Abstract-The problem of optimally adjusting the power spectral density level, referred as power spectrum optimization, is a well-known and difficult nonconvex optimization problem. This paper investigates a class of local optimization methods based on a technique called iterative function evaluation. This idea, first proposed in our previous work, is based on the manipulation of the first order condition of the optimization problem into an algorithm. This paper shows that these manipulations are not unique, and proposes a new iterative-function-evaluationbased method for power spectrum optimization. In addition, this paper proposes methods based on the average channel gains for an orthogonal frequency division multiple access network. Finally, the performance of the proposed methods is evaluated in a wireless backhaul application.