CASIA OpenIR  > 毕业生  > 博士学位论文
基于平行网络的无线资源管理和分配技术研究
杨坚
2017-05
学位类型工学博士
英文摘要

随着通信技术的高速发展,人们对通信业务的需求从最初的语音通话逐渐转换为更加复杂的混合业务(如音频、视频等),同时对服务质量(Quality of Service, QoS)的需求也不断提高。未来无线通信系统面临的最大挑战之一就是提高网络频谱资源利用率,为用户提供更高的QoS。然而,随着用户数量的增加、网络规模的扩大以及业务的创新,基于现有网络架构的无线资源管理和调度技术已无法满足现有及未来网络需求。
本文结合平行系统理论及下一代无线网络技术,提出了一种面向未来的网络架构,并在此基础上对无线资源管理和调度技术进行研究。主要工作如下:
首先,提出了基于平行理论的平行网络架构。首次将平行系统理论应用于新型网络架构设计中,不仅增强了对网络资源的管控,而且也提升了网络自优化性能。基于平行网络,建立适应于现代网络系统的网络模型和复杂认知网络动态计算实验平台,将网络仿真系统发展为具有灵活、便捷和广泛特性的人工网络系统,使得网络不仅在资源管理和控制上更便利,而且也为网络优化提供了支持,令其具有改善网络现状、预估网络状态的能力,从而更好地满足不断提高的网络系统需求。
第二,提出了基于平行网络的资源分配算法。该算法重新构建了基于平行网络的效用函数模型,使得网络提供给所有用户的带宽资源最优。除此之外,平行网络中的人工网络会存储所有网络状态信息及其对应的资源分配机制,基于用户行为的规律性,当同样的网络状态再次出现,控制中心可以直接从人工网络中获取最优资源分配机制。实验结果表明平行网络可以在优化网络资源分配的基础上,进一步降低网络运算复杂度和时延,提升网络灵敏性。
第三,提出了基于用户优先级的资源分配算法。基于实际网络中用户类型及业务的多样性,算法对业务进行分类及排序。当网络处于重负载状态时,网络优先提供给高优先级用户更多带宽,保障高优先级用户的QoS,当网络处于轻负载状态时,用户在分析各个可选基站的实时负载后确定接入基站,从而均衡不同网络之间的资源占用比。
为进一步提高网络资源利用率及用户入网概率,利用网络切换算法对网络资源进行二次优化配置。为了降低算法复杂度,提高实际可行性,采用Q-Learning 算法实现以上功能。


; With the rapid development of wireless communication technology, the demand for communication services is gradually changed from simple voice to complex mixed services (e.g. audio, radio, and so on), and the requirement of Quality of Service (QoS) is gradually improved. The next generation communication networks are envisioned to improve the spectrum efficiency and guarantee the QoS requirement. However, with the expanding network scale, continuously-innovation network business, and increasing network users, the wireless resource management and scheduling technologies based on current network architectures can hardly meet the network requirement.
Based on ACP theory and the technologies of next generation networks, we propose a novel network architecture and several resource allocation mechanisms based on the proposed architecture. The main contributions of the dissertation are as follows:
Firstly, the parallel network architecture based on parallel system theory is proposed. It not only enhances the ability of resource management, but also improves the network optimization performance. With the help of parallel network, we can model the actual network system and construct its corresponding artificial network systems, and make it more convenient to manage and control wireless network resource. Parallel network can revise and predict the network status, and make the real-time network conditions optimized.
Secondly, this thesis propose a novel strategy for resource allocation and access control based on parallel network. The utility function model is constructed, and the required user bandwidth is maximized. In addition, the artificial network can store the network status and the corresponding optimized solutions. As the user behaviors are regular, when an identical network status occurred, the control center can directly provide the optimized solutions without repeatedly executing the resource allocation algorithm.
Thirdly, this thesis propose the user priority-based network resource optimization mechanism. In the mechanism, the user priority is considered. When the network is heavily-loaded, the mechanism can decrease the bandwidth of a few users to make a new user access to this base station, and provide the high priority user a higher QoS in the same condition. When the network is lightly-loaded, the mechanism can provide all the users an optimal QoS and balance the base station load for the whole network system. To increase the network capacity for accommodating more users, the proposed mechanism can hand over a user from one heavily-loaded BS to the lightly-loaded ones. In addition, the Q-Learning algorithm is applied in the mechanism for its practical implementation.
关键词平行网络 资源分配 服务质量 网络优化
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/14822
专题毕业生_博士学位论文
作者单位中国科学院自动化研究所
第一作者单位中国科学院自动化研究所
推荐引用方式
GB/T 7714
杨坚. 基于平行网络的无线资源管理和分配技术研究[D]. 北京. 中国科学院研究生院,2017.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
基于平行网络的无线资源管理和分配技术研究(10885KB)学位论文 限制开放CC BY-NC-SA
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[杨坚]的文章
百度学术
百度学术中相似的文章
[杨坚]的文章
必应学术
必应学术中相似的文章
[杨坚]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。