“…The key idea which guides us consists of taking advantage of the knowledge of the constants T 1,i and T 2,i ≥ T 1,i , which can be derived from information on the delays of a system with poorly known time-varying delays. As in [13], our assumptions are satisfied by functions a which take both positive and negative values (which contrasts with our earlier results on generalized Halanay's inequalities from [8]- [11], which required a to be nonnegative valued) and the largest value of b i can be arbitrarily large over arbitrarily large intervals, provided that the values a(t) of the function a are positive and large on sufficiently long time intervals. Also, our Halanay inequality generalizations [8]- [11] did not use the comparison function approach that we use here, and they use only one gain term, instead of the multiple gain terms that we use here.…”