Domain Decomposition Solver (DDS) is intended for simulation of electrically very large problems, which are too large to be solved by direct MoM Solver. The solver can be applied to both: antenna placement and RCS problems.

Method description

The basic idea behind DDS is that the original model is decomposed into a number of groups. A group is composed of a number of neighboring plates and wires. Basis functions, which correspond to plates and wires from the same group, are grouped into a macro-basis function (MBFs).

All MBFs are multiplied by corresponding weighting coefficients, then replaced into full MoM matrix and residua of all equations are calculated. Weights of MBFs are determined by minimizing residuum of full MoM matrix. The residuum of the final solution in each iteration is used as the excitation in the next iteration. MBFs excited by low values of residua can be eliminated in iterative procedure, which significantly reduce simulation time with negligible influence to simulation accuracy.

The entire iterative procedure finishes when the total residuum falls below the predefined threshold.

In the 0th iteration, subprojects are simulated independently and the coupling between them is not taken into account. Groups which are simulated in the 0th iteration are:

  • the groups which contain excitations – for the case of antenna placement problems,
  • the groups directly illuminated by exciting plane waves – for the case of RCS problems.

The method itself has very fast convergence. Acceptable solution is obtained after 1 to 3 iterations.

Problem with 5 million unknowns (more than 1,000 wavelengths in size) can be solved in one day, on multi-core desktop machine.

As an example, let us consider an anti-collision radar antenna (4×4 patch array) placed on a car bumper. Operating frequency is 77 GHz and the number of unknowns and simulation times are given below.

For more info on Domain Decomposition Solver and its areas of application, please check the following application notes or published publication: