If some of the variables are binary (or at least integer), you may be able to linearize the left side. Otherwise, I suspect the best you'll be able to do with regard to reducing the degree is something like a McCormick linearization, which may be too coarse for your purposes.
The left side of the constraint does not look convex to me, so you likely would be better off looking for a global NLP solver.
#CPLEXOptimizers#DecisionOptimization