Skip to Main Content
A heterogeneous wireless network consists of various wireless networks [e.g., Worldwide Interoperability for Microwave Access (WiMAX) and Wireless Fidelity (WiFi)] and cellular communications [e.g., beyond the third generation (B3G) and the fourth generation (4G)]. Vertical handoff is an important mechanism for achieving continuous seamless transmissions in these networks. In contrast to horizontal handoff, vertical handoff considers not only the received signal strength (RSS) but also the service-class mapping between handoff-in and handoff-out networks. Most previous works have adopted the RSS-based mechanism to determine handoff thresholds, which causes a serious ping-pong effect that increases unnecessary handoff. Although integrating the RSS-based mechanism with a hysteresis method reduces the unnecessary handoff, it suffers from high dropping [i.e., high Sum of Weighted Grade of Service (SWGoS)] and low utilization. Therefore, this paper proposes a cross-layer-based adaptive vertical handoff algorithm with predictive RSS to reduce the unnecessary handoff while significantly increasing utilization and decreasing connection dropping. The proposed approach determines the optimal target network in two phases, i.e., polynomial regression RSS prediction and Markov decision process analysis. Furthermore, fast changes in bandwidth caused by vertical handoff result in inaccurate Transmission Control Protocol (TCP) congestion control and, thus, reduce the TCP goodput. The cross-layer scheme provides a TCP receiver to reply to the TCP sender with the wireless network's protocol type. By using the cross-layer information, the TCP sender can accurately predict the available bandwidth and increase the network goodput. Numerical results indicate that the proposed cross-layer-based approach outperforms the other approaches in the number of vertical handoffs and SWGoS while yielding competitive utilization. In addition, the cross-layer scheme cooperates with existing TCP algo- - rithms to increase goodput by up to 18%.