The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008 
multi-agent system is introduced. Each entity can be 
represented as a corresponding agent through multi-agent 
system. Multi-agent system is adopted to model the behaviours 
of each entity, especially the collation and cooperation of all 
relevant entities during disaster management. For capturing and 
representing each entity’s expertise and professional knowledge, 
rule-based expert system is introduced. By using rule-based 
expert system, the expertise and knowledge of each entity can 
be represented as a set of rules and captured into disaster 
management knowledge base. Through the proposed 
architecture, disaster sensing and decision making can be 
integrated closely. Thus, the existing sensor web can be 
potentially extended and greatly enhanced from mere disaster 
sensing and monitoring to a higher level, i.e., knowledge- 
oriented decision making, which will better support disaster 
management and emergency response. 
The rest of this paper is organized as follows. The following 
section presents an overview on current research of sensor web. 
Then, Section 3 discusses multi-agent system, rule-based expert 
system and their potentials to disaster decision support. In 
Section 4, a conceptual framework of knowledge-oriented 
sensor web for disaster management is proposed. Finally, a 
summary is provided and future study is discussed. 
2. SENSOR WEB: A NOVEL SENSING APPROACH 
Sensor web is attracting more and more research attentions. 
Many organizations and universities step in the study of sensor 
web (Chu et al., 2006; Delin, 2006; Liang et al., 2005; 
McCarthy et al., 2008). Until now, there is no a universal 
definition of sensor web which can be accepted without any 
disagreement. A variety of definitions are also proposed by 
different researchers (Delin et al., 2005; Liang et al., 2005), all 
these definitions substantially help clarify the research 
boundary and goals of sensor web, though the focus of each 
definition are definitely different. No matter how to define a 
sensor web, a well accepted fact is that for any proposed 
definition, the basic features of sensor web should be inclusive. 
The paper attempts to summarize the basic features of sensor 
web instead of proposing a definition. 
2.1 The Basic Features of Sensor Web 
Sensor web has the following basic features: sensing, autonomy, 
dynamic reconfiguration, standards-based interaction, access. 
■ Sensing: for a sensor web, one of its fundamental features 
is the sensing ability 
■ Autonomy: if a sensor web is autonomous, it means that 
the sensor web can operate without external intervention. 
■ Dynamic reconfiguration: it refers to the dynamic 
configuration ability of sensor web based on the changing 
state and monitoring goals. 
■ Standards-based interaction: All components or elements 
within sensor web can interaction through standard service, 
such as OGC web service. 
■ Access: Sensor web must provides a mechanism to 
support the accessibility of sensing data or coordinated 
measurements. 
2.2 OGC Standard: Sensor Web Enablement (SWE) 
Within a sensor web, there are all kinds of working sensors. 
Moreover, the sensors are heterogeneous. In order to achieve a 
common sensing and monitoring goal through sensor web, 
obviously, all the sensors involved should be able to exchange 
data and information smoothly. It means that data 
interoperability is critical for sensor web. More importantly, 
from the users’ perspective, the discovering and accessing of 
sensing data must be standardized so that the exchange of 
sensor web services becomes possible. Unfortunately, the lack 
of data interoperability heavily hinders the practical application 
of sensor web. In order to solve this problem, Open Geospatial 
Consortium (OGC), a widely accepted organization, which 
focuses on the development of open standards for geospatial 
services, proposes a geospatial standard, i.e., Sensor Web 
Enablement (SWE) (Chu et al., 2006; Reed et al., 2007). SWE 
enables the implementation of interoperable and scalable sensor 
web. Just like the using of Hyper Text Markup Language 
(HTML) and Hyper Text Transfer Protocol (HTTP) turns the 
exchange of information on the internet into reality, SWE 
makes the discovery and exchange of sensor observations 
possible and feasible. In fact, SWE has become a de facto 
standard for the development of sensor web. Table 1 
summarizes the details on SWE. 
Specification 
Description 
Standard 
Models and 
XML 
Schema 
Observations & 
Measurements 
Schema (O&M) 
Encoding observations 
and measurements 
Sensor Model 
Language(Sensor 
ML) 
Describing sensors 
Transducer Markup 
Language 
(TransducerML or 
TML) 
Describing transducers 
Standard 
Web 
Service 
Interface 
Sensor Observations 
Service (SOS) 
Requesting and 
retrieving observations 
Sensor Planning 
Service (SPS) 
Requesting user-driven 
acquisitions 
Sensor Alert Service 
(SAS) 
Publishing and 
subscribing to alerts 
from sensors 
Web Notification 
Service (WNS) 
Delivering messages or 
alerts asynchronously 
from SAS, SPS or other 
web services. 
Table 1. OGC SWE Specifications 
2.3 Service Oriented Architecture and Web Service 
Service Oriented Architecture (SOA) is an essential 
infrastructure which can support the integration of distributed 
heterogeneous systems. SOA splits functions into distinct units, 
i.e., services, which are distributed over a network. The services 
can be published, discovered and requested dynamically within 
the network. The communication of different services is 
achieved by using XML messaging mechanism. Through SOA, 
these distributed services can be united and reused to support 
the new applications (Bell, 2008). Figure 1 gives an illustration 
of SOA. 
Web service can be used to support the implementation of SOA. 
Web services are independent from platforms and programming 
languages. Moreover, they are accessible over standard internet 
protocol (Ericsson, 2007).