In this paper we investigate the problem of decision making in dynamic spectrum allocation radios by using energy detection while being close to the coverage area of a primary transmitter. Currently, the signal model used for dynamic frequency selection assumes that the primary signal is either absent or present with a fixed level. However, this model fails to capture the power distribution close to the cell border and thus the prediction capability of the related analytical approaches is rather limited. In this paper we calculate the distribution of the signal power outside the primary cell and then we approximate it by using uniform distribution. On one hand, the proposed approximation allows to estimate the decision performance in a closed-form. On the other hand, the simulations illustrate that the proposed model has better agreement with realistic system behavior compared to the conventional model. The proposed model allows to design the detector such that the system performance in terms of spectrum reuse is improved.