Research Of Mobility Support Mechanism On Locator Identifier Split Network

With the rapid development of Internet, the traditional Internet architecture has several challenges, such as mobility supporting, multi-homing, routing scalability, traffic engineering, location privacy, etc. One of the underlying reasons for those issues is the overloaded semantics of IP addresses which used to both identify a session and locate a host. The dual role problem of IP Address seriously prevents existing network from supporting for mobility. Location/Identifier spilt is one of the effective way to overcome this problem, however, most solutions have focused on how to reduce the layer 3 handoff delay, and cannot meet the real-time business needs.In this paper, we start our research on how to reduce the total delay on both layer2 and layer3 by carrying out locator identifier split network based cross-layer roaming mechanism; and a mapping query optimization mechanism on mobile locator identifier split network is proposed to overcome the problem for the frequent updates of mobile identifier. The major contributions of this thesis are as following:Based on the analysis of 802.11 link-layer handoff and many other representative locator identifier split mechanism supporting mobility, a cross-layer roaming mechanism based on locator identifier split network is proposed with the Layer 2 and Layer 3 handoff coupling organically. By pre-registration on Layer 2, this mechanism effectively reducing the handoff delay and overhead, basically meet the demand of real-time business. Theoretical analysis shows that the total delay of this mechanism is far less than MIPv6, and is 77.5 ms less than SMOS [26].Considering the advantages and disadvantages of the cache management and pre-fetch mechanism, according to the characteristics of mobile network based onlocator identifier split, bilateral pre-fetch control algorithm based on Markov model is proposed. Combining the cache management in edge router and prefetching mapping both on mapping server and correspond node to improve the success rate of prefetching and cache hit rate. The hit rate of identifier mapping table cache would be 16 percent more than LFU. And analysis shows that the mapping lookup delay based on our algorithm is less than PPBM[18] in macro-mobile condition.At last, we design and implement the pre-registration processing module in LAP and Markov forecasting module. The result of prototype module shows the correctness of the algorithm and supporting mobile functionality.