The increasing growth of wireless access networks, proliferation of the Internet and gradual deployment of broadband networks has already given birth to a set of information-centric applications based on data transmission. Efficient scheduling techniques are necessary to endow these applications with advanced data processing capability. Broadly all data transmission applications are divided into (1) push and (2) pull systems. Hybrid scheduling, resulting from an effcient combination of these two types of data delivery, often exploits the advantages of both the schemes. The objective of this dissertation is to investigate and develop a novel hybrid scheduling platform by effectively combining broadcasting (push) of popular data and dissemination (pull) of less popular data. One major advantage of this algorithm is dynamic computation of cut-of-point, used to segregate the popular and less-popular data items, without any prior knowledge or assumptions. In order to achieve a better performance, the framework is enhanced to allow a set of consecutive push and pull operations, depending on the probabilities of the data items present in the system. The framework also incorporates practical issues like clients' impatience leading to clients' departure and transmission of spurious requests. A new client's priority-based service classification scheme is proposed to provide differentiated QoS in wireless data networks. The framework proceeds further to incorporate dynamic hybrid scheduling over multiple channels. Performance modeling, analysis and simulation study points out efficiency of the entire framework.
展开▼
