This paper explores two tools for evaluating irregular sequences of numbers
occurring in the operation of computer networks. The time series might be
interarrival times, packet lengths, or IP destination addresses. The tools
were developed because there is a need for network designers and administra
tors to understand network traffic. an under standing not provided by conve
ntional statistical methods. An application of the tools would be compariso
n of real and synthetic sequences: should a synthetic sequence not yield th
e same outputs as a real sequence. then use of the synthetic sequence in mo
deling would be questionable. The first tool is a modification of fractal d
imension work in other fields. notably radar research. Its application is l
imited to data for which a numerical comparison makes sense, such as time l
engths. The second tool essentially discovers the same thing as the first,
the existence of data points that are visited frequently. However, the seco
nd tool does not rely upon a metric and so can be applied to data such as a
ddresses. (C) 2001 Elsevier Science B.V. All rights reserved.