The computational simulation of biochemical pathways requires a technique for estimating unmeasurable parameters from observable data. In many cases this can be problematic, as parameter estimation problems for biochemical pathways tend to have difficult properties. This paper proposes two techniques that can be used to overcome the difficulties of such problems. The first, a problem decomposition, is used to divide the parameter estimation problem into several subproblems. The second, a logarithmic transformation, is used to decrease the flat area of the objective function of this problem. We apply these two proposed techniques to estimate the parameters of a model of HRG-stimulated ErbB4 receptor signaling.