Yes, a diode ring can be used.
I guess that the generation can be optimized with some phase difference between LO and RF inputs. For example, if RF would be square wave, it should be in quadrature with LO for maximum efficiency (if they are in phase, the 2nd order product would cancel).
As another possibility, you can use a push-push configuration.
Regards
Is the mixer based in a Gilbert cell (as the file name suggests)?
Can you give a more detailed description if the mixer block?
I find strange the direct connection of the two ports (N and LO).
Regards
The two-diode balanced mixer structure can act as a good efficiency doubler. Diodes alternately open and close, so from the input frequency the second harmonic is generated (in fact, all even harmonics). The IF line is not used. Look at "www.spaceklabs.com"; we use such doublers and quadruplers as commercial models.
If you need a frequency tripler, diodes are connected "anti-parallel" and generate the third and odd harmonics.
Similar structures are used in MMICs.