You are on page 1of 3

JOURNAL OF COMPUTING, VOLUME 2, ISSUE 9, SEPTEMBER 2010, ISSN 2151-9617

HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 74

Intelligent Transport Scheduling


Ourida Saidi ben Boubaker and Amine Farhat

Abstract— Constraint Satisfaction represents a great modeling formalism for complex problems in the areas of Artificial
Intelligence and Computer science. A constraint satisfaction problem (CSP) is defined by a set of variables and a set of
constraints which translate all the rules and obligations having to be satisfied by the solutions, each variable has a well defined
domain area. CSP has several applications in many real life fields. Therefore, researchers have worked on CSP and have
developed several solving algorithms and have enlarged the scope of application of CSP formalism. Among the interesting
areas of application of CSP, figure transport and scheduling problems. In this article, we addressed the problem of resources
scheduling in a transport context, which consists in optimal resource assigning in order to accomplish transport duties, called
also tour services. Resources are Transport equipments, Drivers and Time. We have modeled this problem as a CSP, and then
we have implemented a resolving web based system with two algorithms: Generate & Test and Backtracking. Our work has
been strongly appreciated by domain experts.

Index Terms — Computational intelligence, Constraint Satisfaction Problems, Transport problems, Intelligent Scheduling,
algorithms

——————————  ——————————

1 INTRODUCTION

T he CSP [1](Constraint Satisfaction Problem) for-


malism is an approach to address difficult computer
science problems and to find optimal solutions. Sev-
company using the CSP formalism. An operating
schedule is the allocation of adequate resources to ensure
all rounds of services. A tour service, also called section,
eral real-life problems can be modeled by this formula- means a basic unit for which an element of the resource
tion, the implementation of resolution algorithms can DRIVERS is assigned to an element of the resource
generate solutions. In this paper, we present our work in EQUIPEMENTS during a period of TIME in order to
the context of solving a problem of scheduling and trans- achieve a clearly defined transport mission. This resource
portat using the CSP formalism. The problem is to con- allocation must obey a set of rigorous operating rules that
struct an array of tour services for a company of urban will constitute the constraints of our CSP. We propose the
following model of the problem as CSP formalism:
transport by ensuring optimal allocation of different re-
sources: Time, Engines and Drivers, while respecting a set - Variables:
of operating rules previously defined. The characteristics Let e1…….en : n transport equipments
of this problem and the difficulties encountered by the d1……dm : m drivers
experts to solve it have encouraged us to solve it within
the CSP formalism given our deep belief in its effective- t1……..ts : s tour services to accomplish
ness in problems of resource allocation under constraints.
These elements represent the resources of our problem
To do this, we have modeled the problem according to
which must be allocated.
the CSP formalism, then, we have developed a system
that implements core programs for solving CSP. Let the variable Xeidjtk: This variable indicates if the
The details of modeling and solving are presented in driver ei is assigned to the engine dj for the execution of
what follows. The conclusions and perspectives of our the tour service tk. This variable takes the value 0 if the
work are presented in the last section of this paper. last clause is false, and takes the duration of resources
allocation time, if any.
2 PROBLEM SPECIFICATION Thus, the allocation of values needed for the different
variables of the problem Xeidjtk will solve the problem
As mentioned, we tried to solve the problem of automatic
generation of an operating schedule of an urban transport and allows the construction of tour services schedule.
Nevertheless, and according to the CSP formalism, this
————————————————
allocation must obey some constraints that we have mod-
 Ourida Saidi ben Boubaker, Phd, Associate Professor, haed of Computer
Science Departement at the Institut Superieur de Gestion de Tunis, Uni- eled as follows:
versity of Tunis, 11, Av. De la liberté, Cité Bouchoucha, 2000, le bardo, - Constraints:
Tunis, Tunisia.
 Amine Farhat, M. Sc., Phd student in Computer Science Departement in For each i,j, ∑ Xeidjtk < 510 min (1)
Institut Superieur de Gestion de Tunis, University of Tunis, 11, Av. De
la liberté, Cité Bouchoucha, 2000, le bardo, Tunis, Tunisia. For each k, ∑ Xeidjtk = 1 (2)
For each j, ∑ Xeidjtk = 1 (3)
For each i, ∑ Xeidjtk = 1 (4)
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 9, SEPTEMBER 2010, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 75
Constraint (1) allows restricting the overall time for the and resource allocation problems. The business experts
daily work for each element of the resource driver. The have shown great interest to our system due to the
constraints (2), (3) and (4) have been introduced to simplicity of generating appropriate solutions to a very
eliminate the risk of conflicts arising in resource complex problem.
allocation. For the constraints (2), (3) and (4), Xeidjtk The visualization of results is an important factor for
variables are assigned the value 1 once they have a value users in the evaluation of a software system. Because our
great than 0; this operation gives us the possibility to system is aimed at a non-computer specialist users,
apply the SUM function for exclusivity modeling. This accustomed to the use of highly sophisticated business
assignment of the value 1 to Xeidjtk is limited to these applications in terms of graphic interactivity, we
three constraints. proceeded with the development of a graphical interface
hosted on a web server which, treats and displays the
results of our system. For the development of our
2 PROBLEM RESOLUTION graphical interface component, we used Java [3] and Java
Server Faces [4] in order to deploy our system on a web
Our paper presents a real application of using the CSP portal using J2EE [5] high capabilities in this context. Figs.
formalism and demonstrates the high level of adaptation 2 and 3 are examples of implementation of our system.
of constraint programming to provide acceptable Fig. 2 presents the configuration and the start solving
solutions for difficult computer science problems. At the interface of the CSP, while Fig. 3 shows the interface
previous section, the problem of tour services scheduling which displays the solution as a table of generated tour
has been modeled thus we defined clearly all the services.
variables and constraints according to the CSP formalism.
The strength of the constraint programming is in the fact Complexity Analysis
that the user defines the problem and the computer solves CSP generally have exponential complexity since a prob-
it. In order to resolve our CSP, we decided to implement a lem with n variables generates 2n possible combina-
system that encapsulates two CSP solving algorithms: tions[6]. As the computing units capacity is not sufficient
Generate and Test [2] and Backtracking [2]. We used the to examine the validity of all the combinations in a rea-
Java language within Sun NetBeans environment for the sonable period of time. Solving CSP by complete methods
development of our system.
creates a combinatorial explosion of the computation
Then we developed a component for presenting data to
process and does not lead to solutions especially when
users via a graphical user interface which translates the
dealing with problems with many variables. To address
results of running programs of resolution of the CSP
saved as flat files, for purposes of simplification. this problem, we use heuristics and intelligent guiding of
Running the CSP resolution programs generates a the research process. Regarding our case, we have not
consistent assignment of all variables Xeidjtk, which encountered the problem of combinatorial explosion be-
represents the solution of the problem. Once generated, cause we are dealing with a reduced number of variables,
this tour services scheduling will be analyzed and but it is to note that complexity is always calculated at
recorded in a flat file where each line describes a tour worst case. We note that the backtracking algorithm takes
service or a section. In Fig. 1, we present a sample of a much less time to generate solution than the generate and
results file generated through our system. test algorithm which runs through all possible combina-
tions.
68,Eté,D,2009,515,1,14:10,22:20,st,st,14:30,22:15
69,Eté,D,2009,515,3,13:00,21:15,st,st,13:15,21:10
70,Eté,D,2009,42,3,5:20,20:10,st,st,5:30,20:00
71,Eté,D,2009,42,4,5:50,20:40,st,st,6:00,20:30
72,Eté,D,2009,42,5,6:20,21:10,st,st,6:30,21:00
73,Eté,D,2009,42,1,2:20,11:50,st,st,2:30,11:40
74,Eté,D,2009,42,1,14:20,21:50,st,st,14:30,21:40
75,Eté,D,2009,42,6,7:20,22:20,st,st,7:30,22:10
76,Eté,D,2009,42,2,3:20,5:50,st,st,3:30,5:40
77,Eté,D,2009,42,2,21:00,1:40,st,st,21:10,1:30
78,Eté,D,2009,42,1,22:00,0:40,st,st,22:15,0:30
79,Eté,D,2009,16B,1,5:00,20:50,st,st,5:10,20:40
80,Eté,D,2009,42B,1,4:40,13:10,si,si,5:10,12:40

