A practical traffic control system can be envisaged as a hierarchy of control loops, each serving to convert a traffic demand into a potential traffic flow. Each loop is concerned with problems that can be foreseen to arise at some time in the future and the distinctive characteristic of any loop is the distance ahead, in time, to which it prognosticates. Although this model is probably applicable to most traffic systems the present note will apply it in the context of air traffic control.

Figure 1 shows a whole family of possible control loops. (Not all of these are present in all A.T.C. systems.) In each loop there is a mechanism for the prediction of future traffic. This prediction is compared with certain internal criteria and, if these are not satisfied, some control action is taken to coerce the traffic pattern into some more acceptable form. It is characteristic of most of these loops that they cannot be proved to be capable of safely handling all conceivable traffic demands; there is an implicit assumption that the superior elements of the control system will not tolerate a traffic demand which is beyond the handling capability of the inferior loops, but that a residue of unsolved problems can generally be left to be handled further down the chain.