It's interesting, as more low impact development practices (porous pavement, greenroofs, etc) are used it gets harder and harder to adapt CN methods that were developed for conventional, deep soils. For example, one CN may match the volume of runoff for the proposed BMP but yield too high of a peak runoff rate. So you have to look at the CN as well as the Tc, using the Tc to control the runoff peak to a reasonable value. Case in point is using the gravel CN for porous pavement, what is the appropriate Tc to go with a particular CN? Does it depend on the size of the storm (probably should!)? There's still work to be done in making these BMPs model-able. My firm has actually developed an alternative model for green roofs since CN methods are so poorly suited for the processes that occur in these systems, and many of our clients want to install greeen roofs.
All that said, for a porous pavement or paver system you can compute a CN by using the old SCS formula where S = 1000/CN - 10, with S as the potential moisture storage in inches.
So, for example, if you have 12 inches of gravel to store water that percolates through the asphalt, at 30% porosity (typ. for gravel) the equivalent CN would be 73.5. At 6-inches of gravel it works out to 84.7 This doesn't account for underlying soils, which should improve the performance/lower the CN, but it is easiest to demonstrate the CN attributable to the gravel layer alone.
For comparison, the City of Portland BES drainage manual (page C-5) treats it as gravel and allows CN= 76/85/89 for porous pavement over hydrologic soils group A/B/C respectively.
I would go with one of these CNs and also make the case of at least doubling the Tc compared to conventional pavement.