Multi-attribute Decision Making Based On Optimal Heterogeneous Wireless Access Network

Nowadays,the mobile communication technology has changed rapidly.The meanwhile,many different types of networks such as GSM,CDMA,GPRS and WiMAX and WLAN are also developing fast.So how to chose the technology and use it select network access in networks fusion condition is hot and difficult point of current research Especially,there are kinds of access network coverage in the same area on heterogeneous wireless networks,but use of access technology and protocol standard is varies from access networks.Similarly,different services of network also requirement the different network performance index.For the sake of the best connection(always best connect)service,this paper presents a method of received signal strength regression prediction based on Information granulation SVM(Support Vector Machine).First,the received signal should be normalized to increase predictive accuracy,then implement of the regression prediction by using the SVM handle the signal granulation information,which can predict received signal strength in the next 2 time point of mobile terminal in a a relatively small range of error and high accuracy,so that we can advance of access selection and handover to reduce the delay of switch.However,in the field of heterogeneous wireless networks access selection,bandwidth,packet loss,delay and other performance indicators measure are different from varies access network.The proposed algorithm such as TOPSIS in relation research,there are a variety of network attributes are considered for the effect on network access selection,but the result of recommended network tends to be the one whose value of attribute is too large,that makes the recommendation network is not the most close to the optimal scheme of the network.In this paper proposes a network selection algorithm of multiple attribute closest to the optimal(MACO)solution relatively,the core idea of this algorithm is try to find a network has the maximum probability to replace the optimal scheme or relatively large at every attribute,so the access network is the closest to the optimal scheme.The proposed algorithm has verified in this paper,the experimental results show the result of MACO recommend network is the same as the TOPSIS.But in the simulation environment is put forward in this paper,we modified the bandwidth of LTE,results show that the average fluctuations of information gain of MACO algorithm is 1.06,while the the average fluctuation similar to the optimal scheme of TOPSIS is 1.52,compared with MACO algorithm to improve the stability performance of 30%.