Service Discovery Architecture and Protocol Design for Pervasive Computing - Advanced Design Approaches to Emerging Software Systems - page 86

Figure 1. Interactions among the components in service discovery models. (a) Interactions in the client-
service model. (b) Interactions in the client-service-directory model

Directory View With service discovery software installed, the

client does not need to configure service settings.
If a device driver is needed to access a service,
the driver will be installed just-in-time before
the service accesses. Therefore, users will be
released from the burden of installing software
on all client devices.

With directories available, a client queries twice,
the first time asking directories and the second time
contacting the service(s). Without directories, a
client looks for services directly.

• On hearing a service announcement, direc-

tories first check privileges of the service;
and then service information will be updat-
ed or recorded in the directories.

Service View

A service has a name, a list of attributes, and user
privileges. For instance, a printer says it provides
printing service and it is able to provide color
printing at 720 by 720 dpi. It might only allow
people in the marketing department to use it. When
the device needs to use other services, the device
becomes a client. Services work as follows.

• Alternatively, directories may ask what

services are available instead of waiting
for the service to be announced.

• When receiving a query from a client, di-

rectories authenticate, authorize, and reply
to the client.

Service Discovery Protocols • A service announces its information to cli-

ents or directories. For example, every ten
minutes a temperature sensor announces
its information to let clients or directories
know its existence.

Research activities in service discovery have been
very active in academia, industry and international
standards development organizations. Here, we
list a few representative protocols in each of the
communities.
Academia. IBM Research's DEAPspace
(Nidd, 2001) is a service discovery protocol for
single-hop ad hoc environments. Each node that
runs the DEAPspace algorithm caches service
information. Then, it broadcasts on the wireless
channel the cached service information and its

• Alternatively, a service answers directory

solicitation or client queries.

• A service authenticates and authorizes the

user when a client asks for service.

• Finally, the client is granted service access

and uses the service through the service's
interface.