A Metering Infrastructure for Heterogenous Mobile Networks ‡ Andreas Monger * , Marc Fouquet ‡ , Christian Hoene * , Georg Carle ‡ and Morten Schl¨ ager † ‡ Network Architectures and Services, Technical University of Munich * Computer Networks and Internet, Wilhelm Schickard Institute for Computer Science, University of T¨ ubingen, Germany † Nokia Siemens Networks, Berlin Abstract—The next generation of mobile access networks will support different access technologies like UMTS, WLAN, WiMax and LTE. Optimal handover decisions between different access networks have to take various types of information into account, such as the load of possible target cells. However, collecting this information and transporting it to mobility management decision engines is costly in terms of bandwidth. With our flexible Generic Metering Infrastructure (GMI) we are able to collect state information from the core and access networks efficiently. Using the GMI, network operators will be able to control handover decisions for their users. I. I NTRODUCTION Today many operators already run WLAN access points to supplement their GSM and UMTS networks. In the future they will add new radio access technologies such as Long Term Evolution (LTE) [1] or WiMax [2] to improve user experience and offer new services and to increase the capacity of their access networks. To support multiple access technologies, new network control- and management functionalities have to be introduced, which support heterogeneous handovers and resource management to make sure that mobile devices are optimally connected to one or more of the currently available access networks. Future mobile networks should be able to offer seamless handovers between different access technologies. As the man- agement of heterogeneous networks is a major topic in current research a new term named Always Best Connected (ABC) has evolved in recent years [3]. It is a catchphrase that involves both sides in mobile communication to be considered: On the one hand, users would like to be connected to “best” available network in terms of e.g. signal quality and bandwidth. On the other hand the network provider wants to share the available resources between millions of users in a fair or privilege-based fashion. State of the art mobile devices support multiple access technologies and the handover between them. Generally, ei- ther the mobile devices or their users decide which access technology to choose. But a mobile-driven handover decision is not always desirable because the following reasons: First, this kind of network selection poses a burden on the end This work has been developed in cooperation with Nokia Siemens Networks within the BmBF ScaleNet-SYMPATHIE project. user, if it involves user interaction. To hide this complexity from users, smart decision functionalities should be provided that do not involve the user directly. Second, each node has to scan for neighbouring networks, which consumes a considerable amount of power. To prolong the battery run time of mobile devices, radio transceiver that are not needed shall be turned off most of the time. Third, as each mobile node optimises only its own connection without considering the impact of its decisions on other nodes, the decision leads to potentially suboptimal solutions. If, for example, one mobile node connects to a WLAN access network while having a bad link and slow modulation, this can significantly worsen the connection quality of other nodes that are also connected to the same access point [4]. Instead it would be better if the node connected to another access technology in which it would not degrade the connection quality of adjacent fellow nodes. The “Wizard of Oz” view on the network allows for a decision functionality on the network side that can take numerous users and base stations of heterogeneous networks and network-side parameters like the utilisation of certain base stations into account. Thereby it can provide better decisions than a single user may make, which promises better performance and allows for a network-controlled sharing of resources in heterogeneous networks. Currently, a lot of effort in 3GPP [5], [6] and IEEE [7] is spent on integrating multiple existing and future access tech- nologies. Our work contributes to these efforts and provides an architecture that enables the management of distributed, heterogeneous networks. To provide heterogeneous handovers we follow an approach that separates data collection, decision making and execution of handovers. Previous work (such as [3], [8], [9]) has focused on decision making, when to conduct the handovers. These approaches require state information from the access network. For example, Staehle [9] defines how to get this data from an UMTS network. IEEE 802.21 [7] defines a media-independent interface to support handover decision on the mobile device. Our approach differs from the related work as it proposes an abstraction layer between data collection and decision making (similar to IEEE 802.21) but allows for technology-independent support of heterogeneous network-controlled and network-assisted handovers. We present the Generic Metering Infrastructure (GMI) that