Predictive RAT-selection algorithm for enhancing QoS in heterogeneous wireless networks

Multimode terminals designed for heterogeneous wireless networks (HWNs) have the capability to support two or more classes of calls simultaneously. When a multimode terminal having an ongoing call initiates a new class of call, the current radio access technology (RAT) may not support the new class of call thereby necessitating a vertical handoff to a RAT that can support both the current and the newly initiated call. However, vertical handoffs incur signaling overhead, handover delay, packet loss, and service disruption. Therefore, it is important to reduce the frequency of vertical handoffs in HWNs. This paper proposes a predictive algorithm that considers the current state and the possible future state of a multimode terminal in making RAT selection decisions, and thereby reduces the frequency of vertical handoffs in HWNs. The history of a user's call logs is used in making predictions about future states of the user's multimode terminal. The paper develops an analytical model to evaluate the performance of the proposed predictive RAT-selection algorithm, using a four-RAT HWN as an example, and considering three classes of calls namely voice, video, and data. Results show that the proposed predictive algorithm significantly reduces the frequency of vertical handoff calls when compared with non-predictive algorithm in HWNs.