Fingerprinting and reconstruction of functionals of discrete time Markov chains

We explore various fingerprinting options for functionals of Markov chains with a low number of parameters describing both stationary behaviour and correlation over time. We also present reconstruction methods using lazy Markov chains for the various options. The proposed methods allow for efficient simulation of input data with statistical properties similar to actual real data to serve as realistic input of a data processing system. Possible applications include resource allocation in data processing systems. The methods are validated on data from real-life telecommunication systems.