I agree with Alfonso. I'm not sure why encryption and decryption should be modeled as anything more than some potential extra latency, to be applied where these functions occur.
The security protocol determines where the security functions are being carried out, whether you are using AES or other algorithm. So for example, when using TLS (so-called Transport Layer Security), the base stations will play no part in any of this security discussion. Yes, you can definitely encrypt and decrypt at the end hosts only, and affect nothing else along the path.
I don't understand your answer "variable" as you know they are a different paper in this field "cryptography, security attacks", encryption their proposal work in simulation environments with a crypto++ framework, for example, this paper "AES-Based Security Architecture Of WIMAX Using OMNET++" :The target is to encrypt communication between WIMAX based BS (Base station) and MS/SS(Mobile station/subscriber station) using AES where encryption is applied on all the packets exchanged. and so on.
please, can you give me more explanation what do you mean with "variable" word?
You can use an integer variable in the message that identify the encryption. The nodes will have a list of values that they can recognize, if the value is in the list the node can decrypt the frame, in other case, it cannot. With this, you can simulate the encryption of the frames without the necessity to include the AES algorithm