In this paper, we compensate the measurement errors in HMRR and PTSR using 1-step ahead prediction method.
Before we model the link cost function using three inputs as mentioned in the previous section, we validate whether RSS and HMRR are correlated, using the input sensitivity explained in the our previous work [33].
As seen in Table 1, when we vary input [x.sub.1](RSS), the input sensitivity of [x.sub.1] is 0.45 and that of [x.sub.2](HMRR) is 0.36 while that of [x.sub.3](hop distance) is 0.09.