serviceability and efficiency by taking

various measures. Transportation Systems Management

(TSM) measures, originally conceived as a tool applica-

ble on a region-wide scale, can be successfully applied at

major activity centers to avoid, minimize or postpone the

need for more capital-intensive transportation improve-

ments [17]. A transportation management system (TSM)

is intended to provide information on transportation sys-

tem performance and identify alternative actions to alle-

viate congested roadway conditions. Although a TSM

can use data obtained from a traffic operation surveil-

lance and control system, in general a TSM is considered

to be more a planning tool than an operations tool [18].

The main objectives of TSM actions are to coordinate

all individual elements of transportation systems through

operating, regulatory and control policies, so as to achieve

the maximum efficiency and productivity and minimize

environmental pollution and energy consumption. TSM

actions, whether applied on a corridor or at city network

level, aim at maximizing the expected output of the

transportation systems as whole and achieve the specific

objectives [11].

TSM actions include:

 Assuring the safety and the efficiency flow of pas-

senger vehicles and trucks along urban and rural

transportation networks [19].

 Choosing the most efficient improving alterna-

tives solutions of the existing transportation system

problem at a minimum cost [19].

 Increasing the transit system ridership by increasing

the frequency of buses or trains, reducing the fair,

and assuring scheduling time accuracy.

 Minimizing the air pollution, noisy, and all others

undesirable environmental impacts of the existing

transportation system [19]

 Banning of particular class of vehicles at certain

times/certain areas.

 Auto free zones where vehicles are totally prohib-

ited.

 Special bus lanes and streets where spaces are re-

served for public buses and emergency vehicles and

other improvements of facilities for buses viz. bus

bays and bus stops.

 Fiscal measures like parking fee, road pricing and

supplementary licensing etc. which in turn make the

less vehicle usage.

 Parking Control

 Staggering of office hours

 Car pooling

 Travel Bonus

 Controlled Entry

 Banned turns

 Geometric improvements

 Improvements for slow moving traffic

 Improvements to pedestrian facilities

 Improvements in intersection circulation

 Synchronization and redesign of signals

Detailed benefits of the TSM Program can be found

in [19].

To evolve TSM action plan for the selected corridor

there is a need to understand and analyze the characteris-

tics of traffic flow as well as existing transit system, land

use development which generates the traffic on this cor-

ridor and their ill effects.

To integrate the entire above feature of TSM, we first

defined Congestion Management System (CMS) and

Intelligent Decision Support System (IDSS).

A Congestion Management System (CMS) is a sys-

tematic process of monitoring, measuring, and diagnos-

ing the causes of current and future congestion in major

travel corridors [20]; performing a detailed evaluation

analysis and recommendations for different alternatives

solutions strategies that can be implemented to improve

management of current and future congestion [20]; and

monitoring and evaluating the performance of strategies

that have been implemented to manage or mitigate con-

gestion [21]. US Federal transportation legislation re-

quires Metropolitan Planning Organizations (MPO) to

develop and implement a CMS as part of the metropoli-

tan transportation planning process [22]. The CMS pro-

vides a consistent basis to make transportation invest-

ment decisions and ensures that travel demand manage-

ment (TDM) and transportation system management

(TSM) measures are considered prior to roadway capac-

ity expansion, in accordance with goals and objectives

established by any Metropolitan Planning Organization

[20]. The MPO is a transportation policy-making organi-

zation made up of representatives from local government

and transportation authorities.

The dynamic and complex nature of transportation en-

M. K. HASAN.

272

vironment, the complexity of the current transportation

system and difficulty to comprehend the network equi-

librium models and the limited availability of transporta-

tion expertise together created an extremely hard envi-

ronment to deal with efficiently and effectively. How-

ever, the advancement and availability of intelligent

agent-based technologies and solutions for integration

within a comprehensive transportation system analysis

framework, as a Traffic Congestion Management System

(TCMS), provide promising solutions.

Ossowski et al. [23] summarized good literatures ab-

out the Decision Support Systems that include previous

works of French [24], Silver [25], Klein and Methlie [26],

Hernandez and Serrano [27], Hernandez and Serrano

[28], Vlahavas et al. [29], Cuena and Ossowski [30],

Ossowski et al. [31], Iglesias et al. [32], and Hoa Dam

and Winikoff [33].

More details about the decision support system fun-

damentals and the classification of decision support sys-

tems can be found in Holsapple [34].

Recently, Zhang [35] addressed the requirements for

an open and standardized environment for integrating

various DSSs. He also addressed the emergence of intel-

ligent agent technology to fulfill the requirements of de-

veloping innovative and efficient DSS applications. In

his thesis he gave detailed illustrations of the intelligent

agents in offering various advantages, such as mobility,

flexibility, intelligence, etc., to overcome the major pro-

blems in existing DSSs.

Creating an intelligent agent-based DSS for TCMS

that would enable and assist the transportation system

users (operators) to identify critical transportation prob-

lem issues and to respond to them by selecting the right

course of action in a timely, efficient and effective man-

ner would be a response to this critical requirement. To

this end, the objective of this paper is to provide such a

framework.

The main objective of this research paper is to develop

A Framework for Intelligent Decision Support System

for Traffic Congestion Management System, using state

of the art developments in urban transportation system

modeling, that can be a useful decision support tool for

transportation planners and decision makers for the

analysis and evaluation of strategic transportation plans

include transportation projects and policies.

The main contribution of this research is the creation

of Scenario Management Subsystem (SMS) and the In-

telligent Scenario Creation Assistance Agent (ISCAA)

briefly discussed in Section 4.5 and Section 4.6 respec-

tively. We will also show how these components are

integrated with the impact models results where the in-

puts to these impact models are the result of the Multi-

class Simultaneous Transportation Equilibrium Model

(MSTEM) [36] that will be run for each transportation

alternative solution. And finally, we have an intelligent

DSS that can recognize a pattern of solution alternatives

for traffic congestion problem associated with its total

evaluation. This IDSS can be used by any non-

transportation specialist decision makers for any trans-

portation policy evaluations and recommendations.

This paper is organized as follow: Section 2 gives the

main concept for the transportation systems analysis.

Section 3 presents a more detailed framework for trans-

portation system analysis combing the concept of the

transportation planning process and the transportation

system analysis concept. Section 3 presents The Multi-

class Simultaneous Transportation Equilibrium Model

(MSTEM)’s description, formulation as Variational Ine-

quality (VI) and the data collection requirements. Section

4 illustrates the main Architecture of the Intelligent De-

cision Support System (IDSS) implementing the frame-

work for transportation system analysis described in Sec-

tion 2. And finally, the summary, conclusion and future

research are presented in Section 5.

2. Transportation Systems Analysis

2.1. Transportation System

We briefly define the transportation planning as a con-

tinuous process consists of:

1) Problem definition, Generation of Alternatives

2) Building and Calibration of Transportation Models

3) Application of the Calibrated Transportation Mod-

els and Analysis of Alternatives

4) Evaluation and Choice

5) Implementation

A transportation system can be define as the combina-

tion of elements and their interactions, which produce the

demand for travel within a given area and the supply of

the transportation service to satisfy this demand. This

definition is general and flexible enough to be applied to

different context. The specific structure of the system is

defined by the problem itself (or class of problem) for

whose solution is employed.

Almost all of the components of a social and economic

system in a given geographical area interact with diff-

erent levels of intensity. However, it is practically im-

possible to take into account every interacting element to

solve a transportation problem. The typical system ap-

proach is to isolate the most relevant element in the

problem. These elements, and the relation among them,

make up the analysis system. The remaining elements

which belong to the external environment are taken into

account only in term of their interaction with the analysis

system. The transportation system of a given area can

also be seen as a sub-system of a wider territorial system

with which it strongly interacts. The extent, to which

these interactions are included in the analysis system, or

in the external environment, depends on the specific

problem.

M. K. HASAN.273

Transportation is the process of transferring people,

good and information from one place to another. To per-

form its function, any transportation system consists of

several components which together act as a single unit

that is designed and developed to provide a suitable

technology for the “objects” to be transported.

Any transport technology must provide mobility, con-

trol, protection, and land access for the objects. Perhaps

the most widespread transport technology is the one used

for inland transport. That is vehicles and containers op-

erating on highway or railway networks.

Components of such a transport system may be di-

vided into two categories. The first category includes the

physical components such as the network infrastructural

fixed elements (i.e., road and rail links, intersections,

terminals, parking spaces, railway yards, maintenance

shops, stations, etc.), and moving elements (i.e., vehicles

and containers). The second category includes the human

components such as users, operators, owners, and regu-

lators of the system, government and the society at large.

The key players in system analysis are defined as fol-

lows:

1) Users: are those traveler and shipper who represent

demand on the transport system.

2) Operators: are those who own the fleet of vehicles

and hence, are responsible for their operation, mainte-

nance and investment.

3) Owners: are those who own the network elements

and hence, are concerned with construction, upgrading,

operation and maintenance of the infrastructure.

4) Regulators: are responsible for traffic laws, ordi-

nances and regulations.

