In a fourth generation communication system, consisting of diverse variety of wireless access standards, network availability, coverage and Quality of Service (QoS) may rapidly change depending upon the device mobility and fading suffered by the signal. The complexity of the challenge increases further with the consideration of random device mobility and geographical topologies such as environments, buildings etc. In order to achieve seamless connectivity in heterogeneous networks, it is important to enhance the location awareness of the mobile node as well as to reduce the adverse effect of fading so that it is able to predict future network conditions accurately. This paper adopts a proactive, mathematical model-based vertical handover prediction approach to propose an algorithm that equips the mobile node to accurately predict network coverage boundaries. This paper uses a filtering method named `Savitzky-Golay' smoothening to reduce the effect of fading in the received signal, thereby reducing the unwanted handovers considerably. We consider WiMax and Wi-Fi for the case study due to their high bandwidth availability and ease of networking. This algorithm is simulated in MATLAB and an evaluation of its performance reveals superiority over the ordinary Received Signal Strength (RSS) based techniques in their accuracy to predict vertical handovers.