Cache Design In Information Centric Network

In current Internet architecture, the network layer is dominated by IP. IP is anend-to-end communication model, which requires the address of destination node beforetransferring data. Today’s network bandwidth is almost consumed by file transmissionincluding streaming media. Unfortunately, this usage pattern does not match with IP’scommunication model. Therefore, Information Centric Network is proposed.Information Centric Network is a content distribution model. When requesting content,receivers need not to care about from where the data is obtained, they only concernabout the data itself.A typical feature of Information Centric Network is in-network cache, whichcaches content on the routers. Many project groups have already proposed their ownInformation Centric Network designs. Those designs, however, give little attention tothe in-network cache design, most of them just use on path cache, a very simple cachescheme. Pure on path cache cannot adapt itself according to the actual networkenvironment, which leads to inefficient cache use.This work will describe a cache design that solves this issue. The cache design wedeveloped can control and adjust the in-network cache according to the popularity of thecontent. By doing this, the in-network cache can be used more efficiently and the server’load will be reduced accordingly. Moreover, our cache design will use probing packetsto realize content retrieval in a k-anycast way.In cache distribution, we use multi-hash to distribute and locate the cache. Everycontent in the network will be assigned with a set of information that related topopularity. The cache system can adjust itself by checking these information, forexample, when a content’s popularity increases, the cache system will put more cachesin the network, when a content’s popularity decreases, the cache system will stopcaching more copies of the content. In this way, the in-network cache will be used morewisely and efficiently.In content retrieval, the cache system will send probing packet to check the cacheon the routers, receivers can use these results to decide how to retrieve the content. We also conduct experiment and simulation to compare the performance of ourdesign with other two designs. Three improvements can be seen from the results. First,the server’s load reduces considerably; second, the cache hit rate increases significantly;third, the in-network cache becomes more stable by using our design.