Proactive Monitoring of Moving Objects

Fábio da Costa Albuquerque

1,3

, Ivanildo Barbosa

1,2

, Marco Antonio Casanova

1,3

Marcelo Tílio Monteiro de Carvalho

and Jose Antonio Macedo

Departament of Informatics, PUC-Rio, Rio de Janeiro, Brazil

Department of Surveying Engineering, Military Institute of Engineering, Rio de Janeiro, Brazil

TecGraf, PUC-Rio, Rio de Janeiro, Brazil

Department of Computing, University of Ceará, Fortaleza, Brazil

Keywords: Moving Objects, Trajectory Analysis, Real-time Monitoring Systems, Web-based Applications.

Abstract: Positioning systems, combined with inexpensive communication technologies, open interesting possibilities

to implement real-time applications that monitor moving objects and that support decision making. This

paper first discusses basic requirements for proactive real-time monitoring applications. Then, it proposes an

architecture to deploy applications that monitor moving objects, are pro-active, explore trajectory semantics

and are sensitive to environment dynamics. The central argument is that proactive monitoring based on

process models, such as workflows, is a promising strategy to enhance applications that control moving

objects. Finally, to validate the proposed architecture, the paper presents a prototype application to monitor

a fleet of trucks. The application uses workflows to model truck trips and features a module to extract data

from the Web which helps detect changes on road conditions.

1 INTRODUCTION

Positioning systems, combined with inexpensive

communication technologies, open interesting

possibilities to implement real-time applications that

monitor moving objects and that support decision

making. An example would be an application to

monitor a fleet of tank trucks that distribute fuel to

gas stations in an urban environment. Every trip is

carefully planned to follow pre-defined routes,

avoiding sensitive areas (such as school areas) and

periods of the day or routes where the transportation

of dangerous cargo is banned and to pro-actively re-

route the truck in case of traffic accidents and other

events that might cause delays.

We may classify such applications according to

different perspectives. The application may use

trajectory semantics, such as stopping at a point of

interest, or the application may use just raw

trajectory data, such as speed and direction. We cite

Alvares (2011), Siqueira and Bogorny (2011) and

Moreno, Times, Renso and Bogorny (2010) and as

related works in trajectory semantics.

A reactive application uses just the past behavior

of the objects, as opposed to a proactive application

that features models of the predicted (future)

behavior of the objects and perhaps suggests

alternative actions. Proactive computing is

investigated in Tennenhouse (2000), which

advocates a paradigm shift from human-centered to

human-supervised computation. In his perspective, a

system to be proactive must: (1) have a direct

connection with the real world; (2) be able to

execute actions in response to external stimuli; (3)

execute actions faster than the human response. In

other words, a system with proactive behavior must

detect interesting situations before they happen and

must be able to handle such situations without

human supervision.

Finally, the application may be sensitive to

environment dynamics, meaning that it monitors the

current state of the environment (or even estimates

future states of the environment) where the object is

moving to base its decisions. Environmental facts

are considered when they directly affect the moving

object behavior. By contrast, the application may be

insensitive to environment dynamics, in the sense

that it has just a static model of the environment

(such as a road map) where the object is moving.

In this paper, we first discuss basic requirements

for proactive monitoring applications. Then, we

propose an architecture for applications that monitor

moving objects, are pro-active, explore trajectory

191

da Costa Albuquerque F., Barbosa I., Casanova M., Monteiro de Carvalho M. and Macedo J..

Proactive Monitoring of Moving Objects.

DOI: 10.5220/0003988601910194

In Proceedings of the 14th International Conference on Enterprise Information Systems (ICEIS-2012), pages 191-194

ISBN: 978-989-8565-10-5

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

semantics and are sensitive to environment

dynamics.

To achieve proactive behavior, the proposed

architecture includes models of the processes behind

the moving objects. The prototype application uses

workflows to model truck trips. To monitor moving

objects, the architecture includes support for real-

time trajectory data stream processing. Finally, to

account for trajectory semantics and support

sensitivity to environment dynamics, the architecture

features additional data sources, classified as

(geospatial) static structured data sources (SSD

sources) and dynamic structured data sources (DSD

sources). The prototype application uses geospatial

databases and georeferenced facts posted in feeds

and tweets about the road conditions that may affect

the predicted behavior of the trucks.

The contributions of the paper are therefore

threefold: a discussion of the basic requirements for

proactive monitoring applications; a proposal for an

architecture for such applications; and a prototype

application to assess the proposed architecture. The

central argument is that proactive monitoring based

on process models, such as workflows, is a

promising strategy to enhance applications that

control moving objects.

