My initial thought would be to use a semantic web based solution where you define an ontology of possible attack points of the MANET as well as controls for mitigating these attacks. This would allow you to add semantic constraints to the model so that you then could do inference on instances of the ontology etc.
We did some similar work about this in a slightly different domain in the PRECYSE EU project, where we used ontology based reasoning for inferring information about cyber-attacks in critical infrastructures. CSIT at Queens University Belfast (amongst others Kieran Mc Laughlin and Sakir Sezer) have continued this research path after the project finished.