不确定环境中基于学习的绿色动态谱访问序列决策机制研究

In the undeterministic and dynamic wireless communication environment, the undeterministic uncomplete unprecise information obtained from detecting spectrum would definitely affect the policy of allocating channel of the unliciensed users. If one unliciensed user detects all the linciensed channels and visits the best among them, a large amount of time and energy would be spent and also the spectrum efficiency as well as energy efficiengy would be degraded, which cannot meet the requirement of green communications. Therefore, the key point of this project is to study the robust mechanism of sequential decision of detecting spectrum, and to obtain the available information of dynamic spectrum and its evolving rule under the undeterministic environment; to investigate the mechanism of sequential decision of allocating spectrum under the dynamic environment depending on the obtained spectrum information,history information of allocating spectrum, the features of coexisting networks and information of spending energy; to discover the relation of energy efficiency between spectrum-detecting and spectrum-access, and jointly optimize the spectrum efficiency and energy efficiency; next to study the effect of joint optimization of sepectrum efficiency and energy efficiency on sequential decision; finally by simulation of computer and hardware platform, to evaluate and verify the performance of the schemes proposed in this project. Through this research project,we hope to discover the tradeoff between spectrum efficiency and energy efficiency from the viewpoint of sequential decision so as to provide some theoretical insights on green opportunistic spectrum access under the dynamic environment.

在不确定的频谱环境中，频谱检测获得的不确定不完整不准确的频谱信息必然对非授权用户的信道分配策略造成影响。如果非授权用户检测所有的授权信道并访问其中最好的信道，必然耗费大量时间和能量，降低系统的频谱效率和能量效率，不能适应绿色通信的要求。因此，本项目重点研究强鲁棒性的频谱探测序列决策机制，获得不确定环境中动态频谱的可用性信息及变化规律；依据获得的频谱信息、信道分配历史信息、共存网络属性及能耗信息，研究动态环境下能量有效的频谱分配序列决策机制；在此基础上，揭示频谱感知和频谱接入在能量效率方面的相关性，并对频谱效率和能耗效率进行联合优化；然后针对分布式机会频谱访问，研究频谱效率和能耗效率联合优化对序列决策的影响；最后通过软件仿真和硬件验证平台对提出的决策方案进行性能评估和验证。通过该研究可望从序列决策的角度揭示频谱效率和能量效率之间的折中关系，为复杂动态频谱环境下绿色机会谱通信提供理论上的指导。

无线电频谱是一种宝贵的资源，在通信业迅速发展的今天,频谱资源贫乏问题日益严重。认知无线电技术作为一种解决不断增长的频谱需求以及提高频谱利用率的最有潜力的技术应运而生。非授权用户通过感知周围频谱的使用情况,在不影响授权用户通信质量的前提下，机会地使用授权频段，从而极大地提高频谱资源利用率。然而，在不确定的频谱环境中，频谱检测获得的不确定不完整不准确的频谱信息必然对非授权用户的信道分配策略造成影响。如果非授权用户检测所有的授权信道并访问其中最好的信道，必然耗费大量时间和能量，降低系统的频谱效率和能量效率，不能适应绿色通信 的要求。因此，本项目重点研究了强鲁棒性的频谱探测序列决策机制，获得不确定环境中动态频谱的可用性信息及变化规律。依据获得的频谱信息、信道分配历史信息、共存网络属性及能耗信息，研究了动态环境下能量有效的频谱分配序列决策机制。在此基础上，研究揭示频谱感知和频谱接入在能量效率方面的相关性，然后对频谱效率和能耗效率进行联合优化。最后通过软件仿真对提出的决策方案进行性能评估和验证。该研究从序列决策的角度揭示频谱效率和能量效率之间的折中关系，为复杂动态频谱环境下绿色机会谱通信提供理论上的指导。