velopment of internet
ference.
of geography markup
ics (SVG), and Web
| the development of
1s (GIS). Journal of
[ 2008. http://www
(Mar. 2010).
ime, S., 2006. UMN-
roperable, and Open
Analysis System. In:
ger, pp. 400-417.
of GIS (in Chinese),
, Beijing.
N CASE STUDIES
900) as buf from river
with this buffer :
as id, boundary.d as d
ary.d, bf.buf) = true)
district:
st.id as id, intersection
ich district:
'ea
| district:
id as id, intersection
dary, railway
'gment:
(int) as len from seg T
ach district:
(len) as total len from
vays:
re total len>0
Suang-Mei-Shan”:
| 20000) as buf from
» located in this buffer:
n(city.d, buf. buf)=true)
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B4, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
A HYBRID PULL-PUSH SYSTEM FOR
NEAR REAL-TIME NOTIFICATIONS ON SENSOR WEB
C.Y. Huang, S. Liang *
Dept. of Geomatic Engineering, University of Calgary, 2500 University Drive NW, Calgary, Alberta, T2N 1N4 Canada
- (huangcy, steve.liang) @ucalgary.ca
Commission IV, WG IV/2
KEY WORDS: GIS, Sensor, Query, Real-time, Spatial, Temporal, Web based, Application
ABSTRACT:
World-wide sensor web generates tremendous amount of sensor data stream allowing people to observe events that were previously
unobservable. Sensor web has been wildly applied in many monitoring systems; some of them are extremely time-sensitive, e.g.,
disaster management systems. However, with the growing amount of sensor data, the traditional request/response communication
model becomes inefficient as it is based on point-to-point pulling interactions between users and data providers. In order to address
this issue, publish/subscribe communication model has been proposed and applied in many applications, e.g., web blogging. The
publish/subscribe model utilizes an intermediary broker on matching predefined queries with the data pushed to the broker.
However, we argue that the publish/subscribe model is hard to be directly applied to sensor web due to the fact that most sensor web
services are based on pulling interaction model only. For instance, more and more sensor data providers are publishing their sensor
data with the Open Geospatial Consortium (OGC) Sensor Observation Service (SOS) standards, and the OGC SOS services are
based on the request/response model. Therefore, in order to address this issue, we propose a hybrid pull-push system to retrieve
sensor web data in a timely manner. The preliminary experimental results indicate that the proposed system is able to fetch near real
time sensor streams from pull-based sensor web services.
1. INTRODUCTION
1.1 Background
The World-Wide Sensor Web (Liang er al., 2005) is generating
tremendous volumes of real-time sensor data streams ranging
from video camera networks monitoring real-time traffic to
matchbox-sized wireless sensor networks embedded in the
environment to monitor habitats. As these data streams enable
scientists to observe phenomena that are previously
unobservable, the World-Wide Sensor Web is increasingly
attracting interests for a wide range of applications, including:
habitats monitoring systems (Mainwaring et al. 2002),
environment observation systems (Hart and Martinez 2006),
structure health monitoring systems (Hsieh 2002), health
applications (Xu, 2002), fire emergency response systems
(Kassab et al., 2010), etc. Among these applications, many of
them are time-sensitive and require prompt notifications.
However, with the vast amount of sensor data in sensor web, the
traditional request/response communication model becomes
inefficient as it is based on point-to-point pulling interaction
between users and data providers. In order to address this issue,
publish/subscribe communication model (Birman and Joseph,
1987) provides an intermediary broker for users to register
queries and for providers to push new data to. The broker sends
notifications to users as new data meet their query criteria.
While the publish/subscribe model has been widely applied in
other disciplines, e.g., web blogging, this model is relatively
New in the sensor web field. We argue that a major reason of
this slow adoption is that the most current sensor web services
are currently based on pulling model only. Even though sensor
P mar
s Corresponding author.
data streams are pushed to the data repository of web services,
users need to pull the sensor data from the sensor web services
proactively. For example, Open Geospatial Consortium (OGC)
Sensor Observation Service (SOS) (OGC 2007), as one of the
most popular open sensor web standards, defines a standard
protocol for users to retrieve sensor metadata and observations
through Internet. In general, this issue impedes users from
getting real-time notifications about events happening on sensor
web.
In order to address this issue and achieve the goal of timely
notification, this paper proposes a hybrid pull-push system for
near real-time sensor data notification. This system contains
three major components, namely (1) query aggregator, (2)
adaptive feeder, and (3) sensor data cache. Users can first
register queries (i.e., subscriptions) to the system. Before
sending requests to sensor web services to pull sensor data, the
query aggregator aggregates queries in order to avoid redundant
requests. Then based on the aggregated requests, the adaptive
feeder pulls sensor data from sensor web services in a timely
manner. Finally, after receiving responses from services, the
input adaptor preserves sensor data in the sensor data cache
according to users’ query criteria.
In this paper, we use OGC SOS as the sensor data sources. SOS
version 1.0 has been published in 2007 and SOS version 2.0 has
been approved in March 2012. SOS is suitable for our
experiment because SOS is already adopted by many sensor
data providers and current SOS implementations have the
aforementioned challenges.
To sum up, the major objective in this paper is to build a system
allowing users to subscribe sensor data by setting spatio-
