基于分级代理的智能家庭网络模型研究

Research on intelligent home network model based on multi-level agents

针对现有家庭网络中智能设备不能有效学习家庭用户习惯,致使不能满足用户的个性化服务质量的问题,给出了一种基于分级代理的智能家庭网络模型。首先给出了一种智能家庭网络设备的形式化描述,以此为基础提出了分级代理的智能家庭网络模型:全局Agent通过对家庭中的长期数据进行学习,总结出一定的服务规则,指导设备A^et根据家庭成员的生活习惯改变工作方式;设备Agent利用强化学习算法,自主学习,解读家庭环境的状态变化,并做出最优选择。该模型的应用实例及仿真结果表明,应用此模型,设备可以学习用户习惯,为用户提供个性化服务。

This paper presents an intelligent home network model based on multi-level agents to solve the problem that the existing intelligent devices in home networks can not study users＇ habits efficiently so that they can not satisfy users＇ personalized QoS. A formal description for devices in intelligent home networks is given before the model is established. The overall agent in an intelligent home network summarizes the rules of service through the study of the long-term data of the family, which guide the work of equipment agents according to the life habits of the family. The equipment agents learn and detect the change of the family environments automatically and make the optimal choices using the reinforcement learning algorithms. An application scene of the multi-level agent model and the simulation result show that, by applying this model, the devices can study the habits of the user and provide personalized service to the user.