I think from most papers i have read, they use actin as endogenous control for miRNA normalization in QrtPCR while U6 is used for northern blot endogenous control. All the best.
If you have a miRNA you suspect is stably expressed, you could compare it to 1 or more stably expressed mRNAs to check it.
Alternatively, you can use a mRNA as a normaliser for miRNAs.
Be wary of miRNAs suggested as normalisers in the literature. People rarely give a strong argument for their selection. Don't just accept what other people have previously used.
Choosing a housekeeping gene for normalization of miRs is highly depends on the cells/tissues being analyzed. We use both snoU6 and GAPDH as they didn't change in our experimental conditions.
U6 is not always the best choice, especially if you are working with "cell-free" samples. If there has been no published work with reference miRNAs in your particular experimental condition, you ay have to establish the stability of your reference. Using the older, free version of geNorm, or the still-free version of Normfinder. Newer software like what Mikael suggests could work too.