Research Of Wireless Network Convergence And Access Selection Based On IP Flow

With the popularity of the intelligent terminal and the development of mobile Internet, mobile service data traffic is growing in a explosive way. This brings much pressure to the mobile network because of the limited network resources. In order to alleviate pressure on the network, domestic and foreign operators began to deploy 3G, LTE, WLAN networks and so on. Every wireless network has its own advantage, but none of them can meet all the requirements for kinds of services in the future. A combination of networks can make up the inadequate in a single network, and it also can take full advantage of wireless network resources. Heterogeneity and convergence will be the main features in future wireless networks.In this paper, we introduced an new IP flow-based convergence scheme through the research of wireless network convergence. On the basis of IP flow mobile technology defined by 3GPP, we make improvements to the relevant protocols and routing, mobile terminals can connect with multiple access networks in the same time and migrate traffic dynamically.High-speed and mobility are the main features of future traffic, WCDMA/WLAN interworking networks are considered to support this very well. The studies of this issue are focused on multi-radio access selection (MRAS). For MRAS methods, they can be generally parted into several types:traditional single-objective decision, police-based and utility function-based strategies.In this paper, we propose a multiple attribute decision making-based access selection (M-AS) to determine which network is most suitable for a mobile node’s service request in heterogeneous WCDMA and WLAN networks. The M-AS method takes five factors into account, network load, date rate, delay, packet error rate (PER) and mobility. Especially the M-AS method considers the different requirements of real-time services and non-real-time services, so it designs different weights for these factors respectively in different occasions in order to obtain better performance. Simulation results show that the M-AS method achieves smaller PER and delay, larger system throughput than the UFAS method which takes the same factors into account.