Enhancement of Spectrum Decision through Probabilistic Graphical Models

这项研究的目标是提高消费者的服务质量的环境中，越来越多的服务竞争使用的无线电频谱。 人们普遍认为，未来对频谱的监管将从目前的静态环境（打开新的通信信道需要数年时间）转变为高度动态的环境（智能无线电将从一个频道跳到另一个频道）。 在这种动态环境中，消费者将与许多不同类型的设备共享频谱，包括雷达、其他手机、Wi-Fi服务器和许多新的智能设备。新的智能设备将不断改变它们的频率、数据速率和调制方案，从而创造一个动态的、更不确定的环境。科学家们将探索不确定性如何影响无线电频谱使用的特征。重点是问题的感知和决策方面，而不是管理问题。 科学家们将运用他们在人工智能方面的经验，并从其他领域转移动态问题解决技术。拟议的研究有可能开发决策支持模型，为未来的频谱管理提供新的政策。本研究项目的主要目的是通过评估不确定性对自适应无线电不同认知周期变化的影响，提高对无线通信场景的认识和理解，以丰富认知无线电中的频谱决策过程。为了实现这一目标，他们建议识别，分类和表征典型无线场景中存在的随机和确定性变量，并使用概率图形模型（如贝叶斯网络和影响图）对其因果关系进行建模。

The goal of this research is increase the consumer's quality of service in environments where more and more services are competing for use of the radio spectrum. It is widely believed that future regulation of the spectrum will change from the current static environment, where opening new channels for communications takes years, to highly dynamic environment where smart radios will jump from frequency channel to frequency channel. In this dynamic environment, the consumer will share the spectrum with many different types of devices, including radars, other phones, Wi-Fi servers and a host of new smart devices. The new smart devices will constantly change their frequencies, data rates and modulation schemes, thereby creating a dynamic, more uncertain, environment. The scientists will explore how uncertainty influences the characterization of radio spectrum usage. The focus is on the sensing and decision-making aspects of the problem rather than management issues. The scientists will apply their experience with Artificial Intelligence and transfer techniques of dynamic problem solving from other domains. The proposed study has the potential to develop decision support models that will inform new policies for spectrum management in the future. This research project's primary aim is to advance the knowledge and understanding of wireless communication scenarios to enrich the spectrum decision process in cognitive radio by evaluating the impact of uncertainty on the different cognitive cycle changes of adaptive radios. To achieve this goal, they propose to identify, classify, and characterize the random and deterministic variables present in a typical wireless scenario and model their causal relations, using probabilistic graphical models, such as Bayesian networks and influence diagrams.