Online Resource Allocation with Energy Harvesting for Underwater Wireless Networks

水下无线通信网络在环境监测、海上勘探、灾害预防、军事监视等方面有着广泛的应用。这些无线网络需要提供复杂的实时数据收集，并延长部署时间，例如30至50年。本研究的目标是开发永久水下无线网络，其中高速多输入多输出（MIMO）收发器由持久的超级电容器通过从环境能源中收集能量来供电。本项目将开发实时资源分配和随机优化解决方案，以有效地分配水下能量收集无线网络的带宽和功率资源，考虑到能量收集电源的独特动态和水下信道的时变。该方法将建立一种实用的能量收集MIMO收发器设计方法，在高动态水声通信环境中实现高可靠性、高数据速率和高能效。该项目的另一个重要目标是通过一系列教育和推广活动培养下一代研究人员，包括课程开发、研究生和本科生培训、为当地高中机器人团队服务以及参加海洋机器人比赛。这些活动还将提倡负责任的工程方法，以减少海洋中的声音污染。水下无线通信通常需要大型MIMO收发器来实现对可靠性和吞吐量的苛刻要求。因此，能量收集MIMO通信的资源分配成为一个复杂的预编码和带宽分配的联合设计，涉及大量的设计变量。同时，能量收集电源的动态与水下信道的时变具有相似的时间尺度，因此需要在线随机优化，而不是MIMO射频网络中常用的离线资源分配。为了解决永久水下无线网络中这些独特的挑战，本项目将联合在线预编码器设计和带宽分配问题转化为一个多变量随机动态规划（SDP）问题，并提出了一种新颖的嵌套优化算法，以低计算复杂度迭代求解该问题。该项目通过一种新颖的嵌套优化算法，解决具有线性计算复杂性的多变量SDP问题，以及一种在线资源分配方法，该方法在通信信道和能源上使用随机模型，而不是常用的确定性模型，有助于推进无线通信和网络技术。还将建立硬件试验台，对在线资源分配算法和超级电容器能源进行验证。

Underwater wireless communication networks have found a wide range of applications, such as environmental monitoring, offshore exploration, disaster prevention, and military surveillance. These wireless networks are required to provide sophisticated real-time data collection and serve for prolonged deployment time, such as 30 to 50 years. The objective of this research is to develop perpetual underwater wireless networks in which high-speed multiple-input multiple-output (MIMO) transceivers are powered up by long-lasting super-capacitors via energy harvesting from environmental energy sources. This project will develop real-time resource allocation and stochastic optimization solutions to efficiently allocate the bandwidth and power resources in underwater energy-harvesting wireless networks, taking into account the unique dynamics of energy-harvesting power sources and the time variation of the underwater channels. The proposed approach will establish a practical design approach for energy-harvesting MIMO transceivers to achieve high reliability, high data rate, and high energy efficiency, in the presence of highly dynamic underwater acoustic communication environments. Another important goal of the project is to train next-generation researchers through a set of education and outreach activities including course development, graduate and undergraduate student training, service to local high school robotics teams, and participation in marine robotics competition. These activities will also advocate responsible engineering approaches for reducing sound pollution in oceans.Underwater wireless communications often require large MIMO transceivers to achieve the demanding requirements on reliability and throughput. As such, resource allocation for energy-harvesting MIMO communication becomes a complicated joint design of precoding and bandwidth allocation that involves a large number of design variables. Meanwhile, the dynamics of energy-harvesting power sources and the time variation of the underwater channels are on a similar time scale, thus requiring online stochastic optimization instead of the offline resource allocation commonly used in MIMO RF networks. To address these unique challenges in perpetual underwater wireless networks, this project formulates the joint online precoder design and bandwidth allocation into a multi-variable stochastic dynamic programming (SDP) problem, and proposes a novel nested optimization algorithm to solve it iteratively with low computational complexity. This project contributes to advancing wireless communications and network technology by a novel nested optimization algorithm that solves the multi-variable SDP problem with linear computational complexity and an online resource allocation method that uses stochastic models on both communication channels and energy sources rather than commonly used deterministic models. A hardware test bed will also be built to verify the online resource allocation algorithms and super-capacitor energy sources.