5) Government: is responsible for creating the master

plan for the city transportation systems and policies or

projects that affect the transportation systems.

6) Society at large: is responsible location of residents

and activities

It should be emphasized that these components should

interact all together in order to provide effective and ef-

ficient “transportation”. Furthermore, a transport system

may be viewed as one of several components of more

complex socio-economic system of the society. The inter-

action between the transport system and its surrounding

socio-economic environment is, again, evident. Transport

demand is a function of the magnitude and spatial distri-

bution of socio-economic activities which in turn are

greatly influenced by the characteristics of transport sys-

tems. Therefore, the actual performance of any transport

system is a function of several interacting and interde-

pendent factors within the system and those outside the

system.

Like any other complex system, transport system may

not always perform as desired and there are often prob-

lems and issues to be addressed and resolved. Traffic

congestion, limited parking space, high accident rate,

weak connectivity between major development centers,

freight movements, public transit, and air pollution are

but a few examples to mention. The range of possible

remedies is enormous. Construction of new highways,

building multistory garages, introducing new transport

technologies, creating new organizational structures and

traffic regulations are examples of actions which would

be undertaken.

The “best” action (or set of actions) to be implemented

in particular situation, is a question that is, often, not a

simple one to resolve immediately and usually requires a

systematic process of analysis that takes into considera-

tion the interacting affects on the system, that is, trans-

portation planning. Formally defined, planning is “a sys-

tematic analytical process that assists decision makers of

a given system to achieve a specific set of goals and ob-

jectives within a given socio-economic environment in

an optimum fashion”.

2.2. Basic Premises

As indicated earlier any transport system consists of sev-

eral physical as well as human components which inter-

act together in order to produce “transportation”. The

transport system itself is, again, one of several compo-

nents in the socio-economic activity system of the soci-

ety when all interact produce what we call the “devel-

opment” of that society. Therefore, the analysis of trans-

portation systems should be based on two basic prem-

ises:

1) The total transport system within a given socio-

economic environment must be considered and viewed

as a single multimodal system.

2) The interaction between the total transport system

and the surrounding socio-economic activity system

must be taken into account in the analysis.

2.3. Basic Variables and Relationships

Based on the above premises, the resultant effect of the

interactions between the activity and transport systems is

manifested in the flow pattern distributed on the different

elements of the transport system. Therefore, we can de-

fine three basic variables for the analysis:

T: The Transport System,

A: The Activity System

F: The Flow Pattern

The interrelationships among these three variables are

shown on Figure 1 and may be described as follows:

Relationship I: T and A determines F

Relationship II: F cause changes over time in A (e.g.,

an increase of flow on a given route may induce more

activities to shift along that route).

Relationship III: F causes changes over time in T (e.g.,

a congestion of flow may influence the decision to build

a new road or to modify existing one).

M. K. HASAN.

274

Figure 1. Basic variables and relationships in transporta-

tion system analysis.

In order to complete the description of the basic

framework of the analysis it is essential to identify the

major individuals, groups or institutions whose decision

could influence and change any of the three basic vari-

ables of the analysis T, A and F. Six major groups can be

identified: User, operators, owners, regulators, govern-

ment, and society at large (a brief description of each

group has been introduces earlier).

2.4. Basic Issues

Having defined the basic framework of analysis let us

turn our attention to the basic issues involved. The first

basic issue is related to the available “options” through

which the six major groups can influence the system.

The second basic issue is concerned with predicting the

possible “impacts” of these “options” on the same groups.

Thus the analysis can be described by firstly identifying

the available options, secondly identifying the possible

impacts, and thirdly explaining the process of predicting

the set of impacts resulting from a given set of options.

2.5. Options

Options or “decision variables” are those aspects of the

transport and activity systems which can directly be

changed by the decision(s) of one or more individuals or

groups. It is, therefore, natural to divide the available

options into two categories, the first includes those re-

lated to the transport system and the second includes

those related to the activity system.

1) Transportation options:

Transportation options are those decisions which can

mainly influence the transport system’s performance by

changing aspects of the network, technology, operating

policies, and/or institutional structures as follows:

a) The network may be influenced (by the owners)

through the geometric and structural design of different

links, intersections and terminals, the network topology

and hierarchy, traffic signals, signs, markings, parking

facilities, etc.

b) Technological options may include the use of elec-

tric or solar power for vehicles, the use of containers, the

introduction of a new rapid transit system, etc.

c) Operating policies may include vehicle routing,

scheduling, pricing, exit and entry regulations, financial

regulations, laws, ordinances, etc.

d) Institutional options may include the number and

types of institutions, the functions of different instit-

utions, the domain of responsibilities, channels of com-

munications, coordination, control, etc.

2) Activity options:

Activity options are those decisions which can mainly

influence transport demand and which are, in general,

not controlled by the decision makers of the transport

system. People in the society have a wide range of op-

tions about how, when and where they would conduct

their activities. Two types of decision should be consid-

ered:

a) long term decisions:

-The location of residence

-Employment

-Scale and pattern of activities

These decisions determine the spatial distribution of

socio-economic activities and land uses in a given area.

Within this context, the actual transport demand will be

influenced by

b) short term decisions (travel options):

-Trip purpose

-Time of trips

-Frequency of trips

-Trip mode

-Trip route

2.6. Impacts

Impacts are those aspects of the transportation and activ-

ity systems that would be influenced by the implementa-

tion of alternative options and would consequently in-

fluenced the six major groups of the system: users, op-

erators, owners, regulators, government, and the society

at large. In order to predict and evaluate these impacts it

is useful to categorize them according to the affected

groups:

1) User impacts

Users are mainly influenced by the level of service of

the transport system, and hence, their impacts variables

would include:

- Travel time

- Travel cost

- Safety

- Comfort

2) Operators impacts

Operators are mainly concerned with maximizing their

share of demand and/or their net revenues, and hence,

their impact variables would include:

- The costs of several resources consumed in the op-

M. K. HASAN.275

eration

- Maintenance and investment of their vehicle fleets

- Their shares of transport demands

3) Owners impacts

Owners are mainly interested in the usage and condi-

tion to their infrastructural elements, and hence their im-

pacts would be reflected through the investment, main-

tenance and upgrading costs which are, in general, func-

tions of the usage of the transport network.

4) Regulators impacts

Regulators are mainly interested in the safety aspects

of operation.

5) Society impacts

The impacts on the society at large include the effects

caused by the physical presence of transport facilities,

such as:

- Noise

- Pollution

- Relocation of residents and activities

6) Governmental impacts

Governmental impacts will depend upon whether the

given government institution is a user, operator, owner,

regulator or representing the society at large.

It should be obvious that the above impacts will, in

general, have differential effects on different groups, or

even subgroups of the above six major ones. That is,

some will gain while others may lose.

2.7. Prediction Process

As indicated earlier, the core of the analysis is the proc-

ess of predicting the “impacts” of alternative sets of “op-

tions”. A central aspect of this process is the prediction

of an “equilibrium” flow pattern on the transport system

resulting from a given set of options. Once this equilib-

rium flow is predicted, the different impacts variables

can be estimated through a set of impact models. The

prediction process may, then, be described as follows:

1) Specification of the transport system options T es-

tablishes performance functions, J. The function J indi-

cates how the level of service S varies as a function of

the transport options T and the volume of flow V. That is,

S = J (T,V)

2) Specification of the activity system options A estab-

lishes demand function, D. The function D indicates the

volume of flow V as a function of the activity options A

and the level of service S, that is,

V = D (A, S)

3) The flow pattern F consists of the volume Vand the

level of service S on the different elements of the total

transport system. That is,

F = (V, S)

4) For a given set of options T* and A*, the resulting

equilibrium flow pattern, F*, can be computed by solv-

ing the performance and demand function. That is,

(*, )*( *,*)

(*,)

SjTV

VDAS







3. The Methodological Framework for

Urban Transportation Systems Analysis

Figure 2 shows the main components of the method-

ological framework for urban transportation systems ana-

lysis as a Traffic Congestion Management System (TCMS)

that we will develop in this research paper and which

depends on the transportation planning steps described

earlier.

3.1. An Overview

This framework can be described as follow:

1) Socio-Economic Environment Description

This part of the frameworks represents the diagnoses

and the analysis of the socio-economic environment

characteristics and factors of the urban area that should

be considered in the analysis and developments of the

transportation system. These factors may include the

population distribution, income, car ownership, and land

use of the study area. In this part, we can also define the

transportation problems that should be addressed and

clearly and specifically determined the symptoms, cause,

and consequences for each problem in addition to the

objectives and goals that should be satisfied by solving

these problems. Then different alternative can be gener-

ated through different options of the six groups’ users,

operators, owners, regulator, government, and society at

large as in next parts (b) and (c). A related data collec-

tion should be performed in this part.

