Self-Configuration and Administration of Wireless Grids - Cloud, Grid and High Performance Computing: Emerging Applications - page 261

Classification by Usage Pattern urgent search for donors with a rare blood

type. A hospital would issue a query to the
medical history databases in the region
through its mobile network. The mobile
service providers will notify potential do-
nors through the alert messages transmit-
ted to their respective mobile devices, and
the resulting responses would be processed
and reconciled. Internet serves as a mas-
sive data grid where the information re-
sides on multiple servers and such infor-
mation can be accessed using portals and
search engines or by directly request to a
particular IP address.
• Utility Grid: Here the motivation for the
wireless grid is derived from the need to
provide ubiquitous access to specialized
pieces of software and hardware. Users
can request resources when needed (on-de-
mand) and only be charged for the amount
being used. This model can subsume both
Computational and Data grids. For exam-
ple, users might tap Wireless Utility grids
for information such as the traffic condi-
tions and routing, and for making instan-
taneous transactions related to commercial
products and services.

Wireless grids can be classified by usage patterns
as summarized in Table 1.

• Computational Grid: In a computational
grid, the need for creating the wireless grid
is driven primarily by the need to borrow
computational resources from others. This
arises, in part, because of the power con-
straints on mobile devices, which in turn
limits their computational capability. The
computational grid may be cooperative or
parasitic (Barabasi et al, 2001). In a coop-
erative setup, inputs from multiple nodes
are needed to analyze a particular scenario.
For example, sensor network deployed in
the battlefield would present the enemy's
position. Similarly, a wireless sensor net-
work will be used to monitor conditions for
predicting natural calamities like earth-
quakes or volcanoes. In a parasitic setup,
the nodes would rely on each other to man-
age the power constraints. Any remote
setup, will allow for this possibility due to
lack of other power resources. Some kind
of redundancy would be built in such a
setup.
• Data Grid: In this case, the need for creat-
ing the wireless grid is dictated primarily
by the need to provide shared and secure
access to distributed data. Since data can
be presented in various contexts on vari-
ous systems, reconciling the underlying
semantics continues to challenge evolv-
ing technology. One example involves an

TECHNICAL CHALLENGES

Among the many challenges wireless grids face,
these grids must overcome the following set of
initial technical challenges:

• Dynamic Configurability: Wireless grids
are characterized by changing topology

Table 1. Wireless Grid usage patterns

Grid Type Possible Architecture Mainly Provides

Computational Cluster, Intra, Inter Computational Power

Data Cluster, Intra, Inter Data Access and Storage

Utility Intra, Inter On-demand Access various of Resources