I have not seen any hard-role on having the CCC in CRN, it is more or less of whether u implement it in ur work or not, even if u check in IEEE publications, some papers use it and some not. From what I have seen, I can tell that it really depends o the scenario/system model that u are using. That same scenario will dictate the goodness of ur proposed CCC solution/implementation.

A commonly used argument against a fixed CCC is that it forms a single point of failure. Suppose the primary user starts using the channel which you use as a CCC, then there is no possibility to regain control of the network. A variation on the fixed channel CCC is a frequency hopping based CCC, which of course provides more robustness. A third method is to skip the CCC and use rendezvous methods (neighborhood discovery). There exist ND protocols that take into account the channel availability, that is, the channels that are not used by for example the primary user.

Those three items are for in-band control. For out of band control, you could use a fixed CCC, since this would require your own licensed piece of spectrum.

So, if you are not concerned with the single point of failure, you could use a CCC. Otherwise, perhaps an more adaptive approach is preferred.