Huijuan: There are many "reference" genes as bases of monitoring changes of other genes. I'd suggest to monitor the commonly used 18S, GAPDH, actin, etc to see if any of them show a big change (>2 fold) in hyper- and hypo-oxia. Also, carefully monitor the linearity of your qPCR reactions for accuracy and sensitivity (too much cDNA may also hinder the changes). Good luck.
It is more appropriate that the reference gene is stable under your conditions, rather than to use one everyone else uses. To test this, you should run equivalent amounts of target cDNA across all samples (or at least a representative of each condition, but the more samples you use, the better) against as many reference gene options as you can find. You then use the Ct values (which, for a stable reference gene will be the same in all conditions, because the cDNA has been standardized) in a program called RefFinder (online, free to use).
http://fulxie.0fees.us/?type=reference
This will score your reference gene options and choose the best one for you. There are several citations on this webpage that can inform you how the program works to score stability. Cheers!