Performance Analysis of the 3G Network with Complementary WLANs
An analytical modeling method is developed for the evaluation of integrated 3G/WLAN networks. To adapt the feature of the emerging wireless networks efficiently and to cover many possible situations in reality, the cell residence times are modeled generally. The channel occupancy times, the horizontal and vertical handoff rates as well as some interesting performance measures are derived and calculated. Both analysis and numerical results show that the performance of the 3G network is significantly increased when the integrated 3G/WLAN network is employed, and that the performance of integrated 3G/WLAN network is improved greatly when the WLANs are in the hot-spot areas. The method developed here is expected to be useful for modeling and performance evaluation of other type of networks, such as the 3G network with the infrastructurelessmode WLANs or ad-hoc networks.
Future wireless networks will generally be characterized by heterogeneity in radio access technologies. Currently developing 3G technologies such as UMTS, cdma2000 and TD-SCDMA will provide to mobile users wide coverage area. However, the costs of acquiring the necessary radio spectrum and the network equipment upgrades are very high. In the same time, wireless local area networks (WLANs) continue to proliferate in corporate and residential environments due to their low-cost, high-speed wireless Internet access for the localized hot spots. The research 802.11 family has dominated in the United States and is now extending worldwide. In Europe, ETSI has developed a competing standard HiperLan2 . However, the coverage offered by WLANs is quite limited and they lack roaming support. The complementary characteristics of 3G networks (slow, wide coverage) and WLANs (fast, limited coverage) make it attractive to integrate the two technologies to provide ubiquitous wireless access. The 3G networks provide global coverage, mobility and quality of service (QoS), while WLANs are designed to cover localized hot-spot areas, such as airport, shopping center, hotel, etc. Although the WLAN is subject to its small area and limited mobility, it can substantially increase the capacity and performance of the associated 3G networks. Handoffs will be possible (with some trade-off in performance and service) between the 3G network and WLANs. There are many possible interworking approaches between WLAN and 3G networks [2-5]. For any type of interworking approach, a multimode 3G/WLAN mobile terminal is required. The mobile terminal should have at least two logical air-interfaces: Uu-interface (same as the definition by 3GPP) for communications with a Node B at 2GHz (3G band) and Ur-interface (not yet defined in 3GPP) for communications with an access point (AP) at 2.4GHz or 5GHz band. Since the Ur-interface works at different frequency band, it has almost no interference to the transmission signal at the Uu-interface. Some channels in the WLAN-worked band can be reserved for servicing 3G-related sessions, so that the integrated 3G/WLAN network can provide more capacity and better performance while does not consume more bandwidth (of a 3G system) than a conventional 3G network (i.e., without WLANs). In other words, an incoming request blocked will be dropped immediately in the conventional 3G network but not necessarily dropped in the integrated 3G/WLAN network. Following this idea, we build our mathematics model in next section for performance evaluation of the integrated network. Six interworking scenarios were described in . The connectivity between cdma2000 and 802.11 networks was studied by a loosely-coupled interworking in , an internetworking architecture for HiperLan2 and UMTS networks was proposed. In , the complementary use of WLANs in conjunction with UMTS was presented. The capacity of the integrated system has been shown to significantly increase by a set of simulation tools when interworking between hotspots and 3G networks is employed. This is particularly true when a high percentage of the area is covered by the hot-spots. Unfortunately, all the above approaches for performance evaluation of the integrated network are based on simulation modeling, where the simulation parameters selected may become critical and cause a bias on the model or even make it leave the bounds of reality. Furthermore, to the best of our knowledge, there has been no performance evaluation approach based on analytical modeling for the same work. This is the focus on this paper. Specifically, we are interested in determining the quantitative performance measures of the integrated network based on general distribution of channel occupancy time for different classes of users and different types of sessions. The rest of the paper is organized as: section 2 proposes a mathematics model fitting the integrated architecture; section 3 presents the detailed analysis in terms of handoff arrival rates, channel occupancy times, steady-state probabilities and performance measures; section 4 presents the numerical results; and section 5 concludes the paper. II. MODEL DESCRIPTON Consider an integrated 3G/WLAN network consisting of a number of 3G cells. Within each cell there are some isolated