The proliferation of wireless devices and applications has propelled the explosion of data traffic and energy consumption in wireless systems. This will inevitably cause a huge ecological and economical load for human society. Energy harvesting (EH) technology emerges at the right moment to cater for this energy requirement, and stimulates green communications, as well as highly self-sustainable next-generation smart wireless networks. However, the performance of EH-powered wireless networks is limited by the small amount and time-varying properties of the harvested energy. To achieve better communication performance, researchers have devoted a lot of effort on EH wireless networks. Nevertheless, the performance of the point-to-point communication link with EH nodes is fundamentally lim...[ Read more ]

The proliferation of wireless devices and applications has propelled the explosion of data traffic and energy consumption in wireless systems. This will inevitably cause a huge ecological and economical load for human society. Energy harvesting (EH) technology emerges at the right moment to cater for this energy requirement, and stimulates green communications, as well as highly self-sustainable next-generation smart wireless networks. However, the performance of EH-powered wireless networks is limited by the small amount and time-varying properties of the harvested energy. To achieve better communication performance, researchers have devoted a lot of effort on EH wireless networks. Nevertheless, the performance of the point-to-point communication link with EH nodes is fundamentally limited by the low EH rate. Fortunately, cooperative relays are verified as powerful tools to improve the communication performance for wireless systems. Hence, we aim to investigate the transmission strategies of EH wireless networks with cooperative relays, which shall greatly improve the performance while bear low complexity.

In this thesis, we first consider the channel training design in EH wireless networks. The EH constraints pose new challenges for the training design, and such design takes on different properties compared with the non-EH case. The training period and training power are jointly optimized, but their coupling complicates the problem. We obtain two suboptimal solutions based on the instantaneous energy profile and the average energy harvesting rate, respectively. Both solutions are verified to achieve near-optimal performances.

Next, we investigate the single-relay EH networks where the source is an EH node. Joint time scheduling and power allocation problems are formulated to either maximize the total transmitted data within a given duration, or to minimize the transmission time for a given amount of data. Despite the difficulties caused by the half-duplex relay and the coupling nature of the problems, we develop optimal algorithms by constructing a kind of directional water-filling (DWF) EH profile for the EH node, based on the DWF algorithm in single-hop EH communication systems. It is verified that the joint design of time scheduling and power allocation significantly improves the performance over the fixed policies.

In the following, we propose to achieve power minimization for wireless networks consisting of multiple non-EH source-destination (S-D) pairs, without channel side information (CSI) at transmitters, by deploying multiple EH relays. We propose a strategy for different S-D pairs to share those EH relays, and optimize it by jointly designing the transmit power assignment and relay selection. The design problem is rendered challenging by the NP-hard nature of the problem, the coupling effect among all the nodes, and the large variable size. We manage to simplify the problem, and further obtain a general framework to derive sub-optimal solutions, by proposing an easy-to-check sufficient condition for the problem feasibility. The proposed strategy provides performance close to the upper bound and greatly outperforms the greedy policy which only optimizes the instantaneous performance.

Besides the above works, we further study the multi-relay EH network with CSI at transmitters. A dynamic programming problem is formulated to optimize the relay selection for this cooperative EH network. To overcome the high complexity and curse of dimensionality, we propose a novel relay selection rule based on the defined relative throughput gain. This relay selection rule has a large performance gain over the conventional greedy policies. Moreover, it well captures the key characteristics of EH relaying wireless systems, and is also very practical to implement.