16.08.2013 Views

Gugrajah_Yuvaan_ Ramesh_2003.pdf

Gugrajah_Yuvaan_ Ramesh_2003.pdf

Gugrajah_Yuvaan_ Ramesh_2003.pdf

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Routing Protocols for Ad Hoc Networks Chapter 2<br />

There are three kinds of nodes in a cluster: cluster heads, gateways and internal<br />

nodes. The cluster head maintains topological information of the nodes in the cluster<br />

and acts as a local coordinator of transmissions within the cluster. The cluster head<br />

also summarizes link state information within its cluster and propagates it to the<br />

neighbouring cluster heads via the gateway nodes. The knowledge of connectivity<br />

between neighbour cluster heads leads to the formation of the next higher cluster<br />

level. A virtual link is formed between two cluster heads using the internal nodes and<br />

gateways nodes between the cluster heads. Nodes within a cluster exchange virtual<br />

link state information as well as summarized lower level cluster information.<br />

The Hierarchical Identification (HID) is used to identify each node and consists of a<br />

sequence of MAC addresses of the nodes from the top hierarchy to the node itself.<br />

For example, in Figure 2-10, the HID of node 10 is . For the delivery of a<br />

packet from node 5, with HID(5) = to node 10 with HID(10) =,<br />

the packet is first forward to node 1 (top hierarchy) which uses the virtual link<br />

(1,6,2,8,3) to deliver the packet to node 3. Node 3 then delivers the packet to node 10<br />

along the downwards hierarchical path.<br />

The drawback of HSR is that nodes have to maintain longer hierarchical addresses<br />

and have to continuously update the cluster hierarchy and hierarchical address as<br />

nodes move. Since a continuously changing hierarchical address makes it difficult to<br />

locate and keep track of nodes, logical partitioning is used in HSR. Logical<br />

partitioning involves using subnets of the network, each with its own home agent to<br />

manage membership. The subnets correspond to a particular user group, for example,<br />

students from the same class or tanks in the same battalion. Each member of a logical<br />

subnet knows the HID of its home agent from the routing table and registers its own<br />

HID with the home agent. The registration with the home agent is both event driven<br />

and time driven. It is assumed [Iwata99] that since members of the same subnet<br />

would move as a group, registration overhead is modest because the members will<br />

tend to reside in neighbouring clusters. When a source node requires a route to a<br />

destination it uses the logical address of the destination to send the packet to the<br />

home agent of the destination. The home agent redirects the packet to the destination,<br />

which can then continue communication using the newly acquired and current HID<br />

2-29

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!