Identifier Mapping Buffer Management Based on Markov-Repeated-Using-Probability Prediction
Jianwei Zhang, Yunfei Guo, Jianqiang Liu, Baowei Zhang · 2010
The Internet can not support mobility and scalability effectively. One of the reasons is the ambiguity of the IP address signification. It has been accepted that the new network architecture needs to separate the locator from its identifier. Identifier is used to present the information source and information sink while the locator is used for routing. It is necessary to use replacement, not encapsulation, for the locator and identifier in the communication because of the security problems arose from the application of encapsulation. In the process of replacement the backup of the locator/identifier mapping information needs storing in the most necessary place, which is a problem of cache management in fact .In this paper the typical cache management algorithms are discussed firstly, and then introduce the dividing method of Access Switch Router (ASR) cache space. On this basis, a Markov Repeated using Probability Prediction algorithm (MRPP) is proposed, in which it focus on the computing method of state transferring matrix and the computing formula of the querying times in the next period. The simulation result shows that the performance of MRPP is better than that of the existing cache management algorithm such as LFU, LRU and LFU-LRU.