A ubiquitous application needs quality of service (QoS) guarantee to provide the required services to its users efficiently without any user intervention or requests. The ubiquitous applications demand QoS as per the current environment context. Providing QoS in a ubiquitous network is a challenging task as the network consists of several subnetworks of different technologies and controls. In this paper, we discuss a QoS architecture framework, which guarantees the promised QoS requirements of an application. The proposed architecture uses agents to predict the QoS requirements of the applications running on a node (called unode) by using the context and profile information of the user, and the history of the unode and the subnetwork. The framework communicates with the local administration of a subnetwork to allocate resources to the ubiquitous applications running on the unodes. We have simulated the proposed framework in a ubiquitous network consisting of several subnetworks. We have also modeled the QoS architecture to analyze the average service time of different applications running on the unodes and to examine the effect of background load of the local applications on the unodes.