Period assignment in multidimensional periodic scheduling

We discuss the problem of assigning periods in multidimensional periodic scheduling such that storage costs are minimized. This problem originates from the design of high throughput DSP systems, where highly parallel execution of loops is of utmost importance, and thus finding an optimal order of the loops is an important task. We formulate the period assignment problem as a linear programming (LP) problem with some additional, nonlinear constraints. The nonlinear constraints are handled by a branch and bound approach, whereas the LP relaxation is handled by a constraint generation technique. The effectiveness and efficiency of the approach are good, which is illustrated by means of some practical examples.