In this paper, we consider a cellular system, in which base stations (BSs) are powered by both on-grid and renewable energy sources. To efficiently utilize the harvested energy of the BSs, we study adaptive traffic management (TM) and energy cooperation (EC) that aim at minimizing the on-grid energy consumption, while guaranteeing minimum average throughputs. To achieve this, we develop an adaptive TM and EC algorithm that jointly decides the energy sharing among BSs, the user association to BSs, and the sub-channel and power allocation in BSs. Within the algorithm, a network scheduling problem, which is mixed-integer non-linear programming (MINLP), should be solved in each timeslot. To efficiently solve it, we develop a network scheduling algorithm applying generalized Benders decomposition (GBD) that optimally solves the MINLP problem. In addition, we also develop a heuristic network scheduling algorithm that has a much lower computational complexity than the GBD algorithm, while providing comparable performance. Through the numerical results, we show that our algorithms always outperform the algorithms that use only one of TM or EC regardless of the system conditions.

中文翻译：

---

可再生能源蜂窝网络中的自适应流量管理和能源合作

在本文中，我们考虑一个蜂窝系统，其中的基站（BS）由并网能源和可再生能源供电。为了有效利用BS的已收集能量，我们研究了自适应流量管理（TM）和能源合作（EC），旨在最大程度地降低并网能耗，同时确保最低的平均吞吐量。为此，我们开发了一种自适应TM和EC算法，该算法共同决定BS之间的能量共享，用户与BS的关联以及BS中的子信道和功率分配。在该算法内，应该在每个时隙中解决混合整数非线性规划（MINLP）的网络调度问题。为了有效地解决它，我们开发了一种应用广义Benders分解（GBD）的网络调度算法，该算法可以最佳地解决MINLP问题。此外，我们还开发了一种启发式网络调度算法，该算法的计算复杂度比GBD算法低得多，同时提供了可比的性能。通过数值结果，我们表明，无论系统条件如何，我们的算法始终优于仅使用TM或EC之一的算法。