| Summary: | 博士 === 國立成功大學 === 電腦與通信工程研究所 === 101 === This dissertation addresses three issues for next-generation broadband access networks which include Wireless Broadband Access Networks (Wireless-BANs) and Wired Broadband Access Networks (Wired-BANs).
Wireless-BANs, such as LTE and WiMAX, are inherently lossy due to the unreliability of the wireless medium. Although the Hybrid Automatic Repeat reQuest (HARQ) error-control method recovers from packet loss, it has low transmission efficiency and is unsuitable for delay-sensitive applications. Alternatively, network coding techniques improve the throughput of wireless networks, but incur significant overhead and ignore network constraints such as Medium Access Control (MAC) layer transmission opportunities and physical (PHY) layer channel conditions. This dissertation develops an analytical model of Random Network Coding (RNC) and Systematic Network Coding (SNC) decoding probabilities. Based on the results, SNC is selected for developing an adaptive network coding scheme designated as Frame-by-frame Adaptive Systematic Network Coding (FASNC). According to the per frame network constraints, FASNC significantly improves transmission efficiency and reduces the decoding delay compared to existing works.
Relaying provides enhanced system throughput and coverage of emerging Wireless-BANs as described in next generation broadband wireless standards (e.g., LTE-advanced and IEEE 802.16j). Allocating of network resources in an efficient and effective manner still remains a challenging issue. Accordingly, this dissertation proposes two efficient heuristic algorithms, namely Relay Resource Scheduling (RRS) and Adaptive Relay Resource Scheduling (ARRS), to maximize system utility. ARRS enhances upon the performance of RRS by adaptively setting the split ratio between the access zone and the transparent zone in the downlink subframe. Computational complexities of both algorithms are shown to be polynomial and proportional to the number of links and subchannels in the network. Compared to other algorithms, the proposed ARRS algorithm can achieve better system performance with comparable computation complexity.
For Wired-BAN, Ethernet Passive Optical Networks (EPONs) alone are not a good fit since they are cost-prohibitive for remote or rural areas. In order to address this issue, integrated EPON/WiMAX networks are a promising solution proposed in current literature. However, existing bandwidth allocation schemes for integrated EPON/WiMAX networks neglect the interactions between the self-interested EPON and WiMAX service providers. Accordingly, this dissertation presents a two-stage game-theoretic framework for the intra-ONU bandwidth allocation process in which the interactions between the EPON and WiMAX service providers are taken into account. This two-stage framework is capable of resolving the resource sharing problem for integrated EPON/WiMAX Networks by maximizing profit of the EPON and WiMAX service providers, supporting respective differentiated services, and guaranteeing proportional fairness for different traffic classes.
|
|---|
