Studies of Contention Resolution and Bandwidth Allocation Algorithms over DOCSIS HFC Networks

• Main Authors: Chih-Cheng Lo, 羅志成
• Other Authors: 陳文賢
• Format: Others
• Language: en_US
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/98235789959012923861

博士 === 國立中興大學 === 資訊科學系所 === 95 === As broadband multimedia services becoming popular in the past few years, several emerging wired and wireless advanced access technologies have been studied to provide broadband access to subscribers. The Community Antenna Television (CATV) networks have become ubiquitous. Instead of constructing an entirely new broadband network infrastructure, the CATV network has emerged as one of the rapid and economic technologies to converge heterogeneous networks to provide broadband access to subscribers. The HFC network possesses a tree-and-branch architecture and highly asymmetric bandwidth, the upstream bandwidth is rarely and deeply influences the CATV network operation. How to support real-time multimedia applications, especially in an environment with heavy traffic, is a critical issue in modern CATV networks. In this thesis, we propose to improve the upstream bandwidth allocation, contention resolution algorithm, and availability of DOCSIS HFC networks to fulfill the needs of real-time multimedia services. We first propose a Predictive Slot Allocation with Dynamic Contention Resolution Algorithm to predict the number of contention slots in order to better cope with the request contention problem and achieve better overall system performance. It is composed of two mechanisms: Adaptive Contention Slots with Statistics Control Algorithm to improve upstream throughput and Predictive P-Persistent with Dynamic Back-off Algorithm to decrease access delay. To avoid the inherently unpredictable access delay of truncated binary exponential back-off algorithm adopted in DOCSIS, we propose a novel scheme named Adaptive Fast Expansion, Loading Statistics with Dynamic Swapping Algorithm to support real-time interactive services over HFC networks. It consists of two sub-algorithms: Adaptive Fast Expansion with Coordinate Centralized Control Algorithm, and Loading Statistics with Dynamic Swapping Algorithm. In addition to access delay and performance, as HFC networks is to become an infrastructure of broadband service networks, another important issue is the ability of the MAC layer to recover from service breakdown. Finally, we propose a Priority Ranging Scheme, to minimize service disruption time due to a large area malfunction, to enhance availability of DOCSIS HFC networks. Through simulation results, the proposed algorithms have been shown to perform better than that of the MCNS DOCSIS.