SensorGIS

An Integrated Architecture for Information Systems based on Sensor Networks

Jianzhao Huang, Nicholas M. Boers, Eleni Stroulia, Pawel Gburzynski and Ioanis Nikolaidis

Department of Computing Science, University of Alberta, Edmonton, Alberta, Canada

Keywords:

Wireless sensor networks, Geographic information systems, Forums, Visualization.

Abstract:

In this paper, we describe SensorGIS, an integrated architecture for WSN applications. SensorGIS provides

an integrated service-oriented architecture for collecting, archiving, analyzing, and visualizing sensor network

data in a geographic information system (GIS). By using an extendible GIS framework as one of its user

views, SensorGIS can contextually communicate the collected data, its trends, and distinct values of interest.

In addition, it is designed in the service-oriented style and hence is extendible in terms of the analyses and

visualizations. Finally, it integrates an online collaborative forum that enables annotation of the collected data

with the users’ observations and interpretations.

1 INTRODUCTION

We observe that for sensors, to be systematically and

cost-effectively deployed in applications, it is essen-

tial to develop software architectures that treat them,

their data, and their potential services in an integrated

manner. For SensorGIS, we adopt a web-based archi-

tecture that relies on the representational state transfer

(REST) style of services to archive, analyze, and vi-

sualize sensor-network data. At the user-interaction

layer, it integrates (a) a geographic information sys-

tem (GIS) for contextually communicating sensed

data, (b) a visual data analysis component for explor-

ing interesting trends and events in the collected data,

and (c) a collaboration forum for enabling users to

share their observations and interpretations.

The fundamental innovation of SensorGIS is

twofold. First, it relies on an XML data representa-

tion and a REST-style architecture that integrates ser-

vices for data management, analysis, and visualiza-

tion. Second, it integrates three different components

to communicate data to end users: (a) a geographic

model, (b) data-speciﬁc visualizations, and (c) a col-

laborative tool for associating data with user interpre-

tations and analyses.

The data-visualization tools of SensorGIS support

charts and interactive tables that enable users to see an

overviewof observations of interest. Users can access

historical data and statistics, and they can also reg-

ister to receive notiﬁcations when values cross user-

deﬁned thresholds. Finally, users can record their ob-

servations and interpretations of the data in a forum

so that a whole community can collaborate.

2 RELATED WORK

WSNs have applications in a variety of areas, such as

environmental monitoring and healthcare. This tech-

nology also brings forth challenges in visualizing and

interpreting information, which have given rise to a

variety of sensor-network visualization systems.

SensorMap (Nath et al., 2007)is a real-time WSN

visualization web service developed by Microsoft. Its

disadvantages include its limited (or missing) sup-

port for (a) alerting users when measured values cross

user-deﬁned thresholds, (b) aggregating data retrieved

from multiple sensors, (c) analyzing data within the

web-based environment, and (d) facilitating the an-

notation of observations or trends deduced from the

data.

Viewlon (Furuyama et al., 2007), unlike Sen-

sorMap, is not a web service and displays sensor re-

lationships such as masters and sinks alongside the

correlation of their sensed data. It visualizes WSNs

as graphs, which makes it difﬁcult to visualize sen-

sors geographically, and not all WSNs form explicit

and stable graphs (Gburzynski and Olesinski, 2008).

MoteView (Turon, 2005), similar to Viewlon, is

not a web service, and its graphical user interface

(GUI) runs on the client side. Similar to Viewlon, it

199

Huang J., M. Boers N., Stroulia E., Gburzynski P. and Nikolaidis I.

SensorGIS - An Integrated Architecture for Information Systems based on Sensor Networks.

DOI: 10.5220/0002798301990202

In Proceedings of the 6th International Conference on Web Information Systems and Technology (WEBIST 2010), page

ISBN: 978-989-674-025-2

represents sensor distribution as a graph and depends

on speciﬁc sensor-network hardware.

