The method presented here originates from a long specialist experience in the use of various commercial traffic assignment models, after having considered these as inadequate to simulate the real conditions that determine the behaviour of users of transportation networks.

At variance with the majority of traffic assignment models, which prevailingly adopt a deterministic approach, STREAMS relies upon a probabilistic assessment of the factors that may influence the decisions  as to the choice of the most convenient routes by the users of  a given transportation network.

The simple principle, on which STREAMS is based, is that any user of a transportation network makes trips after choosing the routes that are subjectively considered as the most convenient ones. For most of the current assignment models the choice of "the most convenient route" is not defined prior to starting the trip, but during the simulated trip, depending on the simulated traffic conditions encountered by the “virtual user” on the way towards its destination.

Therefore, in most of such assignment models, the user’s route results from a progressive and unpredictable adjustment of traffic streams to arbitrarily simulated accidental events rather than from the user’s rational choice. Moreover, most of such models are biased by the operator’s arbitrary criteria concerning, for instance, priority to be given in the assignment to certain stream categories, demand stocks and so forth. Therefore, it is not appropriate to state that these models account for the user’s “most convenient" choice.

The sequence of links, which connect any trip origin to a destination, includes usually a very high number of possible routes. However, many of such routes are either unlikely or never used in practice. But unlikely or quite unreasonable routes are involved by current assignment simulation models when - on the one hand – the simulation saturates all the likely routes, while - on the other hand - the given demand for travel must be met by the simulation, but the saturated likely routes can no more cope with the demand. The saturation occurs when the volume of the traffic stream achieves the level of the route capacity. In simpler words, although traffic congestion is a frequent occurrence, most of the assignment models are not able to simulate processes of traffic congestion, which instead arise when the loads of the traffic flows exceed the route saturation levels.

An important feature of STREAMS is that this model accounts for preliminary assessments - as supposedly made by users prior to undertaking their travel - concerning the traffic conditions relevant to a selected set of routes, which are considered as the only reasonable alternative routes for each possible travel. These preliminary assessments are not made by separate categories of users and allow for the users’ expectations relative to aggregate traffic streams.
Nevertheless, preliminary assessments of the traffic conditions are often wrong and traffic congestions are not avoided, while it’s also realistic not to exclude that part of the users don’t reject the risk of “acceptable levels” of congestion during their travel. Therefore, the model provides a consistent logical way to simulate the overcoming of traffic congestions, as these emerge during the simulation exercises.

From the methodological point of view, a major feature of STREAMS consists of avoiding the adoption of questionable “optimisation criteria”, such as minimisation of travel cost, of route length or trip duration, etc. Under the constraints fixed by the available data, the mathematical structure of the model exploits the amount of probabilistic uncertainty that is always associated with any distribution of probabilities, such as -  in the model - the unknown distribution of trip probabilities to be determined. It’s an unusual analytical approach, which minimises also the number of basic hypotheses, with a consequent reduction in the degree of logical arbitrariness of the model’s structure.

STREAMS has been conceived to address transportation networks covering large regions, i.e., interurban, interregional or international connections. However, the model can also be adjusted to applications concerning large conurbations, after appropriate zoning and careful identification of travel demand.