2) Users Characteristics and Behavior

This part represents the transportation demand side

where there are, for example, the following alternatives

(options or decision variables):

- Residence Locations

- Economic& Social Activities Locations

- Frequency of trip

- Destination of Trip

- Trip Purpose

- Travel Mode

-Time of Trip

- Work Locations

- Trip Route

3) Owners, Operators and Regulators Behavior

This part represents the transportation system per-

formance side where there are, for example, the follow-

ing alternatives (options or decision variables):

- Network Configuration (e.g., links, nodes, intersec-

tion, traffic signals location, waiting areas)

- Investments and maintenance Policies for the trans-

portation network and facilities

M. K. HASAN.

276

- Vehicle fleets characteristics

- Routing, scheduling and pricing policies

- Laws, regulations and controls

- Institutions

4) Transport Demand

As a result of a combination of alternatives (options) of

users characteristics and behavioral in part (b), a set trans-

port demand functions (in fact a set of transportation de-

mand models) can be defined. These demand functions will

be functions of user behavior and system’s performance.

5) Transportation System Performance

As a result of a combination of alternatives (options)

of owners, operators and regulators behavior in part (c), a

set transport performance functions (in fact a set of

transportation performance models) can be defined.

These performance functions will be functions of trans-

portation supply and transportation demand. A combina-

tion of alternatives from part (b) and part (c) together is

considered a complete scenario (alternative) that would

developed (generated) to solve the problem under con-

sideration in part (a). This is considered as step (b) in

transportation planning steps. Parts (d) and (e) required

building transportation demand and performance models

and calibrating these models using the data collected in

part (a). This is considered as part of step (c) in transpor-

tation planning steps.

6) Transportation Network Equilibrium

To analysis any completed scenario (alternative) gen-

erated by part (b) and (c), an equilibrium between the

built and calibrated transportation demand models (V =

D(A,S)) and transportation performance models (S = J

(T,V) should be performed. The output of this equilib-

rium process is the traffic pattern (F = (V,S)) that can be

used in estimate the impacts of the given scenario on

users, owners, operators, regulators, and society at large

(see part (g) next). Part (d) represents, in general, the

activity system Awhile part (e) represents, in general, the

transportation system T at interacting together to produce

the traffic pattern F. This represents Relation I in Figure 2.

This part represents part of step (c) in transportation

planning steps. To compute this equilibrium we will ex-

plain in the next subsection in detail our MSTEM model

methodology.

7) Impacts on Users, Owners, Operators, Regulators,

Government, and Society at large

The traffic pattern result from part (f) can be used in

different impact models that should be developed for the

six groups: Users, owners, operators, regulators, gover-

nment, and society at large (see Section 1.3 part (e)).

This completes step (c) in transportation planning steps.

8) Evaluation and Choice

As a result of the impacts on the six groups in part (g),

each alternative can be evaluated using specified eval-

uation criteria and the best alternative (scenario) should

be chosen. (Step (d) in transportation planning steps)

9) Implementation

It is same as step (e) in transportation planning steps.

The traffic pattern F that used in the alternatives eval-

uation, choice and implementation will cause chan- ges-

over time in activity system, A, through part (b), this

represents Relation II in Figure 1. This traffic pattern

will also cause changes over time in transportation sys-

tem, T, through part (c), this represents Relation III in

Figure 1.

3.2. The Multiclass Simultaneous Transportation

Equilibrium Model (MSTEM)

Single class travel forecasting models assume that all

travelers are similar in their travel-decision characteris-

tics, such as their money-value of the time and their sen-

sitivity to travel times in choosing their origin, destina-

tion and mode of travel, etc. To obtain more realistic

models, travelers are often divided into classes, either by

socio-economic attributes (e.g., income level, car avail-

ability, etc.) or by the purpose (e.g., home-based-work,

non-home-based-work, home-based-shopping, etc.) of

their travel, assuming that travel-decision characteristics

are the same within each class, but differ among classes.

However, the development of this concept of multiple

classes increases the mathematical complexity of travel

forecasting models. In 1986 researchers in Chile began to

implement multiclass combined models emphasizing

route choices in a congested transit network with several

combinations of transit modes, as found in Santiago [37].

This research led to the development of ESTRAUS and

related software, which has been extensively applied to

Santiago as well as other Chilean cities. Florian, Wu and

He [38] proposed a variant of ESTRAUS intended to be

more efficient computationally.

All existing multiclass combined models including

ESTRAUS, consider that the total originating and termi-

nating flows are known, i.e., the trip generation step of

transportation planning process is exogenous to the com-

bined prediction process. This deficiency is accounted in

the STEM model which is the only model that combined

the trip generation in the prediction process, but it is not

a multiclass model. This encourage Hasan and Dashti [36]

to expand STEM model to be multiple user classes model

in terms of socio-economic group, trip purpose, as well

as pure and combined transportation modes, interacting

over a physically unique network. The developed Multi-

class Simultaneous Transportation Equilibrium Model

(MSTEM) will also combine explicitly the departure

time as one of the main components of the prediction

process for the first time and will be considered as a new

generation of new Case 6 of the multiclass model classi-

fication of Boyce and Bar-Gera [39]. The MSTEM will

include all the features of ESTRAUS in addition to the

other features mentioned above and more flexible struc-

M. K. HASAN.

277

ing given by an entropy maximization model in ES-

TRAUS), and modal split and departure time are given

by Multinomial Logit (MNL) models based on the ran-

dom utility theory (instead of hierarchical Logit for mo-

dal split only in ESTRAUS). The MSTEM is formulated

as a Variational Inequality problem and a diagonalization

(relaxation) algorithm is proposed to solve it.

ture for demand models where the trip generation can

depend upon the system’s performance through an ac-

cessibility measure that is based on the random utility

theory of users’ behavior (instead of being fixed as in

ESTRAUS), trip distribution is given by a more behav-

iorally richer Multinomial Logit (MNL) model based on

the random utility theory (instead of being given by an

entropy maximization model in ESTRAUS), and modal

split and departure time are given by Multinomial Logit

(MNL) models based on the random utility theory (in-

stead of hierarchical Logit for modal split only in ES-

TRAUS). The MSTEM is formulated as a Variational

Inequality problem and a diagonalization (relaxation)

algorithm is proposed to solve it. The MSTEM will in-

clude all the features of ESTRAUS in performance thr-

ough an accessibility measure that is based on the ran-

dom utility theory of users’ behavior (instead of being

fixed as in ESTRAUS), trip distribution is given by a

more behaviorally richer Multinomial Logit (MNL)

model based on the random utility theory (instead of be-

3.2.1. MSTEM Modeling

3.2.1.1. Notation

Let G = (N, A)be a multimodal network consisting of a

set of nodes and a set of A links that can represent

any mode of transport m in an urban area. These modes

can be grouped into different nests n that could be multi-

ple pure and combined (combination of pure) modes. A

typical user of class l with trip purpose traveling

from a given origin i at a specific departure time period

to any destination j that is accessible from i can use

any of these modes for his journey. We will use the fol-

lowing notation for the multiclass models:

Figure 2. A framework for urban transportation system analysis as a TCMS.

M. K. HASAN.

278

G = (N,A) A multimodal network consisting of a set of

nodes and a set of A links N

L = User class (e.g., income level, car availability, etc.)

L = Set of all user classes

O = Trip purpose (e.g., home-based-work, home-ba-

sed-shopping, etc.)

O = Set of all trip purpose

I lo = Set of origin nodes for user class l and trip pur-

pose o

i = An origin node in the set I lo for user class l with

trip purpose o

D= Set of destination nodes that are accessible from

a given origin i for user class l with trip purpose o

j = A destination node in the set for user class

with trip purpose o

Rlo = Set of origin-destination pairs ij for user class l

with trip purpose o, i.e., the set of all origins i∈I lo and

destinations j∈

M = Any transportation mode in the urban area

n= Nest of transportation modes that has a spe-

cific characteristics (e.g., pure modes including private

and public or combined modes) that are available for

user class l with trip purpose o travel between ori-

gin-destination pairs

= Set of all nests of modes n that are available for

user class l with trip purpose o travel between ori-

gin-destination pairs ij

= Set of all transportation modes in the nest

for user class l with trip purpose o travel between

origin-destination pairs ij

t= Departure time period for user class with trip

purpose o using mode m in the nest n to travel between

origin-destination pairs ij

= Time horizon of the departure time periods

for users of class l with trip purpose o using mode m be-

tween origin-destination pairs ij

p= A simple (i.e., no node repeated) multimodal path

(i.e., it may include links with combined modes m) in the

multimodal network(, )NA

lonmt

= Set of simple paths for travel from the origin

node i to destination node j in the multimodal net-

work for users of class l with trip purpose o de-

part at time using mode from the nest

of modes .

