Self-Configuration and Administration of Wireless Grids - Cloud, Grid and High Performance Computing: Emerging Applications - page 263

Self-Configuration and
Administration of Wireless Grid

tion (RS) present in some form as well as a Base
Station (BS). The RS maintains cognizance over
a set of wireless devices and serves as the final
mapping of logical to physical devices. The BS
manages and enforces policy within and among
groups. A grid layout can include a root station
for a community or an actual organization (AO)
of wireless nodes (figure 3). A root station will
maintain up-to-date information about its own
network and the associated nodes as well as
serve as the gateway to the wired network. Mul-
tiple organizations may come together to form a
virtual organization (VO). An AO can belong to
multiple VOs. A base station (BS) can be envis-
aged for a VO. A BS will maintain information
about networks for various organizations and the
associated root stations. For a homogeneous grid,
the same server can perform both the RS and BS
functions. In case of an inter-grid, which can span
multiple virtual organizations, several BSs are
needed to coordinate to maintain the inter-grid
information. Redundancy can be maintained by
having secondary servers to perform the RS and
BS functions. Both RS and BS should not be
resource-constrained devices. Instead, the RS
and the BS could be a simple PC, workstation, or
server equipped with an appropriate interface to
communicate with the edge nodes such as sensor
nodes or other mobile nodes.

As previously stated, wireless grids possess a
unique dynamic quality that is not found readily
in the wired grids. Therefore, technologies that
support self-configuration and self-administration
are critical to the continued growth of the wire-
less grid paradigm. Wireless grids should allow:

• Configuration of addresses for the grid

components: nodes, RS and BS

• Name- to- address resolution for the grid

components

• Maintenance of the state information for

the grid

The address for the nodes can be obtained in
several different ways. It is possible that the ad-
dress may not be an IP address in case the device
is a sensor with no IP stack. We envision that an
IP incapable node could use, as its own address,
either the MAC address of the system chip or
a unique serial number provided at the time of
manufacturing the device. A name, unique to the
AO domain, can be assigned to the device through
an automatic handshake process between the de-
vice and the RS. RS and BS are connected to the
wired infrastructure and can obtain IP addresses
using the DHCP protocol (Droms, 1997).

Figure 2. A hybrid wireless network