A set of new and efficient formulae for buffer size analysis of real-time systems using M/G/1 models

A set of new efficient and compact formulae for buffer size analysis of real time systems using M/G/1 queueing model has been developed. For Poisson random arrival and general service time distribution of a single server system, or an M/G/1 system, and for a certain probability of overflow as the confidence level, the needed size of buffer can be estimated. Two subsets of M/G/1 systems, namely the M/D/1 and M/E/sub k//1 systems, are investigated in detail to illustrate the practicality of this approach. The formulae are derived analytically and are validated using term by term evaluation. The size of buffer needed for M/D/1 and M/E/sub k//1 systems are tabulated for design and validation purposes. The newly derived formulae are more efficient and compact than currently known computation methods.