)NA

tK



mM

a = A link in the set A in the multimodal

lonmt

u= the perceived minimum (generalized) cost of

travel for users of class l with trip purpose o depart at

time t using mode m in the nest n from the origin node i

to destination node j in the set

S= the accessibility of origin as perceived

from user of class l with trip purpose o traveling from

that origin

iI

G= the number of trips generated from origin i for

users of class l with trip purpose o

= the value of the socio-economic variable

that influences trip attraction at destination for users

of class l with trip purpose o

()

lo lo

wwj

gA = a given function specifying how the

socio-economic variable

influences trip attraction

at destination j for users of class l with trip purpose o,

and

= a composite measure of the effect that socio-

economic variables, which, are exogenous to the trans-

port system, have on trip attraction at destination j for

users of class l with trip purpose o.



and lo



for 1, 2, ..., wW



are coefficients to

be estimated, where . 0







= the value of the th



socio-economic variable

that influences the number of trips generated from origin

i for users of class l with trip purpose o







= a given function specifying how the th



socio-economic variable,lo



, influences the number of

trips generated from origin i for users of class l with trip

purpose o, and

E= a composite measure of the effect the socio-

economic variables, which are exogenous to the transport

system, have on the number of trips generated from ori-

gin i for users of class l with trip purpose o



and



for 1, 2, ...,







,oO

are coefficients to

be estimated lL





lonmt

= the number trips of users of class l with trip

purpose otraveling from the origin node to the

destination node and whose already chose the

mode of transport from the nest of modes

iI

jD

mMlo



 and start their trip at the time

tK

lonm

= the number trips of users of class l with trip

purpose otraveling from the origin node to the

destination node and whose already chose the

mode of transport from the nest of modes

iI

jD

mMlo



.

lon

T= the number trips of users of class l with trip

purpose otraveling from the origin node to the

destination node and whose already chose the

iI

jD

M. K. HASAN.279

nest of modes .

n

jD

Km 

T= the number trips of users of class l with trip

purpose o traveling from the origin node to the

destination node.

iI

3.2.1.2. Model Assumptions and Structure

1) Travel cost functions

In single class models are often separable and symmetric,

allowing for convex optimization formulation. In multi-

class models travel costs of one class are affected by

decisions of other classes; hence the cost structure is not

separable, and in general it is not symmetric and does not

allow a convex optimization formulation. We assume the

following:

a) For each link a, the link cost function, A

lonmtlo lolo

amnij

CtMn ijR ,

,lL





, will depend, in general, upon the flow over all

links, the vector f, in the multimodal network (N, A) for

all user class , trip propose, transport mode

nest , transport mode , and departure

time period , that is

oO

oO

n lo

mM

lL

tK

onmt lonmt

ni 

lon mt



 

lonmt lonmt

pap





1 if li

0 ot



lon mt



 

Cf

(lonmt



() ,,

ll olo

tKmM

jRlLoO











)

(1)

where

, ,

,,,)

l ol o

AtKmM

jRlLoO



 

We will also assume that the perceived cost of travel

on any multimodal route (path), is the sum of

travel costs on the links that comprise that path, that is:

lonmt

pP

(), ,,

,,,

lonmtlonmtl o

aij m

lo lo

pPtK

mMijRlL oO







 

(2)

nk belongs to path

herwise

lonmt









,, ,

,,,

lol o

m n

tK mM

jRlLo

 



2) The Jacobian of the link cost functions

()

()() : ,,

,,,

l ol o

CtKmM

nijRlLoO



 

Cf f

is asymmetric

The function form specification of the link cost func-

tion in (1) depends on the application of the model and

how well this link cost function represents the transport

system supply in the urban area of the study. For exam-

ple, if we consider only three nests of transport modes

named and define them as follows:

12 3

,, and nn n



 as pure private transportation modes (e.g., car,

taxi, etc.)



as pure public transportation modes (e.g., bus,

subway, metro, etc.)

nm as combined transportation modes (e.g.,

car/metro, bus/metro, etc.) and used the specified link

cost functions for each of the modes ,

m, and

that used in ESTRAUS [37] in its application to Chilean

city Santiago, MSTEM model will have all the advan-

tages of ESTRAUS from the transport system supply

side representation, especially the capacity constraints

for vehicles of all public transport modes, in addition to,

the advantages of MSTEM model over the ESTRAUS

from the transport system demand side.

That is, the link’s a average operating cost (op-

erating time or generalized a cost), for users of class l,

with trip purpose o, of private transportation mode

(e.g., car, taxi, etc.) depart from his origin at time t. This

is a function of the summation of vehicle flows over all

private transportation modes, user classes and trip pur-

poses (), as well as the fixed flow of public trans-

portation vehicles (

lomt

C



lomt

f

) on linkat time period t, all

measured in equivalent vehicles (e.g., p.c.u.):

(

lomtlomtlomtt

aa a

lom

CC fF

 



(3)

Although the Jacobian of the cost functions vector is

not diagonal, it does turn out to be symmetrical, given

the functional form supposed for cost functions

(every vehicle, from whichever user class, trip purpose

and private transportation mode, produces the same im-

pact on congestion). Nevertheless, this “symmetry” of

the cost functions for private modes, which is a simplifi-

cation, could be relaxed without changing the problem

formulation and its solution algorithm. If more general

cost functions are used, considering, for instance, that

different classes of users of private modes produce dif-

ferent impacts on road congestion, the Jacobian of these

cost functions will be asymmetric just like the one asso-

ciated to the public transport cost functions. Then the

combined problem is asymmetric, independent of the

particular characteristics of the cost functions for private

modes. For every pure public transportation mode

lom

C

the pure service networks can be defined as

(,)

mmm

GNS



where m

N is the set of nodes (m



for ground services that use the road network, such as

buses, and m





where NN







for independ-

ent public transportation services, (e.g., metro)) and m

is the set of transit links (route sections) that belong to

mode m.

M. K. HASAN.

280

The generalized time (cost) functions of the public

transportation links, considering the vehicle capacity

constraints as in De Cea and Fernández [40], (sum of

travel time, waiting time, transfer time, fare, etc.) depend

on the vehicle flow over the road network as well as the

passenger flow in the existing services, as follows:

(,,,)(

()()

()

lomtlomtlomt tlomt

saas

lom

lomt lomt

mtlomtmt ss

sWAIT mt mt

CfFaB

TAR PdCAP





































)

TAR

P







(4)

Where lomt

C: Average or generalized cost on link

for users of class l, with trip purpose o, of public

transportation mode

m (e.g. bus, subway, etc.) at time

period t.

()

lomt

TAR

P: Fare multiplier for users of class l, with trip

purpose o, of public transportation mode m at time

period t.

()

TAR : Fare related to public transportation link s of

mode m at time period t.

()

lomt

WAIT

P: Waiting time multiplier for users of class l,

with trip purpose o, of public transportation mode m at

time period t.



, mt



, tm



: Calibration parameters of waiting time

function, for public transportation mode m at time pe-

riod t.

d: Vehicle frequency of public transportation mode

m over the pubic transportation link s at time period t.

()

CAP : Capacity of public transportation link

mode mat time period t.

lomt

V: Passenger flow of class l, with trip purpose o,

belonging to public transportation mode m at time pe-

riod t, and which use public transportation link s.

lomt

: passenger flow that competes with lomt

V for the

capacity of transit lines belonging to Bs (flow with the

same trip purpose, user class, mode, and time period

that belong to other public transportation links that com-

pete or reduce the Bs line’s capacity, plus flow from oth-

er purposes, classes, modes, and time periods that also

compete for the Bs lines capacity).

It is easily seen, in this case, that the Jacobian of the

cost function vector is non-diagonal and asymmetric. In

addition, the model considers the existence of combined

modes, for example car/metro (private transportation/

public transportation) or bus/metro (public transporta-

tion/public transportation). In each case, the union of the

pure mode networks that compose them forms the com-

bined mode network. The combined modes () are

considered to be formed by two public transportation

modes. Nevertheless, it is important to stress that this

does not limit the model’s general use, since there is no

problem in representing combined modes such as

car-public transportation (as in the application of ES-

TRAUS for the city of Santiago considers combined

modes like car driver-metro and car passenger-metro).

b) Trip Generation (TG)

Following the same line of thought of Safwat and

Magnanti [41], the accessibility of origin as

perceived from user of class l with trip purpose o travel-

ing from that origin can be defined as follows:

Slo

iI

maxmaxmaxmax

lolonmt

jDnmMtK

lololo lo

iij

iIlL oO



 j



















 

where = the expectation operator.





max 0,lnexp

lo lo lonmtlo

iiij j

jD nmM tK

lololo lo

iij

S(θuA )

iI lLoO

 





 



(5)

The number of trips generated from origin for us-

ers of class l with trip purpose o, can be expressed

by:

()

lolo lololo

GS qE

iI lLoO















 

lololololo

iii

GSE iIlLo



 

(6)

Similar tolo

, is assumed to be a fixed constant

during the time period required to achieve short-run

equilibrium, and depends solely on the system’s

performance as measured by the accessibility variable

c) Trip Distribution, Nest/Mode Split, and Departure

Time Logit Models (TD/MS/DT)

Following the same line of thought of Safwat and

Magnanti [41], Oppenheim [42] and Ran and Boyce [43]

our distribution, nest/mode, and departure time Logit

models can be given by:

exp(

lo lonmtlo

iij j

nmMtK

lo lo

ij ilo lonmtlo

iij j

jD nmM tK

lolo lo

lololo lo

iij

θuA )

TG θuA

ijRlLo O

 











 

)

(7)

exp(

,,,

lo lonmtlo

iij j

mM tK

lon lo

ij ijlo lonmtlo

iij j

nmMK

lo lo

lolo lo

θuA )

