Stochastic Optimization of Energy Harvesting Wireless Communication Networks

Energy harvesting from environmental energy sources (e.g., sunlight) or from man-made sources (e.g., RF energy) has been a game-changing paradigm, which enabled the possibility of making the devices in the Internet of Things or wireless sensor networks operate autonomously and with high performance for years or even decades without human intervention. However, an energy harvesting system must be correctly designed to achieve such a goal and therefore the energy management problem has arisen and become a critical aspect to consider in modern wireless networks. In particular, in addition to the hardware (e.g., in terms of circuitry design) and application point of views (e.g., sensor deployment), also the communication protocol perspective must be explicitly taken into account; indeed, the use of the wireless communication interface may play a dominant role in the energy consumption of the devices, and thus must be correctly designed and optimized. This analysis represents the focus of this thesis. Energy harvesting for wireless system has been a very active research topic in the past decade. However, there are still many aspects that have been neglected or not completely analyzed in the literature so far. Our goal is to address and solve some of these new problems using a common stochastic optimization setup based on dynamic programming. In particular, we formulate both the centralized and decentralized optimization problems in an energy harvesting network with multiple devices, and discuss the interrelations between these two schemes; we study the combination of environmental energy harvesting and wireless energy transfer to improve the transmission rate of the network and achieve a balanced situation; we investigate the long-term optimization problem in wireless powered communication networks, in which the receiver supplies wireless energy to the terminal nodes; we deal with the energy storage inefficiencies of the energy harvesting devices, and show that traditional policies may be strongly suboptimal in this context; finally, we investigate how it is possible to increase secrecy in a wireless link where a third malicious party eavesdrops the information transmitted by an energy harvesting node.