Resource Reservation and Management in Mobile Environments

• Main Authors: Gwo-Chuan Lee, 李國川
• Other Authors: Chien-Chao Tseng
• Format: Others
• Language: en_US
• Published: 2002
• Online Access: http://ndltd.ncl.edu.tw/handle/50689994564767983447

博士 === 國立交通大學 === 資訊工程系 === 90 === With the advances in wireless communication technology, a demand for the wireless Internet to support real-time multimedia applications has arisen. We have seen a rapidly growing demand for mobile networks to provide Quality-of Services (QoS) to mobile hosts. Resource reservation protocol (RSVP) is a proposal to guarantee the QoS for fixed hosts on the Internet. Mobile RSVP is proposed for adapting RSVP for mobile environments. However, Mobile RSVP might waste too much bandwidth. We also observe that regional networks do exist in the Internet. Regional networks can be classified into regions with normal handoff delay and large handoff delay. A region might be a Multi-Home network served by various Internet Service Providers. However, Mobile RSVP cannot effectively support the QoS-guaranteed services for mobile hosts in the regional networks. In this dissertation we explore several proposals to make advance resource reservations for guaranteeing mobile host QoS in various regional networks in the wireless Internet. First we propose a Hierarchical Mobile RSVP (HMRSVP) that can achieve mobility independent QoS-guaranteed services for regions with normal handoff delay in mobile environments. In a region with normal handoff delay, the setup time for making resource reservations is normally short for an intra-region handoff. The proposed HMRSVP integrates RSVP with Mobile IP regional registration and makes advance resource reservations only when a mobile host tends to perform an inter-region handoff. We will show that HMRSVP can achieve the same QoS guarantees as MRSVP with fewer resource reservations. Further, we show that HMRSVP outperforms MRSVP in terms of reservation blocking, forced termination and session completion probabilities. Secondly, we propose a pointer forwarding scheme to make advance resource reservations for regions with large handoff delay in mobile environments. In a region with large handoff delay, the setup time for making resource reservations might be long for an intra-region handoff. The pointer forwarding scheme could degrade both the setup time and reservation cost in making advance resource reservations. In other words, the pointer forwarding scheme could improve MRSVP performance significantly if the advance resource reservations consume too many resources. Our scheme performs well when mobile hosts move locally in a region with large handoff delay. In addition, we consider the resource reservations for mobile hosts in a Multi-Home regional network. We found that the pointer forwarding scheme could provide seamless handoffs and improve QoS guarantees when mobile hosts perform inter-home domain handoffs in a Multi-Home regional network. Due to the trends in large area mobility and the increasing population utilizing personal communication services (PCS) networks, the efficiency of mobile communication management schemes is more important than before. Moreover, efficient mobility management for global mobility services is essential for future PCS networks. For the resource management of distributed HLR databases in PCS networks, we propose a new location tracking method for pointer forwarding with distributed HLRs to decrease the location tracking cost. Unlike the traditional distributed HLR schemes, we propagate the most recent location changes from an HLR to reset the forwarding pointers in other HLRs, and thus reduce the location tracking cost for call deliveries. We allow some inconsistency to exist among the HLR location data and defer the propagation of location changes to reduce the propagation cost. For the global mobility services in future PCS networks, we propose a HLR replication protocol, multiple-replica (MR) strategy, to improve the performance of the global mobility management for distributed HLRs. The MR strategy decides whether a profile should be replicated on the HLR of a service area. The replication of user profiles in the MR strategy is dynamic and partial. To evaluate the MR performance we modeled the MR strategy and compared it with the single-replica (SR) strategy.