TT θuA

nijRlLoO



 









 

)

(8)

M. K. HASAN.281

exp

,,,,

lo lonmtlo

iij j

lonm lon

ij ijlo lonmtlo

iij j

mM tK

lo lolo

nij

lo lo

(θuA )

TT (θuA )

mMnijR lLoO













 

(9)

exp() ,

exp

,,,

lo lonmtlo

iij j

lonmt lonm

ij ijlo lonmtlo

iij j

lolo lo

mnij

TT (θuA )

tK mMn

ijRlLoO











 

 

(10)

d) Trip Assignment (TA)

Based on the previous choices assumption, the given

user will choose his or her route according to Wardrop’s

user equilibrium principle. That is, for all users of class l

with trip purpose o traveling from the origin node

to the destination node and whose already

chose the mode of transport from the nest of

modes and start their trip at the time ,

the perceived generalized costs on all used multimodal

paths between the given origin-destination pair are equal

and not grater than those on unused paths. This gives the

following equilibrium conditions:

iI

tK

jD

lo

n

if 0,

if 0

lonmt lonmt

ij p

lonmt

plonmt lonmt

ij p

Cuh

















)

,,,

lonmt lolo

ijm n

lo lo

pPtK mM

nijRlLo



 

(11)

where

,, ,

,,,

lonmtlonmt lonmt

lL oO ijRp P

lo lo

lo lonmt

aAtKmM

nijRlLoO



 

 



 

3.2.1.3. Variational Inequality Formulation for

MTEM

Because of the asymmetry of the link cost functions,

MSTEM cannot be cast as an equivalent optimization

program as STEM. Instead, it can be formulated as the

following variational inequality (VI).

** **

()()()()

feasible ,





CfffUTT T0

fT (12)

where

f: vector of flow on links of the multimodal network

f: vector of equilibrium flow on links of the multimo-

dal network

T: vector of trips between origin-destination pairs of the

multimodal network

(:, ,

ilo lo

iI lLoOTT

T: vector of equilibrium trips between origin - destina-

tion pairs of the multimodal network

()Cf : column-vector of network link's cost functions

(with non-diagonal and asymmetric Jacobian)

()UT: column-vector of inverse demand functions

(with non-diagonal and symmetric Jacobian),



ilo

u, ,

iI lLoO)

The VI problem in (12) is Equivalent to MSTEM (see,

for example, Smith [44,45] and Dafermos [46] for a for-

mal proof of equivalency between VI and traffic equilib-

rium) and can be solved by the relaxation (diagonaliza-

tion) algorithm (see, for example, Dafermos [47], Florian

and Spiess [48], Mahmassani and Mouskos [49], and

Sheffi [50])

At each iteration of the diagonalization algorithm, the

cost functions of (1) result in diagonalized cost

function , which depend only on their own

flows,

lonmt

ˆlonmt

nmtlo

, and the following VI should be solved:

** **

ˆ()()( )()

feasible ,







CfffUTT T0



(13)

This VI can be formulated as the following Equivalent

Optimization Program (ECP):

(,,)

lL oO iIjD nmMtK

lolololo lo

iij

Min Z





 

STf

ˆ()[( )

lonmt lo

aAlLoO iI i

lonmt

Cxdx S















()ln()

lo lolo lololo lolo

iii i

SSESE

 

 

]

1[ln(

lonmt lonmt

ij ij

lL oO iIjD nmMtK

lolololo lo

iij





 )

E,,

iI lLoO 

lo lo

iIjD lLoO

]

lo lonmtlonmt

jij ij

AT T

s.t.

lolo lolo

iji i









lo lon

ij ij

nlo



 i





lon lonm

ij ij





,, ,,

lo lolo

iij

iIjD nlLoO





,,,

lonmlonmtlolo

ij iji

lo lo

nij

TTiIjD

mMnlL oO











M. K. HASAN.

282

I

lonm

T,

lonmtlonmt lo

ij p

lo lolo

lonmt

Thi

jD tKmM

nlLoO







 

 

iI lLoO 

,,,

lo lo

iIjD lLo

,, ,

lo lolo

iij

iIjD nl 

lo lolo

iIjD mM

nlLoO

 

 

, ,

,,,

lo lo

mnij

TiIjD

mMn

 

 



, ,

,,,

lo lo

mnij

lonmt

hiIjD

mM n

oOpP



 

 

,, ,

,,,

lonmtlonmt lonmt

lL oO ij Rp P

lo lo

lo lonmt

aAtK mM

nijRlLo



 

 

 



S

T O

lon

T ,LoO 

lonmt



 

lonmt





 

where

Existence, convexity and uniqueness of ECP problem

as well as equivalence between MSTEM and ECP can be

followed as those of Safwat and Magnanti [41].

The Logit Distribution of Trips (LDT) algorithm that

developed by Safwat and Brademeyer [51] can be modi-

fied as Multiclass Logit Distribution of Trips (MLDT)

algorithm to solve the above ECP. For more details about

the MSTEM methodology see Hasan and Dashti [36].

3.3. Data Collection Requirements

To apply the MSTEM to any urban transportation net-

work we need the following data collection:

1) Household Survey

Trips mad by all household members by all modes of

transport both within the study area and leaving/arriving

to the area during the survey period

 Personal and household characteristics and identi-

fication

- Head of household, wife, son,…, etc.

- Sex, age, possession of a driving license, education

level, and occupation

 Socio-economic information: income, car owner-

ship, family size

 Trip data: origin, destination, purpose, start and

ending times, mode used, amount of money paid for the

trip, and so on

2) Traffic Counts

An actual observed traffic counts in specific locations

in the city road and highways should be collected for at

different time of the day (specifically in the peak hour),

different day of the week, and for different kind of vehi-

cles types.

3) Roadside Interviews

These provide useful information about trips not reg-

istered in the household survey (i.e. external-external in

cordon survey). It involves asking a sample of drivers

and passengers of vehicles (e.g. cars, public transport,

goods vehicles) crossing a roadside station, a limited set

of questions; these include at least origin, destination and

trip purpose.

4) Network inventory

a) Road network

The following link attributes for each link:

1- From node and To node

2- Length

3- Its travel speeds-either free-flow speeds or an ob-

served value for a given flow level

4-The capacity of the link

5- Type of road (e.g., expressway, trunk road, local

street)

6- Road width, or number of lanes, or both

7- An indication of the presence or otherwise of bus

lanes, or prohibitions of use by certain vehicles (e.g.,

Lorries)

8- Banned turns, or turns to be undertaken only when

suitable gaps in the opposing traffic become available

9- Type of junction and junction details including sig-

nal timings

10-

Storage capacity for queues and their presence at

the start of a signal

b) Transit network

1. Lines

2. Fares’

3. Frequency

4. Travel time measurements

5) Land use inventory

a) Residential zones (housing density)

b) Commercial and industrial zones (by type of estab-

lishment)

c) Parking spaces

6) Cordon Surveys

These provide useful information about external-

external and external-internal trips. Their objective is to

determine the number of trips that enter, leave and/or

cross the cordoned area, thus helping to complete the

information coming from the household O-D survey.

7) Screen-Line Surveys

Screen lines divide the area into large natural zones

M. K. HASAN.283

(e.g. at both sides a river or motorway), with few cross-

ing points between them. The procedure is analogous to

that of cordon survey and the data also serve to fill gaps

in and validate the information coming from the house-

hold and cordon surveys.

8) Socio-economic data

To compute lo



, the value of the th



socio-econo-

mic variable that influences the number of trips gener-

ated from origin i for users of class l with trip purpose o,

the Socio economic data for each origin i could be popu-

lation, income, car ownership, ….etc.

To compute lo

, the value of the socio-econ-

omic variable that influences trip attraction at destination

j for users of class l with trip purpose o, The socio-

economic data for destination node j could be the ground

floor area of the land use type of destination j , employ-

ment size, …..etc.

3.4 Analysis and Evaluation of Effective

Solutions for Peak-period Traffic Congestion

Peak-period traffic congestion occurs when travel de-

mand exceeds the existing road system capacity, Sandra

Rosenbloom [52]. There are two basically alter native-

solutions:

I Change demand (Users Characteristics and Beha-

vior in Figure 2) to meet system capability

II Change system capacity (Owners, Operators and

Regulators Behavior in Figure 2) to meet demand.

The demand for road system capacity may be changed

by:

a) Reducing the number of vehicles used to meet the

existing travel demand by increasing vehicle occupancy.

b) Reorienting travel to off-peak periods.

c) Reorienting travel to less congested alternative

routes.

d) Reducing the total demand for travel itself.

System capacity itself can be changed by:

a) Constructing additional roadway, either adding

lanes to existing routes or providing new routes or new

transportation modes.

b) Increasing the capacity of the existing road infra-

structure by engineering techniques that improve traffic

flow.

