2019-02-21-femtocahing

1.对femtocaching的进一步理解

终于明白K.关于femtocaching 的总结了：

It starts with all the caching being empty. Iteratively, it performs the placement of a file to a cache that improves performance(objective function) the most.

Speciallly, it means that select the biggest W(n,f) to make X(n,f)=1, instead that a certain cache node selcets the biggest W(n,f) to make X(n,f)=1 with respect to its local demand.
实际上，femtocaching是在全局实现的贪心算法，反而greedycaching实在局部一个cache node是的的贪心算法。

2.femtocaching 与"On the Complexity of Optimal Content Placement in Hierarchical Caching Networks"的区别

因为论文基于submodular function 采用贪心算法，其与femtocaching的代码实现基本类似，唯一区别在于objective函数的计算方面。论文的缓存策略的setting是基于：finding the caching policy at the leaves given the caching policy at the root()
femtocahing 的贪心算法是基于marginal value， the greedy algorithms selects the highest marginal value with respect to a placement set X.