1. Find the global minimum of CRM(j), CRM _{ min }(j); |
2. Define a window within H that has as first column the corresponding column of CRM _{ min }(j), the same number of rows of H, and a predefined number of columns l (a window of columns); |
3. Solve the equations system H without the packets (variables) used in the sub-matrix of the global minimum; |
4. Point out which equations (rows) cannot be solved; |
5. Move the 1 entry belonging to an equation unsolved (row A) in the sub-matrix of the global minimum to another equation (row B). Row B is a row that does not loose its recovery capability with this change. The objective is to solve the equations system (as in the example in Fig. 6 where row A is the third one, and row B is the second one); |
6. Find the global maximum of CRM(j), CRM _{ max }(i); |
7. Define a new sub-matrix that starts in the column of the global maximum and has the same dimensions of the previous matrix; |
8. Move 1 entry from row B to row A, in order to preserve w _{ r } for all the columns of H. |