Other tools include SNAMP (Yang et al., 2006),

SpyGlass (Buschmann et al., 2005), GIS for ground-

water surveillance (Lawerence et al., 2003), and

SeeMote (Selavo et al., 2006). SNAMP is a desk-

top application without a GIS, and SpyGlass does

not have an extendible database schema for data rep-

resentation. The GIS for groundwater surveillance

is designed to ﬁt a particular WSN and, similar to

SeeMote, visualizes the WSN as a graph.

3 THE ARCHITECTURE

SensorGIS follows four software-design principles:

1. separation of data representation from user-

centric data renderings,

2. adoption of XML as the data-exchange format be-

tween data resources and the user-centric data ren-

derings,

3. independence of database schema design from

the sensor-network deployment and conﬁgura-

tion, and

4. use of an efﬁcient API to extract relevant data

views from the data resource.

It consists of several key components. End-users

access the user interface component that contains

three parts: the multi-layered map, the data-analysis

panel, and the collaboration forum. The second com-

ponent, the data layer, resides in the background and

stores the data received from the network and serves

as the source for visualizations and analyses. The ﬁ-

nal AJAX component glues the above two components

and uses representational state transfer (REST) and an

appropriate XML schema to integrate the user inter-

face with the data layer.

3.1 The User Interface

One of the fundamental innovations of SensorGIS is

its design that integrates the collecting, archiving, an-

alyzing, and visualizing of sensor-network data. The

web-based user interface allows access to the Sensor-

GIS services from multiple locations using different

types of devices and integrates multiple synchronized

views. In the following subsections, we discuss the

three key parts of the SensorGIS web interface and

explain their features.

3.1.1 The Multi-layered Map

SensorGIS builds on the OpenLayers JavaScript li-

brary for its multi-layered map. The base map can be

a user-supplied image or a map from Google, Yahoo,

or NASA. The library supports importing geographic

data from GeoServer, which is a spatial database de-

signed for geographic data storage and retrieval, and

can display the data as overlays. It also supports a

variety of other data sources.

The SensorGIS map shows sensor nodes as mark-

ers pinned at the node locations. If the network con-

tains mobile nodes, the map view may periodically

refresh to reﬂect changing locations.

SensorGIS can be used to monitor both individual

sensor states and sensor group states. When the user

selects a node marker, a pop-up displays the most re-

cent state values. Users can deﬁne sensor groups by

selecting (a) all sensors within a geographic area of

interest or (b) individual node markers.

In addition to sensor and sensor-group queries,

SensorGIS also supports a “watch” functionality to

monitor nodes for extreme values. When enabled,

the map shows sensors with state values outside user-

deﬁned ranges.

The SensorGIS map also allows users to manage

the topology of the deployed WSN. Users can create

new markers on the map for newly deployed nodes

and manually change the location of current sensors.

3.1.2 The Data-analysis Panel

The top-right corner of the interface embeds the data-

analysis panel (Figure 1, top-right). The content of

this panel changes according to user selections made

in the multi-layered map. If the user selects a single

node, it shows either (a) the most recent observations

of the node or (b) its historicaldata as a scalable graph

(we use the ﬂotr JavaScript plotting library).

The data-analysis panel also supports a number

of predeﬁned queries for groups of nodes. The ini-

tial view contains the latest readings of each sensor

at each node. A second view shows summary statis-

tics for a single sensor type for a given period at each

node. Finally, the third view extends the second to

multiple sensor types, and furthermore, it summarizes

them for all nodes in the group. To relate sensor

groups with their geographic location, the map view

labels the sensor group markers with numbers corre-

sponding to table rows in the data-analysis panel.

Besides history graphs and grouping tables, users

can also see any enabled watch function’s status in the

panel.

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

200

Figure 1: The SensorGIS interface with the multi-layered

map in the top-left, the data-analysis panel in the top-right,

and forum in the bottom of the window.

3.1.3 The Collaboration Forum

The integration of a forum (phpBB) is a powerful fea-

ture of SensorGIS (Figure 1, bottom). Unlike typical

forums that simply sequentially sort postings, our in-

tegration associates postings with the source sensor.

When a user observes something within the WSN

data, he can add a topic to the forum. The forum entry

and the selected sensor(s) are then cross-referenced in

the server. Consequently, when another user exam-

ines the particular sensor(s), the software highlights

the appropriate forum entry. Conversely, users read-

ing a particular entry in the forum see the correspond-

ing sensor(s) highlighted on the map.

A type of summary view is also available to vi-

sualize the amount of forum attention given to each

sensor. After enabling this overlay, coloured markers

on the map appear next to each annotated sensor, and

the marker’s colour indicates the number of related

forum entries.

3.2 The Data Layer

To support the described user interface, the collected

sensor data is stored in (and retrieved from) a MySQL

database. Figure 2 shows its entity/relationship dia-

gram. It contains ﬁve entities:

NODE, with attribute Node ID, lists the wireless

nodes in the network.

LOCATION, with attributes for the longitude, lati-

tude, altitude, and a time-stamp, associates wire-

less nodes with their physical locations. We in-

clude time-stamps to support mobile nodes.

STATE lists all of the possible states that a sensor

may observe.

OBSERVATION contains the actual state measure-

ments. The Is User attribute indicates whether

a SensorGIS user (rather than the network)

recorded the observation.

NODE

has

ROLE

Type_ID Type_Desc

observes

STATE

Tstamp

Value

Is_User

Range_Min

Range_Max

State_Desc

State_Name

State_ID

Tstamp

Y Z

Node_ID

LOCATION

OBSERVATION

has

generates

observes

Figure 2: The database schema showing the relationship be-

tween tables and attributes.

ROLElists the possible functions a node may take on

in the network (e.g., sensor, sink, or master). At

any given moment, each node is associated with

one of these roles.

3.3 User-interface/Database Integration

To produce the user interface, server-sidePHP scripts:

1. process HTTP POST requests from the client to

extract arguments for database queries,

2. invoke relevant database queries on our MySQL

database, and

3. construct an XML document containing the rele-

vant data in response to the client-issued request.

Before sending the XML document to the client, the

server validates it against the XML schema.

Two distinct styles exist for implementing web-

based service-oriented applications: the ws* style and

the REST style. Given the relative immaturity of

the sensor-network technologies and applications, we

have chosen the simpler REST style, which builds

well on the set of core operations we have discussed

earlier in this section.

The server contains three modules:

1. The network maintenance module reﬂects the

WSN structure, manages all the raw data from the

network, and feeds the network states with read-

ings.

2. The network states module stores all readings re-

ceived from the network maintenance module.

When query requests unrelated to the forum ar-

rive, the network states module generates query

results and sends the results back in XML format.

SensorGIS - An Integrated Architecture for Information Systems based on Sensor Networks

201

3. Finally, the forum module is standalone and it

manages the forum database and provides data for

interaction between the multi-layer map and the

forum, again in XML format.

According to the REST style, the data exchanged

between server and client are represented in XML.

Inspired by SensorML, we develop a simpler three-

part XML schema to represent sensor-network data

and add in it the support of history and statistics. The

three parts are Operation, Sensor, and Statistics.

The Operation element deﬁnes the visualization

operation to be applied to the retrieved XML doc-

ument. In addition to the operation ID, a transac-

tion ID is also included in the element, to prevent the

client from performing redundant or wrong visualiza-

tion operations.

The Sensor element is used to represent all the in-

dividualsensor information, including history. A Sen-

sor tag has seven potential elements: sensorID, loca-

tion X, location Y, location Z, typeID, typeDesc, and

states. The ﬁrst four elements are mandatory and cor-

respond to the database table LOCATION. The fourth

and ﬁfth elements, typeID and typeDesc, are also re-

quired and correspond to the entity ROLE. The last

element, named states, is optional and corresponds to

the entities STATE and OBSERVATION. The states

tag may consist of multiple child tags if it is carrying

history data. According to the database query per-

formed, child tags will be appended adaptively to the

XML.

