Research On Path Capacity And Mobility For IPv6Networks

Since the first Internet protocol version6(IPv6) was published in the90’s of the20thcentury, the expectations for Next Generation Internet (NGI) has increased with eachpassing day. With the characters of unlimited address space, simplified header andconvenient address auto-configuration mechanism, IPv6will be a good solution tosolve the problems such as IPv4address exhaustion, sacrifice of the end-to-endcommunication model et al which are encountered in recent IPv4network. Recentlymore and more attention is paid to the development trend of IPv6by a large number ofresearch institutes, lots of investigations on the NGI have done at the same time. Thoseinvestigations not only improve the functionality and remedy the defect of NGIprotocol suite, but also accelerate the progress of commercialization of NGI.Same as other frontiers, there are still some problems remaining to be solved in NGIwhen it is coming into our life. Such as how to estimate the parameters of the networkperformance precisely when the packet forwarding rate is sharply rising, how toprovide a high speed and uninterrupted Internet access service when users are movingfast. Based on the theory of IPv6, this paper has done researches on the end-to-endperformance estimation and mobility management of the IPv6network. The maincontributions of this work can be summarized as follows:1. Based on fragmentation mechanism, a novel scheme of end-to-end path capacityestimation for IPv6networks is proposed and the real application of the capabilityparameters is also proved. After analyzing the effects of the packet dispersion on thealgorithm of the path capacity estimation, a conclusion that the accuracy of the pathcapacity estimation can be promoted through decreasing the packet dispersion is drawn.The fact that the time gap between two packets generated by fragmentation mechanismis narrower than that generated by continuous sending policy is found. Based on thisand the usage of Path Maximum Transmission Unit (PMTU) discovery and a specialdesign of type0routing header, this paper gives a detailed strategy of end-to-end pathcapacity estimation for IPv6networks. Finally, extensive experiments have been doneand the results have been analysed, compared with the classical tool called pathrate.2. Based on the redundant router information and pre-buffering strategy, a newscheme called RIRS (router information redundancy scheme) was proposed to improvethe efficiency of IP connectivity resumption in case of erroneous prediction in fast handovers for mobile IPv6(FMIPv6) protocol. Firstly, all the neighboring accessrouter are considered to be a probable next access router (NAR), and the concept ofredundancy routing table for handover is proposed and applied in access router in orderto assist the mobile node in undergoing handover. Secondly, a pre-buffering strategy isalso proposed to decrease the number of the lost packets due to the erroneous handover.The buffers are created in access router according to the average handover latency inhistory and all the duplicate of packets destined for the mobile node during handovershould be buffered in order to avoid packet loss, meanwhile the packets themselves arestill forwarded to the destination. Thirdly, theoretical analysis is applied to theproposed scheme in terms of handover latency and packet loss in four applicationscenarios, compared with FMIPv6. Finally, the overhead cost of the proposed schemeis given.3. The basic principle and application prospect of Proxy Mobile IPv6(PMIPv6) areconsidered. The protocol requires all communications to go through the local mobilityanchor. As this can be suboptimal and results in higher delay and congestion in thenetwork in the case where both endpoints are located in the same PMIPv6domain. Anovel localized routing (LR) scheme for PMIPv6is proposed to solve the abovementioned problem. The unified rules of localized routing initiation are established andfour application scenarios are divided into according to different connect amongmobility entites in same PMIPv6domain. Then the detailed localized routing policiesare designed individually for the above scenarios. Theoretical analysis and simulationresults indicate that the proposed scheme can obviously shorten the response time ofLR compared with Localized Routing Protocol (LRP) proposed by IETF, and make upthe defect that the solution of LR cannot be provided in some special scenario whenusing LRP.4. Based on time and distance weighted k-nearest neighbor prediction, a newhandover scheme is proposed to overcome the shortcomings that the Fast Handoversfor Proxy Mobile IPv6(FPMIPv6) protocol cannot support Network-based LocalizedMobility Management (NETLMM). The prediction mechanism is changed and theMobility Access Gateway (MAG) is choosed to perform related operations, such asprediction and mobility management. Firstly, the mobility access gateways areregarded as a machine learning system and all experiences generated from historicalhandover are stored in it by introducing a Handover Knowledge Database (HKD).Secondly, the network-based location technology is employed to get theangle-of-arrival of a mobile node that is about to undergo handover. That information is then be used to choose the nearest k of experiences to handover point from HKD.Thirdly, those experiences should be weighted by time and distance to determinewhich MAG the mobile node will connect to after handover. The pre-buffering strategyis also employed to protect the packets destined for the mobile node. Fourthly, anoptimized solution is proposed based on feature projection. The coverage area of aMAG is divided into several judgment sections according to the HKD and a fastprediction may be made when handover point is just located in any section. Finally,theoretical analysis is applied to the proposed scheme in terms of handover latency andpacket loss in different scenarios, compared with FPMIPv6.