Energy Harvesting Wireless Networks: Transmission Policies and Coding Scheme

With the wide deployment of battery powered wireless devices, prolonging the lifetime of wireless networks is becoming ever more critical. Systems powered by batteries suffer from a limited lifetime, whereas networks consisting of energy harvesting nodes can survive indefinitely. Along with its unique benefits, energy harvesting introduces a new challenge in terms of managing the intermittently available energy. Energy harvesting wireless networks call for transmission policies and coding schemes tailored to their variable nature of available energy.;A fundamental concern in energy harvesting wireless nodes is to maximize system performance under the inherent energy constraints. This entails utilizing the limited capacity battery to adapt to variations in harvested energy. The first part of this thesis investigates power policies to maximize the throughput of energy harvesting nodes for various system models. First, a single-user communication scenario with an energy harvesting transmitter is considered. Subject to energy causality and battery capacity constraints derived from offline knowledge of energy harvests, the throughput maximizing power policy is found. In particular, properties of the optimal power policy are identified, and the benefit of power control for energy harvesting communications is demonstrated.;Next, extensions to inefficient energy storage, interfering transmitters, and energy cooperation are considered. In particular, for the inefficient energy storage case, the trade optimal power policy is attained with thresholds that regulate battery utilization. Similar thresholds are observed to arise for the fading channel, and for the online problem where harvested energy is revealed causally to the transmitter. As a multi-transmitter setting, the interference channel with two energy harvesting transmitters and two receivers is studied. This configuration requires nodes to not only adapt to their own energy harvests, but to adapt to each others' transmission policies as well. To solve the problem, an iterative algorithm based on directional water-filling is proposed, and is shown to admit simpler interpretations in specific interference regions. A new dimension to energy management in energy harvesting networks is introduced via wireless energy transfer, which has become increasingly efficient in the past decade. The impact of energy cooperation is studied for various channel models by finding the jointly optimal power allocation and energy cooperation policies. Through the use of a subset of optimal policies, this joint optimization problem is simplified and solved with methods similar to the non-cooperating case.;Intermittency of available energy leads to challenges in coding as well. In the second part of this thesis, a simple energy harvesting channel is studied from an information theoretic perspective, with the purpose of identifying the theoretical limits of communication under energy harvesting constraints at the codeword symbol level. The equivalence of this channel to a timing channel is utilized to propose encoding schemes and upper bounds that characterize its capacity within a small gap. Moreover, the amount of information revealed to the decoder about the energy harvesting process is compared for different encoding schemes, addressing the state amplification and state masking problems for energy harvesting channels.