The rest of the paper is organized as follows.

Section 2 describes a motivating scenario. Section 3

discusses basic requirements for proactive

monitoring. Section 4 introduces an architecture for

proactive monitoring applications. Section 5

presents a prototype application to validate the ideas.

Section 6 discusses related work. Finally, Section 7

contains the conclusions.

2 A MOTIVATING APPLICATION

Consider an application to monitor a fleet of

delivery trucks, abstractly defined as follows.

Each truck is modeled as a moving object M and

each trip is described as a workflow W

that defines

the customers to be serviced in the trip and the

routes to be followed. Each step p of W

either

represents delivering merchandize at a customer C

located at place L

, or moving from a place O

called the origin of p, to a place D

, called the

destination of p, through a route R

For each moving object M, the system receives a

data stream containing the date, time, geographic

position and speed. The system transforms this raw

data into meaningful events with the help of a

geospatial database storing the location of points-of-

interest.

The application monitors several trucks, sharing

the same underlying road network and the same

emergency workflows. A centralized application is

desired to integrate the monitoring of the individual

trucks, as well as of the events that affect the road

network where the trucks move. The application also

reduces human interference on the monitoring

process to minimize failures due to fatigue.

Consider now the problem of improving the

truck monitoring application to become proactive

and sensitive to the environment.

Briefly, the first change in the application design

is to use the truck delivery workflows to infer their

future behavior. The second change is to detect

anomalies in the conditions of the roads where the

trucks are expected to drive in the next steps of their

trips (defined by their workflows). As an example,

the system may issue an alert to the driver to

proceed more carefully (or even to take an alternate

route) when detected that a vehicle, carrying a

flammable load, is driving along a road with wet

floor ahead.

Finally, we note that we may describe similar

scenarios related to other classes of moving vehicles,

such as planes and ships. Workflows in this case will

be abstractions for flight or sailing plans.

3 PROPOSED ARCHITECTURE

Figure 1 illustrates the proposed architecture. The

Proative Central Monitor (PCM) is the core

component that, as the name implies, coordinates the

other components to pro-actively monitor moving

objects. The Planning Manager (PM) stores and

controls the workflows that model the behavior of

the moving objects. The Application Databases

contain auxiliary data such as names and addresses

of customers, the road network, etc. The Moving

Objects Monitor (MOM) sends to the PCM the

structured data stream containing information

relative to the real-time monitoring of moving

objects: position, trajectory semantic data (i.e.,

interpreted trajectory data) and other signals from

moving objects. The Mediators facilitate access to

either dynamic or static external data sources.

4 A PROTOTYPE APPLICATION

This section outlines some of the features of a

prototype application to monitor a fleet of delivery

trucks, along the lines of the application presented in

ICEIS2012-14thInternationalConferenceonEnterpriseInformationSystems

192

Section 2. The prototype follows the architecture

proposed in Section 4 and the discussion focuses on

some aspects of the Dynamic Structured Data

Mediator and the Proactive Central Monitor.

Figure 1: General view of architecture proposal.

4.1 Dynamic Structured Data Mediator

Proactivity is two-fold: situations may be detected

from past behavior of the object or from external

agents that affect the application.

Santos and Moreira (2010) propose an input for

proactive computing by predicting the next step of

moving objects based in its current location and road

data. Previous moving object data is not used. The

success of prediction may vary according to the

scenario and variables.

The second approach to proactivity is based on

the extraction of relevant facts that potentially affect

the future behavior of moving objects.

The prototype implementation of the Dynamic

Structured Data Mediator (DSDM) uses Twitter as

the main dynamic structured data source. Similar

applications were deployed by Carvalho, Sarmento

and Rossetti (2010) and MacEachren et al. (2011).

The prototype considers tweets from a predefined

list of institutions, assessed as trustworthy sources,

as well as from users related to the primary sources

(e.g. followers).

The implementation follows the second strategy

listed in Section 4.2, that is, the DSDM is

responsible for post-processing the results returned

by the wrappers. As illustrated in Figure 2, the

DSDM receives raw data containing text body,

source, user, location (when available), number of

re-tweets, hashtags and time stamp. It then filters

tweets according to their creation date and keeps

only the most recent ones. At the classification step,

the DSDM selects only the text body and the source.

It classifies tweets according to the occurrence of

relevant facts in the text body (e.g. car crashes,

floods and road blocks). After filtering the relevant

tweets, the DSDM extracts the spatial reference for

the reported fact, with the help of a street gazetteer

stored in the SSDM. Finally, the DSDM transforms

the extracted data into a predefined structure before

sending the data to the PCM.

4.2 Proactive Central Monitor

The prototype implementation of the Proactive

Central Monitor (PCM) processes facts and events it

receives from the DSDM and the MOM as follows.

For each moving object M, with workflow W

the PCM uses the events the MOM sends to monitor

the step c that W

is executing. It then simulates the

steps of W

that may follow c, up to a certain depth,

and collects the routes that M may traverse.

Next, the PCM verifies if such routes are

affected by a fact that the DSDM has already sent. If

this is the case, the PCM warns the (human)

controller or the driver, or both, that future steps

planned for M may have to be changed or aborted.

For simple facts, the PCM just generates

warnings both to the controller and the driver, but it

does not recommend that W

be necessarily

changed. For example, a fact reporting heavy traffic

in a route generates just a delay warning to the

driver or even suggests an alternative route.

However, some facts may imply restrictions to

traffic, even if temporarily. In this case, the PCM

recommends to the controller that W

be changed or

aborted. The controller then invokes the route

planning component (outside the scope of this paper)

to create a new version of W

The route planning component is prepared to create

routes that consider a list of traffic restrictions

(usually maximum load and maximum height

permitted, forbidden cargo traffic hours, etc…).

Finally, the PCM may also receive events from

the MOM that represent incidents involving M (e.g.

a mechanical problem with M). It then invokes

workflows, stored in the TPL, to mitigate the

incident and eventual damages to the environment

(e.g. to clean up an oil spill).

ProactiveMonitoringofMovingObjects

193

Figure 2: Data flow of the DSDM.

5 CONCLUSIONS

In this paper, we first discussed basic requirements

to achieve proactive monitoring of moving objects.

Then, we proposed an architecture that meets the

requirements. The first key point of the discussion is

to model the process behind a moving object as a

workflow to be able to infer future actions. The

second key point is to monitor or even to predict

changes in the environment by exploring dynamic

data sources.

Finally, we outlined some of the features of a

prototype application to monitor a fleet of delivery

trucks. In particular, the prototype uses Twitter as a

viable dynamic data source to detect changes in the

current road conditions, as well as to register future,

planned changes that may affect the traffic in certain

roads.

We plan to improve the prototype application in

several directions. In particular, we intend to explore

a supervised strategy to address the problem of

classifying facts extracted from tweets. We also plan

to explore RSS feeds as a dynamic data source

(Chen et al, 2007) and to automatically analyze Web

site containing news and weather reports as a viable

source of dynamic information.

REFERENCES

Alvares, L. O., Loy, A. M., Renso, C., and Bogorny, V.,

2011. An algorithm to identify avoidance behavior in

moving object trajectories. In: Journal of the Brazilian

Computer Society 11.

Carvalho, S., Sarmento, L. and Rossetti, R., (2010). Real-

Time Sensing of Traffic Information in Twitter

Messages. In: ATSS @ IEEE ITSC 2010 Proceedings

of 4th Workshop on Artificial Transportation Systems

and Simulation, Madeira, Portugal

Chen, Y. F., Di Fabbrizio, G., Gibbon, D., Jora, S.,

Renger, B., Wei, B., 2007. Geotracker: geospatial and

temporal RSS navigation. In 16th International

Conference on World Wide Web. pp. 41-50. Alberta.

MacEachren, A. M., Robinson, A. C., Jaiswal, A.,

Pezanowski, S., Savelyev, A., Blanford, J., Mitra, P.,

2011. Geo-Twitter analytics: applications in crisis

management, In 25th ICC, International Cartographic

Conference [available online]

Moreno, B., Times, V. C., Renso, C., and Bogorny, V.,

2010. Looking inside the stops of trajectories of

moving objects. In XI Brazilian Symposium on

Geoinformatics, pp. 9–20, Campos do Jordão.

Santos, M. Y., and Moreira, A., (2010). GUESS: on the

prediction of mobile users’ movement in space, In:

Wachowicz, M. (Ed.) Movement-Aware Applications

for Sustainable Mobility: Technologies and

Approaches, IGI Global Publishing.

Siqueira, F. L. and Bogorny, V., 2011. Discovering

chasing behavior in moving object trajectories. In

Transactions in GIS, 15(5).

Tennenhouse, D., 2000. Proactive computing. In: Comm.

ACM, vol. 43, May. 2000, pp. 43–50.

ICEIS2012-14thInternationalConferenceonEnterpriseInformationSystems

194