中文摘要：

Resource reservation is an effective measure to ensure end-to-end quality of service(QoS),however,the burstyness of the traffic makes the reservation idle some time,and forms a waste of resources.Based on the analysis of active queue management(AQM)of DiffServ network,we propose a resource management scheme,which allows borrowing resources from unused reservation,according to the characteristics and advantages of cognitive networks.First,some nodes reserve certain proportion capacity for some special services(for instance,some services pay additional money)to guarantee the priority of these applications.Then resources are assigned according to the different parameters of services.If the available resource can not meet the requirements of new services,realtime ones are admitted with higher priority and allow borrowing the unused reservation from other nodes appropriately.Simulations show that,the proposed scheme has good performance at network resource utilization,the admission rate of new applications and QoS of users.

英文摘要：

Resource reservation is an effective measure to ensure end-to-end quality of service （QoS）, however, the burstyness of the traffic makes the reservation idle some time, and forms a waste of re- sources. Based on the analysis of active queue management （AQM） of DiffServ network, we propose a resource management scheme, which allows borrowing resources from unused reservation, accord- ing to the characteristics and advantages of cognitive networks. First, some nodes reserve certain proportion capacity for some special services （for instance, some services pay additional money） to guarantee the priority of these applications. Then resources are assigned according to the different parameters of services. If the available resource can not meet the requirements of new services, real- time ones are admitted with higher priority and allow borrowing the unused reservation from other nodes appropriately. Simulations show that, the proposed scheme has good performance at network resource utilization, the admission rate of new aDolications and OoS of users.