Hierarchial radar resource management uses multi object Markov decision scheduling with a constraint on the resources. In this paper we give a detailed description of constrained multi-object Markov decision scheduling in its general form and the separation that is achieved in the dynamic programming level using Lagrange multipliers. We then apply this general model to obtain a simultaneous beam and waveform scheduling method for radars based on an objective function that depends on both state and action. This method extends on a previous hierarchial method for beam scheduling with an objective function defined only on state. We further improve the objective function based on entropy reduction. This criterion makes the resource management to be more flexible in favor of measurements that carry more information.