Resumen de Collaborative beamforming schemes for wireless sensor networks with energy harvesting capabilities
Lazar Berbakov
In recent years, wireless sensor networks have attracted considerable attention in the research community. Their development, induced by technological advances in microelectronics, wireless networking and battery fabrication, is mainly motivated by a large number of possible applications such as environmental monitoring, industrial process control, goods tracking, healthcare applications, to name a few. Due to the unattended nature of wireless sensor networks, battery replacement can be either too costly or simply not feasible. In order to cope with this problem and prolong the network lifetime, energy efficient data transmission protocols have to be designed. Motivated by this ultimate goal, this PhD dissertation focuses on the design of collaborative beamforming schemes for wireless sensor networks with energy harvesting capabilities. On the one hand, by resorting to collaborative beamforming, sensors are able to convey a common message to a distant base station, in an energy efficient fashion. On the other, sensor nodes with energy harvesting capabilities promise virtually infinite network lifetime. Nevertheless, in order to realize collaborative beamforming, it is necessary that sensors align their transmitted signals so that they are coherently combined at the destination. Moreover, sensor nodes have to adapt their transmissions according to the amounts of harvested energy over time. First, this dissertation addresses the scenario where two sensor nodes (one of them capable of harvesting ambient energy) collaboratively transmit a common message to a distant base station. In this setting, we show that the optimal power allocation policy at the energy harvesting sensor can be computed independently (i.e., without the knowledge of the optimal policy at the battery operated one). Furthermore, we propose an iterative algorithm that allows us to compute the optimal policy at the battery operated sensor, as well. The insights gained by the aforementioned scenario allow us to generalize the analysis to a system with multiple energy harvesting sensors. In particular, we develop an iterative algorithm which sequentially optimizes the policies for all the sensors until some convergence criterion is satisfied. For the previous scenarios, this PhD dissertation evaluates the impact of total energy harvested, number of sensors and limited energy storage capacity on the system performance. Finally, we consider some practical schemes for carrier synchronization, required in order to implement collaborative beamforming in wireless sensor networks. To that end, we analyze two algorithms for decentralized phase synchronization: (i) the one bit of feedback algorithm previously proposed in the literature; and (ii) a decentralized phase synchronization algorithm that we propose. As for the former, we analyze the impact of additive noise on the beamforming gain and algorithm’s convergence properties, and, subsequently, we propose a variation that performs sidelobe control. As for the latter, the sensors are allowed to choose their respective training timeslots randomly, relieving the base station of the burden associated with centralized coordination. In this context, this PhD dissertation addresses the impact of number of timeslots and additive noise on the achieved received signal strength and throughput
