提出了一种基于案例推理的最优策略产生方法,用于异构无线网络系统自主高效的无线资源管理及优化.该方法利用案例检索与匹配算法,从案例库中寻找最合适的再用策略;利用基于相似度概率的策略再用算法,改进新策略学习的效率和质量;应用贪婪算法,产生没有可用案例情况下的应对策略.此外,该方法在学习的基础上对策略案例库进行实时的更新.仿真结果表明,该算法具有高效的在线学习能力,能够有效提升网络在频谱效用和阻塞率方面的性能,实现自主的无线资源管理.
A case-based reasoning approach for generating the optimal policy is presented,which can be used for radio resource management and optimization in an autonomous and efficient mode in heterogeneous wireless networks.It searches for the most appropriate reuse policy in the case library by utilizing case retrieval and matching algorithms,and improves the new policy's efficiency and quality of learning by utilizing the policy reusing algorithm based on the probability of similarity.The approach also introduces the greedy algorithm for generating coping policies when there is no case available.In addition,it updates in real time the policy case library based on learning.Simulation results show that the proposed approach owns efficient online learning ability,and can effectively improve the network blocking rate together with spectrum utility performance.It achieves the autonomous management of radio resources.