新型移动网络能耗改善机制研究

The convenience of mobile services leads to the explosive growth of the number of mobile devices and the increasing and creative mobile services. This thrives on the new wireless communication technologies and wireless networks, but also consequently roars the energy consumption of wireless networks. This proposal studies the problem of energy consumption reduction in future wireless networks. It plans to address this problem from the following three perspectives: the energy consumption reduction of mobile devices, internal wireless networks, and coordinate among multiple wireless networks. The detailed research works are as follows. 1) To work on reducing energy consumption of mobile devices by the offloading technology. This proposal considers the computing task unit as the granularity, and models the computing-transmission energy consumption. Moreover, it also plans to transform the execution sequence of computing task units into a graph, and calculate the latency of each task execution flow. At last, it plans to design a scheme for task offloading based energy consumption reduction with the constraint of the latency. 2) To work on reducing the energy consumption of data transmission over a wireless network by exploring cross-layer design. This proposal plans to develop a new architecture for the future wireless network. Moreover, it plans to design a novel scheme for energy consumption optimization for data transmission in future wireless networks by jointly considering routing, link scheduling, and physical layer parameters. Finally, it plans to apply the column generation based method to solve this formulated problem. 3) To work on energy consumption of heterogeneous wireless networks by using coordination. This proposal plans to establish the network utility model with the consideration of service QoS. Moreover, it plans to develop the optimal network selection strategy based on a dynamic differential game to achieve energy consumption reduction. The research results of this proposal will guide the development of green future wireless networks.

移动服务的便利性带来了移动设备数量持续爆炸式增长和移动业务不断推陈出新，促进了新型移动通信技术和网络飞速发展，但同时也导致移动网络的能耗问题越来越严重。本项目拟从移动设备-移动网内-移动网间三个层面开展新型移动网络能耗改善机制研究。具体研究内容和目标如下：1）基于计算卸载的移动设备节能研究，以计算单元为粒度，建立计算-传输能耗模型，将计算单元执行顺序转化为有向图并建模计算执行流的时延，设计在时延受限条件下计算卸载能耗改善机制；2）基于跨层设计的移动网络传输节能研究，设计新型多跳混合网络架构，提出综合考虑网络路由、无线链路、链路参数，以及无线资源的新型移动网络传输能耗优化机制，并应用列生成方法快速求解该优化问题；3）基于异构移动网络中移动网络间协同的节能研究，建立对应服务质量需求的网络效用模型，提出基于动态微分博弈的能耗最优网络选择策略。项目研究成果将为实现“绿色”新型移动网络供科学指导。

本项目的主要研究内容包括三个方面：1）基于计算卸载的移动设备能耗降低机制；2）移动网络中数据传输的能耗优化机制；3）基于异构无线网络协同的能耗改善机制。取得以下重要结果：1）在计算卸载方面，对于该方面的研究主要分为三个方面，分别为计算-传输的能耗模型、基于计算卸载的时延模型，以及时延受限的能量高效计算卸载机制；2）在数据传输的能耗优化机制方面，深入分析了新型移动网络中影响能量效率的因素以及研究层与层之间的相互影响，并设计出了优化TCP传输能量效率的统一框架；3）在异构无线网络协同的能耗改善机制方面，为研究异构无线网络协调的能耗改善机制，在研究过程中考虑5G环境下智慧城市服务中心的视频传输场景，研究5G场景下的人体行为识别问题，提出了基于稀疏神经网络的深度学习识别框架，以降低异构网络中的时延及能量消耗问题。本研究从三个方面出发，以减小整体新型移动网络的能耗为目标，对能耗优化的关键技术进行深入研究，实现动态、非确定环境下的能耗优化的数据传输，为新型移动网络框架提供科学指导。本项目在国内外重要学术期刊上发表学术论文4篇，其中SCI收录4篇（2篇JCR1区，2篇JCR2区）。培养硕士生10名，博士生2名，其中2名硕士研究生已经毕业，取得相应学位。