The third part of the schema is the Statistics ele-

ment, and it is for sensor group data including sum-

mary statistics. It has two child tags: Sensor List and

Observations. Sensor List records all the node IDs,

and Observations has multiple child tags that repre-

sent different group statistics retrieved from the OB-

SERVATION table in the database.

While the Operation element is mandatory, the

Sensor and Statistics elements are optional and mu-

tually exclusive. This exclusiveness results from each

XML response document only containing the result of

a single query. For example, an XML document con-

taining the current state values of a sensor does not

need to include the Statistics tag. Similarly, an XML

document including the statistics of a sensor does not

need the Sensor tag. Moreover, if an error occurs and

no sensor data can be returned, the XML document

will only include the Operation tag.

4 CONCLUSIONS

We described SensorGIS, an extendible service-

oriented architecture for managing and analyzing the

data collected by WSNs. SensorGIS aims to improve

productivity by providing a library of functions that

collect, archive, visualize, and analyze WSN data. We

have incorporated SensorGIS into two signiﬁcantly

different WSNs: the Intelligent Mousetrap and the

Smart Condo, representing respectively an outdoor

and an indoor WSN. In the context of the ﬁrst project,

we augmented traps, used by biologists to live-catch

small animals for study purposes, with switches and

a radio, and we visualized the network and the traps’

states within SensorGIS. Biologists can view the data

in SensorGIS, either in the ﬁeld or via the Internet,

and then focus only on occupied traps. The inter-

ested reader can ﬁnd more information about the latter

project at (Boers et al., 2009). Its integration with the

intelligent mousetrap and Smart Condo project pro-

vides strong support for its applicability to diverse

WSNs.

REFERENCES

Boers, N. M., Chodos, D., Huang, J., Stroulia, E., Gburzyn-

ski, P., and Nikolaidis, I. (2009). The Smart Condo:

Visualizing independent living environments in a vir-

tual world. In PervasiveHealth ’09: Proceedings from

the 3rd International Conference on Pervasive Com-

puting Technologies for Healthcare, London, UK.

Buschmann, C., Pﬁsterer, D., Fischer, S., Fekete, S. P., and

Kr¨oller, A. (2005). SpyGlass: a wireless sensor net-

work visualizer. SIGBED Rev., 2(1):1–6.

Furuyama, M., Mukai, J., and Imai, M. (2007). New

Frontiers in Artiﬁcial Intelligence, volume 4384 of

LNCS, chapter Viewlon: Visualizing Information on

Semantic Sensor Network, pages 65–76. Springer

Berlin/Heidelberg.

Gburzynski, P. and Olesinski, W. (2008). On a practical ap-

proach to low-cost ad hoc wireless networking. JTIT,

2008(1):29–42.

Lawerence, A., Varadhan, A., and Gadiparthi, S. (2003). A

GIS application for remote surveillance of groundwa-

ter contamination. In Proc. 2003 ESRI User Conf.

Nath, S., Liu, J., and Zhao, F. (2007). SensorMap for wide-

area sensor webs. Computer, 40(7):90–93.

Selavo, L., Zhou, G., and Stankovic, J. (2006). SeeMote:

In-situ visualization and logging device for wireless

sensor networks. In Proc. of the 3rd Int. Conf. on

Broadband Comm. Networks and Syst., pages 1–9.

Turon, M. (2005). MOTE-VIEW: a sensor network moni-

toring and management tool. In Proc. 2nd IEEE Work-

shop Embedded Networked Sensors, pages 11–17.

Yang, Y., Xia, P., Huang, L., Zhou, Q., Xu, Y., and Li,

X. (2006). SNAMP: A multi-sniffer and multi-view

visualization platform for wireless sensor networks:

A multi-sniffer and multi-view visualization platform

for wireless sensor networks. In Proc. 1st IEEE Conf.

on Industrial Electron. and Applications, pages 1–4.

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

202