Sandra Rosenbloom [52] shows that 22 techniques

(see Table 1) which appeared to have the potential to

reduce or redistribute demand were fell into four major

categories of approaches:

a) social approaches which seek to alleviate conges-

tion utilizing techniques that change social behavior or

personal interactions

b) socio-economic approaches which utilize tech-

niques that induce favorable travel changes by manipu-

lating broadly defined economic penalties and incentives

c) socio-technical approaches which utilize technol-

ogy resources to modify social and economic behavior in

ways that ultimately reduce congestion

d) technical approaches which use technical devices to

modify directly dysfunctional travel behavior at the im-

mediate congestion site or source.

These Twenty two techniques were grouped under

these four major approaches cover a wide range of social,

behavioral and economic incentives, it was extremely

difficult to make comparisons among them. This diffi-

culty was heightened because these techniques affected

peak-period traffic congestion directly and indirectly in

different ways. Sandra Rosenbloom [46] illustrated the

differing impacts of these techniques in practice where

most of the non-engineering techniques considered to be

applicable to Peak-Period traffic congestion were not

designed with the reduction of peak period traffic con-

gestion as their primary goal; the reduction of traffic

congestion was usually considered to be a secondary or

indirect impact of their implementation. For such tech-

niques to significantly affect traffic congestion, they

must have first successfully met some other goal－their

primary goal. However, it was often difficult to get pro-

ject goals clearly articulated, and the relationship be-

tween those goals and traffic congestion identified.

Based on this analysis (see Sandra Rosenbloom [46]),

the study team evaluated 17 of these techniques as both

effective and feasible in a U.S. institutional context.

However, none of these 17 offered more than marginal

reductions in peak-period traffic congestion when ap-

plied individually. Some techniques affected so small a

percent-age of travelers that reductions in congestion

would not be discernible. Other techniques promised

significant congestion reductions in theory but did not

realize that promise in practice. It was concluded that

many techniques could be implemented together with the

potential for far greater combined effectiveness. An

analysis was performed to determine how best to "pack-

age" or jointly implement promising techniques to opti-

mize their combined effectiveness. It was found that all

promising techniques could not be applied together be-

cause of conflicts in their impact. This analysis suggested

eight sample "packages" or combinations of mutually

supportive techniques. These eight packages were sub-

jected to evaluations similar to those performed for indi-

vidual techniques; while the packages are merely exam-

ples of potential combinations, the evaluation methodol-

ogy employed should be of continuing use to local

transportation planners.

The development of the packaging concept is consis-

tent with the development of the Transportation System

Management (TSM) concept in the U.S. The conclusions

of the study should also have merit for metropolitan ar-

eas considering management schemes. They are:

- Promising congestion reduction techniques or pack-

ages must be applied to the appropriate congestion situa-

M. K. HASAN.

284

tions.

- Promising techniques have not been successful alone

in significantly affecting peak-period traffic congestion

- All promising techniques cannot be implemented to-

gether because some seriously conflict with one, and -

Promising congestion reduction techniques may have

serious social, economic and physical side-effects which

must be recognized prior to implementation

The above study summary shows the guidelines based

on empirical results that can be helpful for city planner,

traffic engineers or transportation authorities' decision

makers. But how we help these people to perform these

analysis and guidelines in some systematic or even intel-

ligent way?

4. The Architecture of the Intelligent

Decision Support System for Traffic

Congestion Management System

The architecture and the information flow shown in Fig-

ure 3 represent a high level blueprint for the implemen-

tation of the framework for an Intelligent Decision Sup-

port system Traffic Congestion Management System.

The framework is derived and guided by the methodo-

logical framework for urban transportation system analy-

sis presented in Section 3 and depicted in Figure 2 and

the adoption of the main components of the Decision

Support System (DSS) is guided by standard DSS text-

book such as Turban et al. [53]. A fourth component, the

scenario management was added to package the func-

tionally required by scenarios creation, storage, retrieval,

analysis, evaluation and reporting. An Intelligent Agent

for supporting scenario creation is also included in the

frame-work. Figure 3 shows the main components of the

DSS, their interactions (data and control flows) with the

Transportation Object Repository, with each other and

with the User Interface Management Subsystem (UIMS)

directly or indirectly. In the following section

we briefly describe each of these components.

4.1. Urban Transportation Object Repository

(UTOR)

The Urban Transportation Object Repository (UTOR) is

an object oriented repository storing various transporta-

tion objects such as nodes, links, and zones for multi-

modal urban networks; equilibrium models; impact

models, scenarios and user interfaces that are managed

by the four subsystems discussed bellow. The four

UTOR object stores are distinct but are integrated. These

are:

1) Urban Transportation GIS-Data Store: This object

store contains two distinct components:

Table 1. Four approaches and 22 techniques reduction of

peak period traffic congestion.

Social Approaches

- Staggered Work Hours

- Shortened Work Weeks

Socio-Economic Approaches

Pricing and Regulatory Mechanisms

- Road Pricing

- Parking Controls

- Restricting Access

-Traffic Cells

- Auto-Free Zones with Facilities for Pedestrians

Land Use Planning

-New Towns

-Planned Communities

-Planned Neighborhoods

-Zoning and Building Codes

Marketing

-Incentives to Off-Peak Travel and Usage of Facilities

-Incentives to Mass Transit Usage

-Carpooling and Other Forms of Ride-Sharing

Approaches

Socio-Technical Approaches

-Communications in Lieu of Travel

Technical Approaches

Traffic Engineering Techniques

-Freeway Surveillance and Control

-Traffic Simulation Models

-Maximum Use of Existing Facilities

Transit Operations

-Extended Area Services

-Priority Systems: Expressways and Freeways

-Priority Systems: Arterials

Circulation Systems

Vehicle Design Factors

a) GIS Object Base: Contains various GIS Objects.

b) Transportation Data Warehouse (TDW). TDW is a

multidimensional, object-oriented, nonvolatile integrated

database containing various current and historical data

about the transportation Objects (Inmon [54]). The TDW

and GIS Objects are populated by the Extract Transform

and Loading (ETL) component of Data Management

subsystem (DMS) mentioned bellow.

2) Urban Transportation Model Base: contains various

transportation equilibrium and impact models.

3) Scenarios Base: contains various scenarios objects

that are created over times.

4) User Interface Base: contains various user interface

(UI) objects created over time.

4.2. Data Management Subsystem (DMS)

The data management subsystem (DMS) is responsible

for the data administration such as creation, storage, re-

trieval of node object, links object, and zone object for

different modal network. DMS manages the Urban

Transportation GIS-Data Store which contains two dis-

tinct but integrated data bases:

a) a GIS database contain the spatial data;

b) a transportation data warehouse mentioned above.

M. K. HASAN.285

4.3. Extract, Transform and Load (ETL)

Component

ETL component of the DMS extracts data from multiple

sources, cleanse them and transform the data from its

original to a form that could be place in TDW guided by

the metadata, and then load the data into the TDW. The

purpose of ETL is to populate TDW with integrated and

cleansed data required by the transportation objects ga-

thered from three main sources, the socio-economic, the

demand data and the supply data.

4.4. Model Management Subsystem (MMS)

MMS is responsible for the creation, storage, retrieval of

the transportation object models which are stored in the

Model Object Base (MOB) for utilization or reuse.

Two types of models are managed by MMS. These are:

a) The Transportation Network Equilibrium Models

(MSTEM)

b) Impacts Models: such as impact on Users, impact

on operators, impact on owners, impact on society, and

impact on government discussed earlier in Section 3.

4.5. Scenario Management Subsystem (SMS)

Much like MMS, SMS is responsible for the creation,

storage, retrieval, analysis, evaluation of scenarios. An

important part of SMS is the impact evaluation compo-

nent that assesses the impact models and presents the

assessment results to the SMS. Objects are stored in

Scenario Object Base (SOB) for future utilization or re-

use by SMS and Intelligent Scenario Creation Assistance

Agent (ISCAA) described bellow. SMS retrieves previ-

ously created scenarios and pass them to the User Inter-

face Management Subsystem (UIMS) as initial scenarios

on which further what-if analysis could be performed.

4.6. Intelligent Scenario Creation Assistance

Agent (ISCAA)

The complexity of creating the right scenario or retriev-

ing the right scenario from the previous created ones

stored in the scenario base is a complex process requires

human expertise which is scarce. An intelligent compo-

nent within the DSS framework that would look at the

historical scenario objects and assist and guide the deci-

sion maker in choosing the best alternatives from this

pool of historical scenario object to be included in the

initial scenario setup is extremely valuable. A solution

for this problem is to create an Intelligent Scenario Crea-

tion. Assistance Agent (ISCAA) that would encode and

encapsulates the expertise for scenario creation and

would provide the necessary assistance for creating the

right scenario. ISCAA would be a hybrid intelligent

agent containing multiple computational intelligent tools

(such as ANN, Rough Set, Fuzzy Logic, etc.) as well as a

set of scenario creation rules.

4.7. User Interface Management Subsystem

(UIMS)

The UIMS packages and manages the functionalities

require for creating a data-rich intensive (maps, graphs,

text, and structured data) with various visualization ca-

pabilities user interface. Since The DSS is to be used by

users with various roles (Transportation Planners, Trans-

portation Engineers, Transportation Decision Makers or

Traffic Administrators), the complexity involves in dy-

namically creating the right graphical user interface (GUI)

for the right role lies within the functionality of UIMS.

For example, the transportation planner is responsible for

creating models and capturing the right data for those

models, as such he/she would directly interact with

Model Management and Data Management and as such

the GUI for this role would configure that would allow

for that only. On the other hand, the transportation deci-

sion maker role deals with scenarios and as such the

UIMS would create the proper GUI allowing various

scenario related activities such as scenario creation, re-

trieval, storage and execution and presenting the result in

a dashboard view allowing for a comprehensive, at-a-

glance, GIS-Based graphical view of the solution gener-

ated. UIMS also provides various analysis tools such as

what-if analysis, sensitivity analysis, reporting the result

of impact evaluation and providing various Ad hoc que-

ries and reports. The Graphical Interface Objects, that are

created, are stored as UI Objects in the UTOR and man-

aged by UIMS.

All components of CMS can be adapted through the

Users Characteristics and Behavior (Activity System

Options) and Owners; Operators; and Regulators Behav-

ior (Transportation System Options) in Figure 2, in ad-

dition, to any roadway capacity expansion through the

Transportation System Options. In other word, our frame-

work in Figure 2 can be adapted to any of the single 22

techniques in table one or any combinations (packages)

of these techniques, but the evaluating and recommend-

ing alternative strategies for CMS will be done through

the results of the total evaluation of Impact Models (the

positive and negative parts) on Users, Owners, Operators,

Regulators, Government, and Society at large for each

alternative or packages of alternations. All the alterna-

tive can be reflected through the MSTEM model demand

and performance (link cost functions) parameters and

socio-economic and land use variables.

5. Summary, Conclusion and Future

Research

5.1. Summary

Traffic congestion is a very complicated problem that

M. K. HASAN.

286

affects the social, economic and development aspects of

many countries around the world. The relationship be-

tween the traffic congestion problem and socioeconomic

characteristics of trip makers and the rapid increase of

urbanization of the land use make the problem becomes

much more complicated and calls of urgent solution.

Many techniques for solving this problem were sug-

gested starting from empirical studies for a single tech-

nique or packages of techniques , to advanced Transpor-

tation System Management (TSM) and finally to Con-

gestion System Management (CSM). Most of these tech-

niques tried to maximize the utilization of the existing

transport system (Roads, Highways, and Transit systems)

by certain management techniques prior to any expan-

sion or addition to the transport system.

US Federal transportation legislation requires Metro-

politan Planning Organizations (MPO) to develop and

implement a CMS as part of the metropolitan transporta-

tion planning process [22]. Most of MPOs use traditional

transportation planning modeling approach and a com-

mercial specialized software as a part of this process to

evaluate traffic relive and the length and cost for deferent

alternative solutions prior to submitting their recom-

mendations. This process requires highly educated,

trained and expert personnel in transportation planning

and engineering field. It would also require very user

friendly transportation planning software that includes a

state of the art transportation network equilibrium mod-

els. Models that are significantly more advanced than the

traditional four steps sequential. Models that combine

single class Trip Generation- Trip Distribution- Modal

Split-Trip Assignment (TG-TD-MS-TA) [41], or a mul-

ticlass combined model [36] that combine TG-TD- MS-

TA in addition to Departure Time for different class of

traveler depend on income, car ownership, trip pur-

pose, … etc.

As Boyce [55] says, “combined or integrated models

of origin-destination, mode, route and time period choice

can be implemented and solved with practitioner soft-

ware systems such as CUBE, EMME/2, PTV Vision,

QRS II, SATURN or TransCAD, if the travel forecaster

has a detailed understanding of the models and the solu-

tion algorithm. Ongoing discussions with practitioners

reveal that few practitioners have this requisite expertise.

The observed inability of many practitioners to solve the

sequential procedure with feedback in a convergent

manner is one indication of this dilemma. An alternative

approach is to design a software system which directly

solves such an integrated model, including providing a

number of options to the user. This approach was taken

by the Chilean software vendor, MCT, in the develop-

ment of its product ESTRAUS. The distinction between

these two software development philosophies may be

more subtle than is generally appreciated. The former

approach offers the forecaster a “tool kit” with which to

“build” a model. Then the practitioner must acquire the

expertise to use it, an uncertain and arduous process with

many pitfalls. The latter provides a “canned” model and

an algorithm for solving it. When presented with a de-

scription of ESTRAUS, a typical user’s comment is,

“This is interesting, but I don’t know how we would use

it to solve our model.” This comment may offer a clue to

the dilemma of vastly upgrading or revolutionizing prac-

tice, which presumably is one of the aims of academics

as well as software vendors. Would a standard set of

models with numerous user options provide a more ef-

fective pathway to improving travel forecasting practice

than attempting to upgrade practice model by model?

Certainly, this option is worth considering, so long as

several of the vendors can effectively participate in the

software market”.

5.2. Conclusions

The main conclusions of this research paper can be

summarized as follows:

1) In this research paper, it was possible to integrate

all of the above requirements in a unified framework of

analyses that capture the concept of the interaction of the

transport system with the activity system in one Trans-

portation System Analysis (TSA) approached. Develop-

ing this TSA framework as a TCMS within an Intelligent

Decision Support System (IDSS), that can be a useful

decision support tool for transportation planners and

transportation decision makers for the analysis and

evaluation of strategic transportation plans include

transportation projects and policies and utilizing the stat

of the art of the Multiclass Transportation Equilibrium

Model MSTEM [36] within an easy to use environment

that would eliminate or reduce the necessity for highly

educated and trained transportation specialists and deci-

sion makers, is the cornerstone of this paper.

2) The paper presents many transportation concepts

and integrate all of them in such a comprehensive

framework that is easy to follow and undetectable for

practitioners transportation planners, transportation En-

gineers, or transportation decision makers.

3) The Intelligent Scenario Creation Assistance Agent

(ISCAA) and the Scenario Management Subsystem

(SMS) components of the IDSS architecture will be the

most important components that will reduce the need of

the expertise and high education level of the transporta-

tion planners and transportation engineers or any trans-

portation decision makers. This IDSS will reduce the gap

between the practitioners and software vendors and in-

crease the usability of the most sophisticated and more

behaviorally relevant Multiclass Transportation Equilib-

rium Model like MSTEM.

5.3. Future Research

The author started a development of data collection pro-

M. K. HASAN. 287

Figure 3. The architecture and information flow of the intelligent decision support system for urban transportation systems

nalysis as TCMS. a

gram for Saint Louis Region area during his sabbatical

year at University of Missouri at Saint Louis. These data

will be used in the calibration and validation process for

STEM, MSTEM, and ESTRAUS models for future

comparisons among these models. The results of these

comparisons will be reported in upcoming papers

The development of a prototype, derived from the

framework described in this paper is also under way.

The result of the experimentation with the prototype, in

terms of its usability, degree of ease and its effectiveness

in providing optimum solutions/scenarios to transporta-

tion and decision makers will be reported in upcoming

papers.

6. Acknowledgements

The author wish to thank the Research Administration

(RA), Kuwait University for granting the financial assis-

tance and sponsoring this research. This research is sup-

ported by Kuwait University, Research Grant No.

[IQ04/04].

. References

[1] E. F. Ogunbodede, “Transport Management Techniques

of Rapidly Changing Urban Land Use,” The Case of

Akure International Journal of Environment and Devel-

opment, Vol. 1, No. 1, 1998, pp. 81-86.

[2] E. S. Mills, “Studies in the Structure of the Urban Econ-

omy,” The Johns Hopkins Press, Baltimore, 1972.

[3] D. C. Gazis, “Optimum Control of a System of Over-

saturated Intersections,” Operations Research, Vol. 12,

1964, pp. 815-831.

[4] R. L. Gordon, “A Technique for Control of Traffic at

Critical Intersections,” Transportation Science, Vol. 4,

1969, pp. 279-287.

[5] D. Longley, “A Control Strategy for a Congested Com-

puter Controlled Traffic Network,” Transportation Re-

M. K. HASAN.

288

search, Vol. 2, 1968, pp. 391-408.

[6] L. J. Pignataro, W. R. McShane, K. W. Crowley, B. Lee

and T. W. Casey, “Traffic control in over saturated con-

ditions,” NCHRP Report, No. 194, 1978.

[7] N. H. Gartner, C. Stamatiadis and P. J. Tarnoff, “Devel-

opment of Advanced Traffic Signal Control Strategies for

Intelligent Transportation Systems: Multilevel Design,”

Transportation Research Record, Vol. 1494, 1995, pp.

98-105.

[8] D. J. Quinn, “A review of queue management strategies,”

Traffic Engineering and Control, Vol. 33, No. 11, 1992,

pp. 600-605.

[9] A. K. Rathi, “A Control Strategy for High Traffic Density

Sector,” Transportation Research B, Vol. 22, No. 2, 1988,

pp. 81-101.

[10] A. K. Rathi, “Traffic metering: An Effectiveness Study,”

Transportation Quarterly, Vol. 45, No. 3, 1991, pp. 421-

440.

