A large number of models for the description of time-headway distributions have been proposed. This report presents a new and straightforward derivation of Branston's Generlised Queuing model. Both the headway sustained by constrained vehicles and the free-flowing headway are assumed to be randomly distributed. It is established that traditional statistical methods do not efficiently estimate the parameters of the model. A new statistical procedure suited for the estimation of parameters in headway distribution models is proposed, significantly reducing computation time, thereby enabling on-line application of Branston's headway model. The new procedure also provides better results than other estimation procedures. Using results from the reproduction study, performance of the method is compared to the performance of the Maximum-Likelihood method. The model is validated by comparing the results from application of the method on the real-life headway sample to results from the reproduction study. (A)
Abstract