Fig. 1. A sample of the tour services scheduling flat file

Every record shown in Figure 1 shows the details of a


generated tour service, it contains: Equipemnt Number,
Driver Number, Concerned Department Number, Start time, Fig. 2. Graphic user interface: Setting up parameters and solving
CSP
End time, concerned line number, Date... Thus, we see
clearly the interestingness of the CSP formalism in
modeling and optimization of the resolution of transport
JOURNAL OF COMPUTING, VOLUME 2, ISSUE 9, SEPTEMBER 2010, ISSN 2151-9617
HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/
WWW.JOURNALOFCOMPUTING.ORG 76

Authors:

Ourida Saidi Ben Boubaker, Ph. D in


Computer Science from the Paul Sabatier
University of Toulouse, France. She is As-
sociate professor and head of the Depart-
ment of Computer Science in the ISG of
Tunis, University of Tunis. She is a member
of SOIE research Laboratory. Her research
interests are Artificial Intelligence, Informa
tion Systems Security and Data mining.

Amine Farhat, Master of Science in Computer


Science, is a researcher and Ph.D. Student at
the BESTMOD Laboratory, University of Tu-
nis. He is assistant professor in Computer
Science in the ISG of Tunis. He is also a se-
nior computer science analyst Engineer, Head
of Software Development and Decision Sup-
port Systems Projects, in a public Office. His
research Interests are mainly Knowledge Dis-
Fig. 3. Graphic user interface: Deploying results of CSP resolving
covery in Databases, Data mining algorithms,
algorithms
Artificial Intelligence and Data warehousing.

3 CONCLUSION
In this paper, we reported our work in the context of re-
sources scheduling in a transport problem. We have
modeled our problem as a Constraint Satisfaction Prob-
lem. Then, we have implemented two algorithms for ge-
nerating solutions: Test & generate and backtracking.
Real experiments of our system have generated adequate
solutions with high accuracy and without any conflicts
between resources allocations. For presentation reasons,
we have developed a web-based component oriented to
end users and domain experts in order to setting up pa-
rameters, executing algorithms and deploying results. In
order to improve our work, we are actually working on
dealing with more complex problems in terms of va-
riables and constraints. We plan also to apply heuristics
and incomplete methods for resolving the transport sche-
duling problem.

REFERENCES
[1] V., Kumar, “Algorithms for constraint-satisfaction problems: A
survey”, AI Magazine, Vol. 13, pp. 32-44, 1992.
[2] F., Benhamou, N., Jussien, B., O'Sullivan, “Trends in Constraint
Programming”, ISTE Publishing Company, 2007.
[3] B., Spell, “Pro Java Programming", Second edition, APress edi-
tion, 2005.
[4] B. Dudney, J., Lehr, B., Willis, L. Mattingly, "Mastering Java
Server Faces", Wiley Press Edition, 2004.
[5] E., Armstrong, J., Ball, S., Bodoff, D. B., Carson, I., Evans, D.,
Green, K., Haase, E., Jendrock, "The J2EE Tutorial", Sun Micro-
systems, 2005
[6] O., Goldreich, “Computational Complexity", Cambridge Uni-
versity Press, 2008.

You might also like