利用信道学习获取超蜂窝网络休眠基站的信道信息

发布时间：2018-10-30 14:28

【摘要】：信道信息作为刻画无线信道状态的重要信息,在新一代基于大规模多天线技术的系统中扮演了极其重要的角色.而在超蜂窝架构中,业务基站需要动态休眠以提高系统的能效.当基站休眠时,传统的基于导频的信道信息获取方法无法使用.本文针对这一问题提出了基于机器学习的信道信息获取策略,通过活跃基站多维度的信道信息来推测休眠基站的信道信息.我们以用户选择业务基站为例,利用人工神经网络架构设计了一套具体的信道学习解决方案.与位置辅助信道估计方案不同,该方案采用控制基站侧的信道信息作为输入,避免了用户定位过程中存在的一些问题.通过基于随机地理散射物的信道模型,我们对该方案进行了仿真验证,发现其预测准确率优于K近邻等其他预测算法,性能接近于已知准确用户地理信息的位置辅助估计方案.
[Abstract]:As an important information to describe the state of wireless channel, channel information plays an extremely important role in the new generation of systems based on large-scale multi-antenna technology. In the supercellular architecture, the service base station needs to sleep dynamically to improve the energy efficiency of the system. When the base station is dormant, the traditional channel information acquisition method based on pilot frequency can not be used. In order to solve this problem, a new channel information acquisition strategy based on machine learning is proposed in this paper. The channel information of dormant base stations can be inferred by the channel information of active base stations. Taking the user selected service base station as an example, we design a specific channel learning solution using artificial neural network architecture. Different from the position-aided channel estimation scheme, the channel information of the control base station is used as input in this scheme, which avoids some problems in the process of user location. Through the channel model based on random geographical scattering, the simulation results show that the prediction accuracy of this scheme is better than that of other prediction algorithms such as K-nearest neighbor, and the performance is close to that of the location aided estimation scheme with known accurate geographic information.
【作者单位】：清华大学电子工程系;清华大学信息科学与技术国家实验室(筹);
【基金】：国家重点基础研究发展计划(973)(批准号:2012CB316000) 国家自然科学基金(批准号:61461136004,61571265,91638204)资助项目
【分类号】：TN929.5

【相似文献】