QoS Provisioning For Secondary Users In Cognitive Radio Networks

With the rapid increasing of wireless services, the contradiction between the growing de-mand for spectrum resources and the low spectral efficiency resulted by the current fixed spec-trum allocation method is becoming more and more serious. To address this issue, in recent years, cognitive radio technology has drawn more and more attentions. With the help of the cognitive radio technology, secondary users (unlicensed users) are able to obtain the spectrum usage condition around through spectrum sensing and can opportunistically access the spectra while not disturbing the normal usage of primary users (licensed users), which can increase the spectral efficiency tremendously. Most majority of the existing research works focus on the QoS provisioning for primary users. However, secondary users’QoS provisioning is also necessary, since it is the premise for cognitive radio technology to be put into practical use. In this disserta-tion, the QoS provisioning of secondary users in cognitive radio networks (CRNs) is thoroughly studied on packet level, session level and end-to-end level on the transport layer, respectively.The queueing analysis and its application for multi-user multi-channel CRNs are studied. The impacts of several important internal factors (transmission mechanisms and parameters of CRNs) and external factors (the radio environment around) on queueing dynamics of secondary users are investigated. An analytical framework for queueing dynamics of CRNs is established, with the following transmission mechanisms being considered:spectrum sensing (especially the impacts of spectrum sensing errors on the queue), medium access control (MAC) scheme and link adaptation technologies (including automatic repeat request and adaptive modulation and coding). The queueing dynamics are modeled as a finite state Markov chain (FSMC), and the analytical expressions of some key parameters are derived including average queue length, packet dropping rate and packet collision rate. Then, the key QoS parameters, such as average queueing delay, packet loss rate and effective throughput, are expressed analytically. These expressions reveal both the way and the degree of the influences of the aforementioned internal and external factors on the QoS of secondary user. The application of the proposed analytical framework is demonstrated with a case study on TCP throughput maximization through adjusting the lower-layer transmission mechanisms and parameters appropriately.The call admission control (CAC) in CRNs is investigated. First, a general framework is established to analyze CAC schemes in CRNs, taking channel reservation, buffers for both newly arrival secondary users and handoff secondary users, and departure of the queued secondary users caused by impatience into account. Hence, almost all the existing works about CAC for CRNs can be viewed as special cases of this framework. Using this framework, the impacts of buffer sizes, number of reserved channels, and impatience degree of queued secondary users on the session level QoS are studied. Second, due to the high dynamics of the network resource, CAC parameters should be adjusted in time, which leads to a high computational complexity. To handle this problem, a low-complexity calculation method is proposed, through which the optimal number of reserved channels to minimize the dropping probability under the constraint on blocking probability can be obtained analytically. This method helps with finding the optimal CAC parameter according to the real-time condition of the network resource.The method of modeling and improving TCP performance on the transport layer in CRNs is studied. It is pointed out in this dissertation that there is a complicated interaction in CRNs between the back-off operation of TCP and the transmission mechanisms in lower layers. On one hand, a long-time unavailability of channels may trigger TCP back-off. On the other hand, during the TCP back-off, if the lower-layer buffer is empty, transmission opportunity cannot be used since there is no packet to send, which results in the wastage of transmission opportunity. In this dissertation, this interaction is analyzed, through which the expression of the effective data transmission time of secondary users is obtained. Then, a new TCP throughput model for CRNs is obtained by modifying the traditional TCP throughput model. As a demonstration of using this proposed model, two optimization problems are formulated to maximize the TCP throughput and the energy utility respectively, through appropriate configuration of lower-layer parameters. The model and methods can be employed in system design and cross-layer optimization for QoS provisioning of secondary users.