ABSTRACT
The design technique of Chap. 6 inherently involves some overdesign, as seen from Eq. (6.1), in which t21, t31 can be any members of their acceptable sets T21> T31 and qi2,qi3 any member of their uncertainty sets. As noted in Sec. 6-2, it is therefore necessary to use the worst case values which leads to overdesign. Actually, in the real world, there is a correlation between the t2i,t31 and the qu,qi3, etc. For example, in Eq. (5.55) it is possible that q12 is large when t21 is large in the expression for cu. Such correlation a can only help make cu smaller. Thus, for Method 1, it is not possible to use this correlation, and so one must take the largest t21, the smallest q12, etc. This is the price paid for converting the MDVIO problem into the much simpler MISO problems, and avoiding having to work with the horrendous denominator in Eq. (5.25).
