| dc.description.abstract | In the future cyber-physical systems, such as smart grids, a large amount of sensors will
be distributively deployed in di erent locations throughout the systems for the purpose
of monitoring and control. Conventionally, sensors are powered by xed energy supplies,
e.g., regular batteries, which can provide stable energy output. However, such energy
sources require periodical recharging or replacement, which incurs high maintenance cost
and may become impractical in hazardous environments. Self-sustaining devices powered
by energy harvesting (EH) sources are thus highly desirable. However, energy provided
by energy harvesters is
uctuating over time and thus introduces the EH constraints to
the systems, i.e., the total energy consumed until an arbitrary time cannot be larger
than the harvested amount up to this time, which invokes the need of advanced power
control and scheduling schemes. This thesis studies both the o ine and online resource
allocation strategies for wireless communication systems empowered by EH sources.
First, the resource allocation problems for a Gaussian multiple access channel (MAC),
where the two transmitters are powered by a shared energy harvester, are studied. For
both in nite and nite battery capacity cases, the optimal o ine resource allocation
schemes for maximising the weighted sum throughput over a nite time horizon are
derived. It is proved that there exists a capping rate for the user with stronger channel
gain. Moreover, the duality property between the MAC with a shared energy harvester
and its dual broadcast channel powered is demonstrated. Numerical results are presented
to compare the performance of several online schemes. Moreover, the utility of a
greedy scheme against the optimal o ine one is measured by using competitive analysis
technique, where the competitive ratios of the online greedy scheme, i.e., the maximum
ratios between the pro ts obtained by the o ine and online schemes over arbitrary
energy arrival pro les, are derived. Then, the resource allocation schemes for the Gaussian MAC with conferencing links,
where the two transmitters could talk to each other via some wired rate-limited channels
and share a common EH source, are studied. The optimal o ine resource allocations are
developed for both in nite and nite battery cases. It is shown that the optimal resource
allocation in this scenario is more complicated than in the traditional MAC scenario and
there exists a capping rate at one of the two transmitters, depending on the weighting
factors. Online resource allocation strategies are also examined. Numerical results are
used to illustrate the performance comparison of the online schemes. Furthermore, the
competitive ratios of the online greedy scheme are derived under di erent weighting
factors. | |
