Information Technology Reference
In-Depth Information
own service information to one-hop neighbors. All
nodes acquire their knowledge from other nodes,
and thus service lookup is done by searching local
cache. Intentional Naming System (INS) (Adjie-
Winoto, Schwartz, Balakrishnan, & Lilley, 1999)
from MIT is a new naming system to name and
discover different services. The innovative char-
acteristic of INS is late binding, which enables
service and service location mapping just before
the service access. A following project, INS/Twine
(Balazinska, Balakrishnan, & Karger, 2002),
hashes and stores service attributes in mesh struc-
ture directories. It uses peer-to-peer technology to
look up services. Peer-to-peer technology allows
INS/Twine to be able to scale up to millions of
services. But on the other hand, service lookups
may have to go through several directories and
thus have additional latency. Researchers at UC
Berkeley proposed SSDS (Czerwinski, Zhao,
Hodes, Joseph, & Katz, 1999). SSDS focuses more
on security and scalability issues. Privacy and
security are enabled by public key and symmetric
key encryption. Different hierarchical directory
structures are considered to support scalability.
Industry. Sun Microsystems' Jini is based on
Java technology (Sun Microsystems, 2001). Java
technology makes Jini platform independent of
the underlying operating systems and hardware,
but all the clients, services, and directories need
Java runtime environments directly or indirectly.
Microsoft Corporation ships operating systems
with UPnP (Miller, Nixon, Tai, & Wood, 2001).
UPnP targets unmanaged networking environ-
ments, such as home environments. UPnP uses
XML format to store service information and
communicate among services and clients. Thus,
UPnP is platform and programming language inde-
pendent and device-oriented. Rendezvous at Apple
Computer's is a DNS-based service discovery
protocol (Cheshire, 2002). The ubiquity of DNS
servers might facilitate the Zero Configuration
networking (Zeroconf) prototocol to be adopted
(Apple Computer Inc, 2003).
Organizations. Bluetooth, from the Bluetooth
Special Interest Group (SIG), is now widely used.
It allows nearby devices to discover and commu-
nicate with each other at low power consumption
(Bluetooth SIG, 2001). Salutation protocol, pro-
posed by the Salutation Consortium (Salutation
Consortium, 1999), is an open source protocol.
Salutation protocol implements two interfaces,
one of which is designed to be independent to the
transport layer, so that it can be used on various
transport protocols. Service Location Protocol
(SLP) is posted by IETF in enterprise environ-
ments (Guttman, Perkins, Veizades, & Day, 1999).
SLP defines a way to locate a service, but it leaves
the interaction between clients and services after
service discovery to the application developers.
AN ANALYSIS OF ARCHITECTURE
AND PROTOCOL DESIGNS
Much active service discovery research has been
occurring as we discussed in the last section.
Targeted at different environments, these service
discovery protocols have different design criteria
and choices.
Issues, Controversies, Problems
Service discovery protocols provide desired func-
tionalities, yet they face great challenges. First,
unlike traditional network services, the services
and devices are highly dynamic. New devices and
services may be added without users' knowledge.
Sensors, services, network connections may not
be available all the time. Second, the pervasive
computing environments are extremely heteroge-
neous. Different types of operating systems, net-
work topologies, network protocols and devices,
owned and administrated by different people or
organizations.
Most service discovery protocols are designed
as application layer protocols. Thus, many hetero-
geneous issues are handled by underlying network
Search WWH ::




Custom Search