Comparing Apples to Apples in ICN

Many algorithms and applications in ICN directly depend on content names. All the more surprising that there is no common foundation to evaluate them against each other. Without a common foundation, apples are compared to oranges. As this paper has the goal to enable real, realistic and comparable evaluations in ICN, also methods to compare sets of content names are needed. There are two ways to reach that goal: A standard collection of content names, or an abstract description for data set characteristics, facilitating comparison. We provide solutions for both ways. We present the derivation of the Content Name Collection (CNC), a collection of content names based on real data. We gain generalized insights through empirical observations. On this foundation, two possible mathematical abstractions realistically describing data set characteristics are introduced and discussed. With a slightly extended formula for skewed normal distributions, we found a potent candidate supporting our goals. Moreover, all abstractions can be used not only to describe existing data sets, but also for realistic simulations of data sets.