Conventional traffic simulation techniques, employing continuous movement of vehicles, do not have a sufficiently fast responce time for the real-time operational control of an urban network. A new simulation technique is proposed using field-programmable gate arrays (FPGAs). This radical approach offers simulation at a speed orders of magnitude faster than real-time and at a significantly lower cost than simulation with parallel architectures. A discrete model of traffic movement is introduced in which the roadway is modelled as a series of connected cells, each representing a short length of roadway, which either contains a vehicle or is empty. Individual vehicles progress from cell to cell using an established protocol. It is possible to develop models of individual road links and intersections. These components have been tested under a range of typical geometric configurations and traffic loadings. The validity and tractability of these network components is demonstrated. a network can be constructed in a heirarchical fashion by compling these components together. This work has been funded by EPSRC Grant Number GR/J09239.