<p>Solar energy is a clean and abundant renewable energy source which is currently used in many types of photovoltaic (PV) designs. In practical PV systems, solar panels are used to harvest solar energy and convert it into a usable form of electricity. Due to the intermittent nature of solar energy input however, battery storage, in combination with solar panels, must be used to provide an uninterrupted source of power.</p> <p>The process of assigning solar panel and battery configurations for a PV system is referred to as energy resource provisioning. Unfortunately, energy provisioning costs are still relatively high, and this is one of the main obstacles that inhibits the adoption of solar power for many applications. These costs however, can be substantially reduced through cost-efficient resource provisioning methods. The focus of this thesis is on the development of efficient algorithms and energy management methods that will reduce energy provisioning costs in solar powered systems.</p> <p>First, we consider resource provisioning in solar powered wireless mesh networks. In practical solar powered systems, there are usually restrictions in the way that the mesh nodes can be positioned, and this results in a time-varying and node-dependent attenuation of the available solar energy. Unfortunately, conventional resource provisioning methods cannot take this into account and therefore the deployed system may be unnecessarily expensive. In this part of the thesis, the resource provisioning problem is considered from this point of view. We first review conventional resource provisioning mechanisms and give an example which shows the value of introducing positional solar insolation awareness. A Position Aware Provisioning (PAP) algorithm is then introduced that takes known positional variations into consideration when performing the energy provisioning. Simulation results show that reductions in total network provisioning cost can be obtained using the proposed methodology compared to conventional algorithms.</p> <p>In the second part of the thesis, we consider communication infrastructure that is operated from the power grid with a solar powered addition. Resource provisioning and energy management algorithms are introduced to minimize the capital expenditure (CAPEX) and operating expenditure (OPEX) costs. We first derive lower bounds on the costs using a linear programming (LP) formulation where solar components are sized using solar insolation and projected loading data. A variety of different node configurations are considered. Three energy scheduling algorithms are then introduced to optimize online OPEX costs, namely, Grid Purchase Last (GPL), Solar Load Optimization (SLO) and Solar Load Simulation (SLS) algorithms. Simulation results show the extent to which a solar powered add-on can reduce total cost.</p> <p>Finally, we consider solar powered systems where part of their energy demands are deferrable, up to some maximum tolerable delay. The objective is to exploit the flexibility of deferrable energy demands in a way that decreases the total provisioning cost. A mixed integer linear optimization program is derived which gives a lower bound on the provisioning cost. A Delay Aware Provisioning (DAP) algorithm is then proposed to determine practical cost-efficient energy provisioning. The performance of DAP is compared to the provisioning bound and the conventional Stand-alone Node Provisioning (SNP) algorithm. Results are presented which show the significant provisioning cost savings that can be obtained.</p> / Doctor of Philosophy (PhD)