Support of resource-aware vertical handovers in WLAN hotspots

End-user devices such as smart phones and tablets have become very popular as they offer a variety of wireless Internet accesses ranging from the WLAN standards to WWAN technologies such as LTE or even WiMAX. Due to these different wireless access options and new emerging applications—e.g., from the areas of video streaming, social networks, as well as Internet clouds—people are increasingly connecting to the Internet with their devices while being on the move. In line with this, the number of devices as well as the traffic demand of end users have been reported to increase rapidly over the last years which imposes a strong challenge especially for the operators of WWANs. Thereby, end users frequently tend to use settings that favor a connectivity to the Internet whenever possible rather over WLAN than over WWAN access. Further, the cellular standardization bodies of the 3GPP envision solutions to hand over on-going wireless sessions from cellular to other small cell accesses such as WLANs or femto cells. This is also known as traffic offloading essentially freeing capacity in terms of users with a certain service in the cellular accesses. Nevertheless this offloading follows a rather simple strategy to utilize additional capacity of heterogeneous accesses such as WLANs whenever being available for a given device. This thesis shows that stronger gains can be expected if the selection of devices to be served in WLANs is conducted in a resource-aware fashion including an evaluation of the WLAN traffic in terms of the channel occupation time and MAC overhead as result of contention, interference, and fluctuating channels. In this context, this thesis envisions to onload unfavorable devices negatively affecting the WLAN capacity back to WWAN accesses. A support of such an onloading imposes challenges in different dimensions. From the hardware design of devices, there is a strong trend to limit the number of separate network interface cards (NICs) due to space and cost issues. We refer to a multi-mode NIC if it covers multiple technologies, while at a given time only access to one technology is possible. Thus, smoothly onloading a device with such a NIC is by far not trivial. We present an approach that conducts handover preparation mechanisms, while also allowing a continuous WLAN communication over a multi-mode NIC. Further, it is by far not trivial to judge which subset of associated devices is negatively affecting the capacity of a WLAN hotspot. Thus, a careful evaluation of devices regarding a selection for an onloading back to WWAN accesses imposes a challenge yet. In this direction, we present a performance metric that identifies devices degrading the WLAN capacity. While our performance metric tackles a reactive selection, it falls short to support a predictive evaluation, e.g., of devices which just joined the WLAN cell. Even worse, proprietary algorithms inside a WLAN stack impose a severe challenge as their internal routines are usually not conveyed via typical management interfaces. A well-known example for this category of algorithms are the link data rate adaptation schemes, with which WLAN devices adjust the modulation and coding scheme (MCS) for their transmissions. As MCSs resulting in low link data rates may specifically degrade the capacity of a WLAN cell, we focus on an estimation regarding the data rate selection of a device as a third contribution of this thesis. This estimation enables to select devices that will likely degrade the capacity of the WLAN hotspot for an onloading in advance.