Use of neural nets in channel routing

The channel routing problem is solved using a massively parallel processor based on the artificial neural system (ANS) computational model. In this model the functional behavior of the human designer is emulated by expressing problem constraints as interconnection weights between neural cells (computational elements), using very simple computational elements for neural cells and massive parallelism to solve problems. The algorithm constraints are expressed as interconnection weights. By making the artificial neural cells work collectively on a task, a solution is obtained after each of the neural cells settles to a stable state.<>