“…Although there are some MDSQ algorithms also for nonconvex cases (e.g., see [7], [8], [4]), several papers propose algorithms for convex MDSQs, and most MRSQ algorithms address the problem of finding the optimal one among convex MRSQs, that is, they hence lead to overall optimality if there is a convex optimal MDSQ/MRSQ (e.g., see [2], [9], [10], [11], [5], [12], [6], [13]). Under this constraint, optimal SQ design is equivalent to designing the optimal threshold sequence and corresponding codewords, so the search space is usually reduced significantly.…”