Abstract:
Reliable data delivery is an important feature of mission critical
wireless sensing applications. Achieving high degree of reliability
is a challenging task due to scarcity of resources in tiny
sensor devices in general and energy de ciency in particular.
This dissertation investigates inner dynamics of standard reliable
data delivery mechanisms based on theoretical frameworks
and empirical analysis. To achieve this purpose, we develop realistic
wireless channel models to identify the underlying bit-error
process of wireless communication in sensor networks. Using
knowledge of communication channels and reliable data delivery
mechanisms, we propose improved schemes for reliable data
delivery in wireless sensor networks which are more energy ef-
cient as compared to existing schemes. Compatibility with
existing standards is considered as a basic constraint in the design
of the new protocols. Each protocol proposed in this work
is stochastically modeled and veri ed empirically using a comprehensive
set of wireless traces. Our results show up to 40%
improvement in energy e ciency. However, we observe that the
energy e cient design has its limits as the reduction in energy
consumption through energy e cient design does not signi -
cantly improve network wide performance measures. Therefore,
we conclude that the fundamental problem lies with the coupled
energy-sensing design philosophy presently followed in sensor
networks. To alleviate this problem, we propose a paradigm
shifting concept of decoupling energy sources and sensing in
wireless sensor networks. Through theoretical analysis and empirical
evaluation, we show that energy decoupling is not only
a feasible energy enhancement idea, but also has the potential
to completely change the energy conscious design focus of the
sensor network community.
