Well, I have limited expertise in exosomes, But, I have recently gone through a paper where they also isolated exosomes from urine of chronic kidney disease patients. In their paper, they have used miR-29, miR-200 and RNU6B as endogenous control.
First, you should consider that microRNAs in urine may reflect various issues associated with diseases, i.e. they may be disease-specific. I would suggest to find the best controls directly in your samples and not according to other papers. There is no consensus for this, and what may be considered as suitable control somewhere may not be proved and functional in other analyses.
Second, many different approaches exist for exosomes isolation and they differ in results for miRNAs expression. It is another reason why one should follow their own samples and methods applied.
Third, normalization in large-scale manner (e.g. miRNAs arrays) may be different than in individual miRNA assays. In large-scale experiments you may use global/geometric mean normalization, while in the latter case individually validated controls may be suggested. Using more individual miRNA assays you may apply also geometric mean normalization when this is functional and in congruence with normalization alternatives, as in our paper: http://www.ncbi.nlm.nih.gov/pubmed/25827090. But generally, this is not recommended where individual or several selected controls perform well, and the number of assays is limited.
Finally, you can follow several algorithms for the selection of suitable endogenous controls, many cited here on RG for gene expression.