[11] T. S. Reddy, E. Madhu and J. V. R. Reddy, “TSM Ac-

tions to Improve the Traffic Operations－A Case Study,”

ACE, Vol. 2, 2002, pp. 922-932.

[12] J. V. R. Reddy, “Application of TSM Actions－A Case

Study on Mathura Road in Delhi,” Master’s Thesis, Uni-

versity of Baroda, Vadodra, 1998.

[13] J. M. Thomson, “Reflections on the Economics of Traffic

congestion,” Journal of Transport Economics and Policy,

Vol.32, No. 1, 1998, pp. 93-112.

[14] Adefolalu, “Traffic Management and Enhanced Urban

Environment in Nigeria: A Discussion,” In: E. O. Adeniyi

and I. B. Bell-Imam Ed., Op Cit, 1977.

[15] S. O. Oyefesobi, “Measures to Improve Traffic Flow and

Reduce Road Accidents in Nigeria,” In: S. O. Onako-

maiya and K. F. Ekanem Ed., Transportation in Nigerian.

National Development Proceedings of a National Con-

ference, Ibadan, July 1977, pp. 4-9.

[16] J. M. A. Orioke, “Traffic Education and Flow in Ibadan

City,” In: S. O. Onakomaiya and N. F. Ekanm Ed., Op.

Cited, 1981.

[17] M. A. Kennedy and K. Walter, “TSM Measures for Ma-

jor Activity Centers,” Transportation Engineering Jour-

nal, Vol. 105, No. 5, 1979, pp. 499-511.

[18] E. Lindquist, “Assessing Effectiveness Measures in the

ISTEA Management Systems,” Southwest Region Uni-

versity Transportation Center, Texas Transportation In-

stitute, College Station, Texas, 1999.

[19] http://www.countyofkings.com/kcag/Plans_Programs/Re

gional%20Transportation %20Plan%20Section/Final-TS

[20] http://www.madisonareampo.org/Plan%20Elements/Con

gstMngt.pdf

[21] http://www.marc.org/transportation/cms/policy.pdf

[22] http://www.gvmc.org/transportation/CongestionManage

[23] S. Ossowski, J. Hernandez, M. Belmonte, A. Fernandez,

A. Garcıa-Serrano, J. Pe´rez-de-la-Cruz, J. Serrano and F.

Triguero, “Decision Support for Traffic Management

Based on Organizational and Communicative Multiagent

Abstractions,” Transportation Research Part C, Vol. 13,

2005, pp. 272-298.

[24] S. French, “Decision Analysis and Decision Support

Systems,” University of Manchester, Manchester, 2000.

[25] M. Silver, “Systems that Support Decision Makers,” John

Wiley, Berlin, 1991.

[26] M. Klein and L. Methlie, “Knowledge-Based Decision

Support Systems,” John Wiley, Berlin, 1995.

[27] J. Hernandez and J. Serrano, “Environmental Emergency

Management Supported by Knowledge Modelling Tech-

niques,” AI Communications, Vol. 14, No. 1, 2001, pp.

13-22.

[28] J. Hernandez and J. Serrano, “Reflective Knowledge

Models to Support an Advanced HCI for Decision Man-

agement,” Expert Systems with Applications, Vol. 19, No.

4, 2000, pp. 289-304.

[29] I. Vlahavas, N. Bassiliades, I. akellariou, M. Molina, S.

Ossowski, I. Futo, Z. Pasztor, J. Szeredi, I. Velbitskiyi,

S. Yershov and I. Netesin, “Expernet—An Intelligent

multi-agent System for WAN Management,” IEEE Intel-

ligent Systems, Vol. 16, No. 7, 2002, p. 62.

[30] J. Cuena, S. Ossowski, “Distributed models for decision

support,” In: Weiss Ed., MIT Press, Cambridge, 1999, pp.

459-504.

[31] S. Ossowski, J. Hernandez, C. Iglesias and A. Fernandez,

“Engineering Agent Systems for Decision Support,” In: T.

Zambonelli Eds., Engineering Societies in an Agent

World III, Springer, Berlin, 2002, pp. 234-274.

[32] C. Iglesias, M. Garijo Ayestara´n and J. Gonza´lez, “A

Survey of Agent-Oriented Methodologies,” Intelligent

Agents V, Springer, Berlin, 1999, pp. 317-330.

[33] K. Hoa Dam, M. Winikoff, “Comparing agent-oriented

methodologies,” In: Giorgini et al. Eds., Agent-Oriented

Information Systems, Springer, Berlin, 2003, pp. 78-93.

[34] C. W. Holsapple, “Decision Support Systems. Encyclo-

pedia of Information Systems,” Vol. 1, 2003, pp. 551-

565.

[35] H. L. Zhang, “Agent-Based Open Connectivity for Deci-

sion Support Systems,” Ph.D. Thesis, School of Com-

puter Science and Mathematics, Victoria University,

2007.

[36] M. K. Hasan and H. M. Dashti, “A Multiclass Simulta-

neous Transportation Equilibrium Model,” Networks and

Spatial Economics, Vol. 7, No. 3, 2007, pp. 197-211.

[37] J. De Cea, J. E. Fernandez, V. Dekock, A. Soto and T.

L.Friesz, “ESTRAUS: A Computer Package for Solving

Supply-Demand Equilibrium Problem on Multimodal

urban Transportation Networks with Multiple User

classes,” The Annual Meeting of the Transportation Re-

search Board, Washington, DC, 2003.

[38] M. Florian, J. H. Wu and S. He, “A Multi-Class Mul-

ti-Mode Variable Demand Network Equilibrium Model

with Hierarchical Logit Structures,” In: Current Trends,

M. Gendreau and P. Marcotte Eds. Dordrecht, 2002, pp.

1131-1119.

[39] D. E. Boyce, H. Bar-Gera, “Multiclass Combined Models

M. K. HASAN.

289

for Urban Travel Forecasting,” Networks and Spatial

Economics, Vol. 4, 2004, pp.115-124.

[40] J. De Cea, and J. E. Fernandez, “Transit Assignment for

Congested Public Transport Systems: An Equilibrium

Model,” Transportation Science, Vol. 27, No. 2, 1993, pp.

133-147.

[41] K. N. A. Safwat, and T. L. Magnanti, “A Combined Trip

Generation, Trip Distribution, Modal Split and Traffic

Assignment Model,” Transportation Science, Vol. 22, No.

1, 1988, pp. 14-30.

[42] N. Oppenheim, “Urban Travel Demand Modeling: From

Individual Choices to General Equilibrium,” John Wiley

and Sons, New York, 1995.

[43] B. Ran and D. E. Boyce, “Modeling Dynamic Trans-

portation Networks,” Springer-Verlag, Berlin, 1996.

[44] M. J. Smith, “The Existence, Uniqueness and Stability of

Traffic Equilibria,” Transportation Research B, Vol. 13,

1979, pp. 295-304.

[45] M. J. Smith, “The Existence and Calculation of Traffic

Equilibria,” Transportation Research B, Vol. 17, 1983,

pp. 291-303.

[46] S. C. Dafermos, “An Iterative Scheme for Variational

Inequalities,” Mathematical Programming, Vol. 26, 1983,

pp. 40-47.

[47] S. C. Dafermos, “Relaxation Algorithm for the General

Asymmetric Traffic Equilibrium Problem,” Transporta-

tion Science, Vol. 16, No. 2, 1982, pp. 231-240.

[48] M. Florian and H. Spiess, “The Convergence of Diago-

nalization Algorithms for Asymmetric Network Equilib-

rium Problems,” Transportation Research B, Vol. 16,

1982, pp. 447-483.

[49] H. S. Mahmassani and K. C. Mouskos, “Some Numerical

Results on the Diagonalization Network Assignment Al-

gorithm with Asymmetric Interactions between Cars and

Trucks,” Transportation Research B, Vol. 22, 1988, pp.

275-290.

[50] Y. Sheffi, “Urban Transportation Networks: Equilibrium

Analysis with Mathematical Programming Methods,”

Prentice-Hall, Englewoods Cliffs, 1985.

[51] K. N. A. Safwat and B. Brademeyer, “Proof of Global

Convergence of an Efficient Algorithm for Predicting

Trip Generation, Trip Distribution, Modal Split and Traf-

fic Assignment Simultaneously on Large-Scale Net-

works,” International Journal of Computer and Mathe-

matics with Applications, Vol. 16, No. 4, 1988, pp. 269-

277.

[52] S. Rosenbloom, “Peak-Period Traffic Congestion: A

State-Of-The-Art Analysis and Evaluation Of Effective

Solutions,” Transportation, Vol. 7, 1978, pp. 167-191.

[53] T. Efraim, et al., “Decision Support and Business Intelli-

gence Systems,” 8th Edition, Pearson Prentice Hall, New

Jersey, 2007.

[54] W. Inmon, “Building the Data Warehouse, 4th Edition,

Wiley, New York, 2005.

[55] D. Boyce, “Forecasting Travel on Congested Urban

Transportation Networks: Review and Prospects for

Network Equilibrium Models,” Networks and Spatial

Economics, Vol. 7, 2007, pp. 99-128.