Tactical networks are typically of an ad-hoc nature operating in highly restricted environments and constrained resources. The frequent presence of communication disruptions and network partitioning must also be expected and managed, while core functionalities must be maintained, providing asynchronous invocation and access to services in a distributed manner. Supporting the required functionalities of the contemporary tactical environment, requires the dynamic evaluation of security policies, incorporating semantic knowledge from various network layers, together with facts and rules that are defined axiomatically a priori. However, the required basis for such policy decisions can be excessively extended and dynamic. Thus, it is desirable to locally minimize the scope of the policy maximizing efficiency. In this paper, we therefore analyze criteria and optimization goals for the a priori distribution and partitioning of security policies, ensuring the continuous support of the required capabilities, given the operational tasks of each deployed actor.