In the above article [1], the adopted i-IpDFT estimator was implemented strictly according to the pseudocodes reported in [2], i.e., Algorithms 1 and 2. However, a more detailed implementation of the i-IpDFT estimator is reported in [3]. One crucial detail missing in [2] but reported in [3] is: for q > 1 every time the e-IpDFT is called, line 3 of Algorithm 1 in [3] (corresponding to line 2 of Algorithm 1 in [2]) should be skipped, and the estimates of the previous iteration should be directly used in the spectral interference compensation routine in line 5 of Algorithm 1 in [3] (corresponding to line 4 of Algorithm 1 in [2]). This crucial detail not only ensures that the i-IpDFT estimator has a faster convergence rate but also provides higher estimation accuracy.