Mobility Management And Packet Reordering Processing In FiWi Networks

Despite the good application prospect of hybrid Fiber-Wireless (FiWi) networks, there are still lots of design challenges to provide high quality communications (e.g., high bandwidth, low latency and seamless handover) for fixed/mobile users. Specifi-cally, to improve the quality of communication for mobile users, it is a critical issue to achieve seamless handover which means no or negligible degradation in service qual-ity during handover. To improve the quality of communication for fixed users, it is a challenging issue to increase the bandwidth since packet reordering may be generated by employing multi-path routing protocols.In this dissertation, we will investigate mobility management and packet reorder-ing issues to improve the quality of communication for both fixed and mobile users in FiWi networks. Specifically, we identify two problems as our research questions:(1) design of a novel handover framework to improve the handover performance and achieve seamless handover for mobile users,(2) design of packet reordering processing mechanisms to mitigate packet reordering and improve the TCP performance for fixed users.(1) Mobility management in FiWi networksAs a candidate wireless subnetwork of FiWi networks, Mobile WiMAX (MWiMAX) is a promising technology and also one of the key candidates for4G cellular networks. In MWiMAX networks, a fundamental issue is seamless handover, which has been investi-gated extensively in the literature. However, there is a lack of comprehensive handover solutions that integrate the design of MWiMAX network, backhaul technology, and Layer-3(i.e., IP) mobility management. This dissertation proposes a novel handover framework, which integrates a hybrid Passive Optical Network (PON) that enables ef-ficient multicast, localized mobility management, and direct inter-base-station commu-nication; the framework supports both hard handover and soft handover in MWiMAX: and the framework takes advantages of Mobile IP technologies and Media Independent Handover (MIH) services of IEEE802.21. Within the proposed framework, we de-velop an efficient mobility management protocol (MIH-FHMIPv6) for hard handover, and also design an efficient packet distribution algorithm for soft handover. The MIH-FHMIPv6mobility management protocol reduces handover latency and signaling over-head in the integrated architecture by combining the benefits of FMIPv6and HMIPv6, and employing the IEEE802.21MIH services; avoids potential service degradation by eliminating both packet loss and packet reordering. The packet distribution algorithm reduces the amount of packet duplication by exploiting the optical multicasting abili-ty of the hybrid PON backhaul, and achieves the perfect data content synchronization by utilizing the centralized control ability of the tree topology. Extensive performance analysis and numerical results show that (1) our mobility management protocol can significantly improve handover performance in terms of handover latency, signaling overhead, packet loss and packet reordering, and (2) the proposed packet distribution algorithm can effectively reduce the amount of packet duplication.(2) Packet reordering processing in FiWi networksIn a FiWi network, multi-path routing may be applied in the wireless subnetwork to improve bandwidth. Due to different delays along multiple paths, packets may ar-rive at the destination out-of-order, which may cause TCP performance degradation. In this dissertation, we propose three mechanisms to improve in-order departure of pack-ets from the OLT in a FiWi network and enhance the TCP performance. Firstly, we propose an effective scheduling algorithm (SPCS) at the OLT, which gives lower pri-orities to the packets which may potentially cause duplicate ACK (dupack) and reduce the congestion window size (cwnd). Secondly, we design a DBA scheme (IPACT-MPR) in the optical subnetwork (i.e., an EPON) which gives higher priorities to flows which may trigger TCP’s fast retransmit and fast recovery algorithms in upstream bandwidth allocation. Finally, we propose an integrated flow assignment and fast resequencing approach which jointly determines the probability of sending packets along each path from the source and needs virtually zero resequencing delay at the OLT to reduce the out-of-order probability when packets are injected to the Internet from FiWi networks. Simulation results show that the proposed three mechanisms are effective in mitigating packet reordering and improving the TCP performance.