| The LEO (Low Earth Orbit) satellite constellation has several advantages, such as lower orbit latitude, reasonable latency, efficient frequency reuse and simple terminal power requirements. Due to low orbit latitude, single LEO satellite generally has small footprint which is defined as the servicing area of a corresponding satellite on the earth. To provide global coverage, a greater number of satellites are needed to work together. In other words, these satellites constitute LEO satellite network to achieve global coverage. By doing so, the characteristic of single satellite coverage deficiency is overcome. To satisfy the communication requirements of those users within different footprints, these satellites should have the ability to exchange information, and independently to forward data. As the important part of satellite network technology, the dynamic routing onboard is very critical to improve the timeliness and reliability of data transmission. Therefore, it is very necessary to undertake the research on LEO satellite network dynamic routing technology.This dissertation provides a comprehensive overview of the satellite network routing. Also, current satellite network protocol simulation tool is elaborated. Afterwards, the dissertation applies agent technology to LEO satellite onboard routing problem, which focuses on load balancing routing, QoS routing, multi-service routing and satellite network simulation tool. The main contributions are shown as follows.(1) Considering the uniform distribution of traffic in LEO satellite network, ALBR (Agent-based Load Balancing Routing) for LEO satellite network is proposed. Firstly, satellite network traffic load model for hot spot areas is presented based on the characteristics of traffic distribution on the earth surface. Secondly, two cost modification factors are proposed considering the traffic model and the characteristics of the polar orbit satellite constellation, named Inter-Plane ISL(Inter-Satellite Link) cost modification factor and Intra-Plane ISL cost modification factor, irrespectively. Afterwards, the path cost computing strategy of ALBR is given. Thirdly, the dissertation also designs agent-based satellite network routing architecture and elaborates ALBR algorithm. In ALBR, mobile agents cooperate with stationary agents to explore the routing path, gather routing information and update the routing table. Through complexity analysis and simulation, ALBR achieves better load balancing, and has lower onboard computation, storage and signaling requirements than other on-board routing schemes.(2) Focusing on poor QoS performance because of handover, AQR (Agent-based QoS Routing) for LEO satellite network is presented. Considering call duration, UDL and ISL service time, ISL available probability is presented. Afterwards, the weighted path cost is evaluated using ISL available probability and ISL cost. In AQR, mobile agents migrate using ALBR routing table to construct QoS candidate path set, from which the candidate one with the least weighted path cost is selected as the data transmission path. Consequently, several goals are achieved, such as, congestion avoidance, low ISL handover frequency and good QoS performance. Moreover, the rerouting strategy is given which combines partial path extension and full path rebuilding to improve the handover success probability as well as to decrease latency time. Furthermore, AQR is shown to achieve lower rerouting frequency and overhead, and more QoS guarantees than the traditional QoS methods on delay jitter, new call blocking probability (CBP) and handover blocking probability (HBP) with bandwidth and end-to-end delay constraints.(3) To improve the ability of service differentiation, AMSR (Agent-based Multi-Service Routing) for LEO satellite network is proposed. AMSR achieves service differentiation by providing different routing mechanisms for delay-sensitive, bandwidth-sensitive and best-effort delivery applications. Firstly, according to the characteristics of the former two applications, this dissertation proposes delay-sensitivity and bandwidth-sensitivity path metrics based on ISL available probability. Secondly, two multi-path detection mechanisms with bandwidth constraint are presented to construct candidate path sets for delay-sensitive and bandwidth-sensitive application, irrespectively. Thirdly, the data transmission paths are selected from candidate path sets using path metrics. Simulation results show that AMSR exhibits better differential service behavior, and the general performances of AMSR are better than other algorithms.(4) Research on satellite network simulation platform supporting agent technology. The simulation of agent-based satellite network routing algorithm is carried out based on NS2, in which mobile agent is implemented by extending the Packet structure. In addition, the satellite network simulation platform supporting agent tenchonlogy is developed based on Grasshopper and Linux systems, called GSNSP (Grasshopper-based Satellite Network Simulation Platform). Moreover, GSNSP implements Iridium constellation and successfully launches ALBR algorithm in a Local Area Network environment. The test results demonstrate that GSNSP is designed well, and GSNSP can achieve the simulation performance similar to NS2 if the Grasshopper's ability of multi-agent supporting and agent migration under high link delay could be improved.
|
PDF Full Text Request