| dc.description.abstract | An Opportunistic Mobile Network (OMN) refers to the network paradigm where
wireless devices communicate with each other through the opportunistically formed
wireless links. Routing in OMN relies on node mobility and the store-and-forward
mechanism. It is paramount to have energy efficient, robust and cost effective routing
protocols in such environments. Previous research usually assumes that the
connectivity in such networks is extremely sparse and that the network is purely
infrastructure-less. However, real world deployments of OMNs actually exhibit di-
verse connectivity, i.e., connectivity may range from sparsely connected to well connected
or the network may coexist with infrastructure. Consequently, the simplified
assumptions of previous solutions lead to suboptimal behaviors of routing protocols,
which includes redundant transmissions, too much or insufficient data replications,
poor forwarding decisions, etc.
In this dissertation, in order to address the aforementioned problems, we propose
a communication framework for OMNs with diverse connectivity, which consists of
a series of algorithms and protocols that aim to provide energy efficient, robust and
cost-aware communication services to applications. In this framework, we propose:
a) algorithms that carefully schedule transmissions in an opportunistic contact involving
multiple nodes; b) routing protocols that consider simultaneously mobile
nodes' delivery capability and traffic load; c) mathematical tools that characterize
not only Inter-Contact Times but also their correlations; d) adaptive mechanisms to
realize dynamic data replication; and e) forwarding strategies that optimally trade-o_
energy consumption and delay in a cost-aware fashion when utilizing infrastructure.
We evaluate the proposed routing protocols and algorithms through extensive simulations using both synthetic network models and real world mobility traces. We also
conduct real world experiments on a wireless testbed to demonstrate their practicability.
The evaluation results show that, with the assumption of diverse connectivity
in mind, the proposed algorithms and protocols greatly improve the networking performance
and efficiency. The consideration of delay correlations and a mechanism
for dynamic replication are critical for a routing protocol to perform well with a wide
range of network connectivity. When infrastructure is present, our proposed forwarding
strategy helps improve the energy-delay trade-off when cost is a constraint. | en |
