随着无线通信工业的迅速发展，能量消耗快速增加。作为缓解由此带来的环境和能源短缺问题的有效解决方案，绿色通信成为未来发展的趋势。应用随机优化理论，本课题研究绿色无线通信中网络层缓存时延和物理层功率的跨层优化问题。发送端产生的数据存放在先进先出的缓存队列中，在每个发送周期从缓存中取一些数据发送到接收端。发送数据所需总功率包括数据传输功耗和电路的功耗。发送端可以使用可再生能量也可以从电网购买能量。本课题研究平均电网能量约束下通过设定每个发送周期中发送数据的数量和可再生能量的分配（其余能量从电网获得）使得数据在缓存中的平均等待时间（缓存时延）最短。首先，对相关物理过程（数据产生、可再生能量到达、无线信道变化）进行随机模型描述；在此基础上，建立研究问题的随机优化模型并在马尔科夫决策框架下对其进行分析、求解，提出发送数据数量和可再生能量分配的设定机制；最后，结合李雅普诺夫优化提出实用的低复杂度的算法。

Energy consumption increases greatly with the rapid development of wireless communications industry. As a promising solution to alleviate the severe environmental problems and energy shortage, green communications becomes a prospective topic. In this project, we investigate the physical layer power and the network layer buffer delay in green wireless communications by utilizing stochastic optimization theory. The data generate randomly in the transmitter, and are stored in a first-in-first-out buffer before transmitting. In each transmission period, the transmitter chooses some data in the buffer head, and transmits to the receiver. The total power consumption includes transmission power and circuit power. The transmitter could utilize renewable energy, and it can purchase the grid power. In the project, we study the minimization of mean data waiting time in the buffer (buffer delay) through the arrangement of transmitting data amount and renewable power allocation (the rest power is purchased from grid) in each period. Meanwhile, mean grid power constraint should be considered. The related processes (e.g., the data arrival, the renewable energy arrival, and the fading wireless channel) are modeled as proper stochastic processes. Then, we formulate a stochastic optimization problem, analyze the formulated problem under the Markov decision process (MDP) framework, and propose a arrangement mechanism for transmitting data and renewable allocation thereby. Finally, low complexity algorithms are given utilizing the Lyapunov optimization.