PEER-TO-PEER BASED GIS WEB SERVICES 
J.H.Guan ?, S.G. Zhou ^, L.C.Wang *, F.L.Bian * 
* School of Computer, Spatial Information & Digital Engineering Center, Wuhan University, Wuhan, 430079, China- 
jhguan@wtusm.edu.cn 
° Department of Computer Science & Engineering, Fudan University, Shanghai, 200433, China-sgzhou@fudan.edu.cn 
TS, WG 11/3 
KEY WORDS: GIS, Research, Interoperability, Infrastructure, Internet, Web based, Distributed 
ABSTRACT: 
Geographic Information System (GIS) based on Internet is a promising research field, which brings new approaches to access, share 
and disseminate geographic information. However, current Internet-based GIS relies heavily on centralized server, which inevitably 
has drawbacks such as single points of failure and network congestion etc. In recent years, Peer-to-Peer (simply P2P) and Web 
services are two of hot research topics in network computing and appear as two extremes of distributed computing paradigm. By 
adopting a decentralized network-based style, P2P technology can improve scalability/reliability, enhance the overall reliability and 
fault-tolerance, increase autonomy and enable ad-hoc communication and collaboration. On the contrary, current proposals for Web 
services infrastructures are mainly based on centralized approaches, which are prone to introducing single points of failure, hotspots 
in the network, and exposing vulnerability to malicious attacks. In this paper, we explore the techniques of building GIS Web 
services systems in P2P environment. By combining Web services and P2P technologies into GIS, we aim to add more flexibility 
and autonomy to GIS Web services systems, and alleviate to some degree the inherent limitations of centralized systems. We 
propose a P2P based GIS Web Service architecture, and as a case study, we present our ongoing project BP-GServices, i.e., BestPeer 
based GIS Web Services, and the major techniques of BP-GServices implementation. 
1. INTRODUCTION direct communication among clients, and enable resource 
aggregation, so thus provide promising opportunities for novel 
applications to be developed (Ooi, 2002). 
In conventional GISs, geographical data is typically generated 
and stored locally and then utilized by a limited number of On the other hand, Web services technologies provide a 
specialized computations or services on that site. For decades, language-neutral and platform-independent programming model 
GIS has relied heavily on centralized paradigm with the that can accelerate application integration inside and outside the 
complete set of data stored on one single server. Remote enterprise (Gottschalk, 2002). It is convenient to construct 
computers can access the data via the interconnection flexible and loosely coupled business systems by application 
networking with quite high networking access costs. To get the integration under Web services framework. However, current 
required geographic information or services, we have to gather proposals for Web services infrastructures are mainly based on 
information from various locations and process them by using centralized approaches such as UDDI (UDDI, 2004): a central 
centralized server paradigm, which has inherent drawbacks repository is used to store services descriptions, which will be 
such as single points of failure, network congestion, and data queried to discover or, in a later stage, compose services. Such 
inconsistency etc. With the popularity of Internet and the centralized architecture is prone to introducing single points of 
diversity of GISs, more and more geo-referenced information failure, hotspots in the network and exposing vulnerability to 
sources spread over the network. The inherent disadvantages of malicious attacks. Furthermore, making full use of Web 
traditional GISs need to be solved for new applications on services capabilities using a centralized system does not scale 
Internet or Web. gracefully to large number services and users. This difficulty is 
severed by the evolving trend to ubiquitous computing in which 
Nowadays Peer-to-Peer (P2P) and Web services are two ofthe more and more devices and entities become services and service 
hot research topics in network computing. Roughly, they appear ^ networks become extremely dynamic due to constantly arriving 
as two extremes of distributed computing paradigm. and leaving service providers. 
Conceptually, P2P refers to a class of systems and applications 
that employ distributed resources to perform a critical function To overcome the limitations of Web services systems causing 
in a decentralized way. A P2P distributed system typically by their centralized architecture, we explore the techniques of 
consists of a large number of nodes that can potentially be building GIS Web services applications under P2P environment. 
pooled together to share their resources, information and ^ By fitting GIS Web services into P2P environment, we aim to 
services. These nodes, taking the roles of both consumer and add more flexibility and autonomy to GIS Web services systems, 
provider of data and/or services, may join and depart the P2P and alleviate to some degree the inherent limitations of the 
network at any time, resulting in a truly dynamic and ad-hoc centralized systems. In this paper, we explore the techniques of 
environment. In addition, the distributed nature of such a design building GIS Web services systems in P2P environment. We 
can eliminate the need for costly infrastructure by enabling propose a P2P based GIS Web Service architecture, and as a 
820 
Inter 
  
case 
Best! 
BP-C 
The 
over 
GSeı 
impl 
that, 
the c 
Best] 
impl 
Univ 
appli 
illust 
LI 
Bestl 
of no 
of I 
serve 
and v 
other 
data ; 
publi 
LIGL 
Locat 
It pre 
Identi 
such | 
onlint 
LIGL 
BestF 
has ft 
1)