|
The prime goal of todays communication is to keep anybody capable of
communicating with anyone, anywhere, anytime. The wireless is the
only medium that can satisfy all of these requirements. A Mobile Ad
Hoc Network (MANET) is a decentralized communication network with
changeable topology. Flooding is a low level primitive of a MANET.
Therefore, successful research in the area can significantly improve
the overall performance of the network.
Due to the variations of mobility, movement and speed of mobile
nodes, there is a change of node distribution in a network. Varying
network topology makes it hard for typical networking tasks such as
information diffusion or flooding. That is why nodes have to adapt
their protocols on-the-fly. A node can choose its own diffusion
strategy based on the local or global topology knowledge. But both
strategies yield inefficient performances.
This leads us to make a compromise between local and global
strategy: region-based adaptation. In region-based adaptation, the
whole network is partitioned into regions based on node
distributions (node density). We choose Gini Coefficient as a means
of estimating inequalities of node distribution. Regions are formed
with network parts of almost equally distributed mobile nodes.
Based on region node distribution, an appropriate diffusion strategy
is selected. In this thesis, we choose HyperGossiping as an
application of regioning concept. We analyze performances of
HyperGossiping with regioning and without regioning. Simulation
results indicate that up to 21% (mean 14%) of rebroadcasts per
packet per node could be saved assuming the availability of
regioning information.
|