Study On Optical Network Resource Allocation Strategies Based On Game Theories

The convergence of Information Technology (IT) and Communication Technology (CT) is the most conspicuous tendency of the industry development nowadays. Under this convergent condition, unified smart pipe systems could be used to carry various services. However, the transmission quality requirements of different services are different so that when various traffics share the network resources, the infrastructure resource allocation and control strategies will be of great significance to the customer experiences and satisfaction as well as the whole system resource optimization. On the other hand, as the network, which bears the convergent services, faces the huge amount of end users, the big scope of users and services form big challenges to the pricing system, therefore, the pricing strategies of the network resource is also a hot research theme.This thesis is focused on the intelligent control strategies of the bandwidth resource allocation respectively in the three parts of the fiber optics transmission network, namely fiber access network with EPON (Ethernet Passive Optical Network) technologies, optical metro network with PTN (Packet Transport Network) technologies and optical backbone network with WDM (Wavelength Division Multiplex) technologies. The research adopted the game theories and used the price signals as incentives for the automatic systems. The main contents are as follows:1. The bandwidth sharing of the EPON upstream link under auction mechanism. The efficient time slots allocation of the time division multiplexing in the next generation multiple services broadband access network is analyzed via auction mechanism in the EPON upstream capacity sharing. The OLT (Optical Line Terminal) and ONU (Optical Network Unit) constitutes the Vickrey auction gaming model, which realizes the economic incentive efficiency based on arbitration. These economic incentives guarantee that the traffics are transmitted according to different service values and quality levels. This research proved that the Nash equilibrium exists in the auction. The simulation verifies that under this auction environment, the time slots resources can be preferentially assigned to those ONUs with higher level of quality requirements.2. The distributed bandwidth allocation in PTN network based on QoS (Quality of Service). The distributed bandwidth resources allocation mechanism supporting various QoS levels is analyzed in the packet transport network with PW (Persudo Wire) and LSP (Label Switching Path) structure. The users and network provider are set in a Stackelberg gaming model. The comparison simulation verified that the bandwidth allocation and pricing mechanism based on Stackelberg game model could provide better QoS for the higher grade traffics and obtain higher revenues than the fixed rate pricing mechanism does.3. The stability analysis of the gaming equilibrium in the hybrid optical backbone network based on WDM technologies. WDM network platform could not only carry multiple services types but also reasonably allocate network resources to different sub-networks by dynamic wavelengths assignments according to the changes of the traffics with different QoS requirements. The stability analysis of the gaming equilibrium between the sub-network provider and users is made based on Lyapunov theory. It is proved that the gaming equilibrium is asymptotic stable so that the theoretical optimality could be obtained in the system. Based on the gaming stability analysis, it is suggested that the bandwidth resource allocation between different sub-networks be relative stable within certain period. In this way, the properties of the Paris Metro Pricing could be achieved and stable Nash equilibrium could be reached.In this thesis, the game theory and network resource allocation are combined for the research. From the perspective of system equilibrium, the economic efficiency is investigated in the resource allocation, which practically reflects the benefits coordination between the network provider and users. The research results might be applied in the intelligent control of network resource management based on QoS.