Researches On Discrete-Time Two-level-polling System With Exhaustive And Limited (K=1) Service

In the late 1950s, the polling technique was first used by the British cotton industry for investigation and repairing of machines, and it turned out that the polling technique had great potential in many industrial fields. In the 1960s, the research of the polling technique achieved significant progress and it evolved into the theoretical and systematic stage. In particular, researchers used polling models to characterize the process of equipment fault detection, industrial process control, multiple access control, resource arbitration and load sharing, etc. The performance was analyzed using the probability theory, queuing theory and stochastic process. This made polling technique an important model for various applications. In the past decades, polling system provides useful mathematical models to analyze the performance of industrial automation and control, time-sharing system, communication and computer networks. The objective in analyzing polling models is to set up mathematical functions of controlling mechanism, and derive closed form expressions to obtain the system parameters, such as the mean queue length (MQL), the mean cyclic period (MCP) and the mean waiting time (MWT). Among them, the MWT is the most important performance measure for analyzing the underlying system. In recent years, polling models have been widely used to analyze the performance of a variety of systems. Especially, it is used to provide performance evaluation criteria for demand-based, multiple-access schemes in communication and computer networks. In the last 60 years, research indicates that polling system model has become an effective tool for analyzing system performance, and it provides a theoretical framework for resource allocation and assignment. Thanks to its impartiality, flexibility and practicality, polling systems have been widely used to analyze the performance of a variety of systems, from early equipment fault detection to transportation, production flows control, communication networks, computer networks, wireless sensor networks, Ad Hoc networks and social resources allocation. In all of these fields, the polling models have considerably advanced the underlying technology. Nowadays, the research for polling system and theory is still expanding with emerging polling systems, analysis approaches and applications. With coming of 21st century, the research of mobility, self-organization, high efficiency and energy saving for wireless communication becomes hot area, and the up-to-date as well efficient MAC protocol becomes an important subject.Given polling technique's unique capability in access control and resource allocation in communication and computer networks, this dissertation focuses on priority control and resource allocation in Broadband Wireless Access (BWA) Network Systems and Wireless Sensor Networks. The theory and technique for achieving real-time implementation, impartiality, priority differentiation and QoS guarantee are investigated as follows. First, based on the analysis of the traditional polling multiple access systems, a novel and efficient model of Discrete-Time Two-level-polling System with Exhaustive and Limited (K=1) Service and its extended model are proposed. Second, the embedded Markov Chain Theory and the Probability Generating Function Methodology are employed to build the mathematical model. The parameters of the model are obtained using closed form expressions. Particularly, this involves probability theory, queuing theory, stochastic process, theory of polling control, Broadband Wireless Access Network Systems as well as computer networks. Finally, the aforementioned mathematical model is applied to the WLAN (Wireless Local Area Network) and WSN (Wireless Sensor Networks). The analytical results of their MAC (Media Access Control Protocol) protocols are verified through extensive computer simulations. The performance results clearly show that our proposed two-level-polling scheme with mixed services can achieve the differentiation for real-time and priority services (such as voice, video et al.), and it leads to higher performance in terms of better impartiality, flexibility and real time transmission of information packets for their MAC protocols. From this point of view, the dissertation effectively expands the research and applications of the polling systems.The dissertation consists of 6 parts, and it is organized as follows:Chapter 1 introduces the concept and the state-of-the-art research for polling systems, and it also presents the background of the dissertation.Chapter 2 analyses the Gated, Exhaustive and Limited (K=1) Service Polling Systems using the embedded Markov Chain Theory, the Probability Generating Function and System State Equations methodology, and gives performance evaluation. These results are further used in the following chapters.Chapter 3 deals with 4 topics. The first topic proposes a novel and efficient model of Discrete-time Two-level-polling System with Exhaustive and Limited (K=1) Service, and analyses its performance using the embedded Markov Chain Theory, the Probability Generating Function and System State Equations methodology. The second topic extends the proposed polling model by considering the arrival process of Geom/G/1. Performance is analyzed in a similar way as in the first topic. The third topic presents the set up of the experiments, and the analytical results are verified through extensive computer simulations. The forth topic focuses on performance evaluation of two types of polling systems (ZY Polling System Model and ZL Polling System Model) proposed by us.Chapter 4 first introduces the development of WBAN (Wireless Body Area Network), WPAN (Wireless Personal Area Network), WLAN (Wireless Local Area Network), WMAN (Wireless Metropolitan Area Network), WWAN (Wireless Wide Area Network) and Ad Hoc, and then the proposed polling model is used to improve the performance of IEEE 802.11 PCF MAC protocol.After presenting an overview of the development, research and application for WSN in Chapter 5, the proposed polling model is employed to improve the function of Clustering Algorithm for WSN.Chapter 6 draws the conclusion of the dissertation, and it also points out the future trends of the polling systems and puts forward some challenges.