We present two methods that can be useful when the network or system performance is captured by a model that is not Markovian. Although most performance models are based on Markov chains or Markov processes, in some cases one cannot maintain the Markov property and the efficient algorithmic solvability simultaneously. This can occur, for example, when the system exhibits long range dependencies or has too many states. For such situations our methods can provide useful tools.
展开▼