Title: Transfer Graph Approach for Multimodal Transport Problems
1Transfer Graph Approach for MultimodalTransport
Problems
MCO'08
- Hedi Ayed,
- Djamel Khadraoui,
- Zineb Habbas,
- Pascal Bouvry,
-
- Jean Francois Merche
2Plan
MCO'08
2- Problem description and our objectives
3- Existing solutions
4- Our solution Transfer Graph
5- Conclusion
3Introduction
MCO'08
- -What is multimodal transport problem ?
-What this paper represents ?
-What is Carlink ?
-Route guidance in Carlink ?
4What is multimodal transport problem ?
MCO'08
- Monomodal one transport mode.
Multimodal many transport modes.
5What this paper represents ?
MCO'08
- When investigating existing approaches and
algorithms on the topic, we observe that none of
them is applicable to multi modal route guidance
problems subject to the following constraints - i) The multimodal network is assumed to be flat.
- ii) Involved unimodal networks may be kept
separated and accessed separately. - iii) If there are multiple network information
sources within a single mode, they may be kept
and accessed separately.
This paper presents our contribution to
multimodal route guidance problem.
6What is Carlink ?
MCO'08
The present work has been done in the context of
Carlink
- Carlink is European project aiming to develop an
intelligent wireless traffic service platform
between cars supported by wireless transceivers
beside the road(s)..
-real-time local weather data -the urban
transport traffic management -the urban
information broadcasting for the mobile users
7The Objective of Carlink
MCO'08
Objective of Carlink is the travelling user
mobility management. The main idea behind this
objective is to provide route guidance services
to a given traveling mobile equipped user. The
primary challenge of this work has been to set
solid and new basis to address multimodal route
advisory problem under Carlinks route guidance
requirements.
8Route guidance in Carlink
MCO'08
What is the traveling mobile user ?
traveling mobile user any entity equipped
with a mobile technology device which is
planning to move or is already moving from one
geographical location to another
What route calculation mean ?
The scenario referred as route planning
The user is preparing a travel, route calculation
service consist to propose a set of possible
route according to a user defined source and
target location.
The scenario referred as travel monitoring
The user is already on his way to the
destination, route calculation service consist
to propose a set of alternative routes from
users current location to a new destination or
to the current destination, in case the user has
revised his choices or when some traffic
disturbance.
9Problem description and our objectives
MCO'08
- Shortest path problem In graph theory, the
shortest path problem is the problem of finding a
path between two nodes such that the sum of the
weights of its edges is minimized.
- Multi objective and time dependent In many
works on SPP, the path cost is a single scalar
function. However, in the multimodal transport
problem we need to optimize path according to
more than one scalar functions. In this case the
problem becomes a multi objective optimization
problem.
-Our objectives
10Main existing solutions
MCO'08
- - Multigraph based approach
The multimodal transportation network is viewed
and treated as a multigraph, In general, a
multigraph is simply a graph in which it is
allowed to have more than one arcs between two
nodes. So the problem is reduced to the classical
shortest path problem or to one of its variants.
- CSP based approach
Most works consider multimodal route planning
from graph theory perspective. But the problem
can also be seen as a constraint satisfaction
problem (CSP). So the problem is viewed as
finding a combination of variables value in a
search space which satisfies a set of
constraints.
11Our solution Transfer Graph
MCO'08
- The problem keep all existing unimodal
transportation networks separated. - The Solution we introduce an unusual graph
structure which we call transfer graph.
Let G(N, A) denotes a graph
Gs G1, G2, ... ,Gq a set of sub graphs
Each Gg (Ng, Ag) is called a component
Given two distinct component s Gg , Gg
Ng n Ng f is not mandatory.
Ag n Ag f is always hold.
i in Ng n Ng is a transfer point
We denote a transfer graph by TG(N, A, GS, TS)
12Our solution Transfer Graph
MCO'08
- - Shortest path algorithm in transfer graph
Consider a transfer graph TG(N, A, GS, TS), let
s, t be an origin-destination pair and Gg be a
component of TG
- inter components paths
- intra components paths
- full paths
- partial paths
13Our solution Transfer Graph
MCO'08
- Full paths
- Relevant Head paths
-Relevant intermediate paths
- Relevant tail path
Assume that for all components Gg in GS we have
computed the following relevant path sets
Pg s.t the set of best intra component full
path within Gg
Pg s.- the set of all best intra component head
paths from s within Gg
Pg .t the set of all best intra component
tail paths to t within Gg
Pg .- the set of all best intra component
intermediate paths within Gg
14Our solution Transfer Graph
MCO'08
The relevant graph (RG)
RG (RV, RE)
RV (URVg Gg in GS) Us, t
RE (UREg Gg in GS)
15Implementations and results
MCO'08
First implementation a basic algorithm
The idea
-Get the request of the user
-Compute all the best paths
-Build the relevant graph
-Answer the request user
Disadvantages
-Very slow
-Many reputations
16Implementations and results
MCO'08
Second implementation an algorithm with
database
The idea
-Compute all the best paths for all pairs of
nodes
-Store the best paths in a database
-Get the request of the user
-Build the relevant graph
-Answer the request user
17Conclusion
MCO'08
Summary
This work has been done in context of Carlink
We presented an algorithm to solve the shortest
path in multimodal network
Support multi objective and time dependent
Future works
New decomposition geographic decomposition
18MCO'08