The Research Of Key Technologies For Performance Optimization Of Large-scale Wireless LANs

With the birth and development of wireless LANs, as well as popularity of smart phones, tablet PCs, notebook computers and other mobile wireless terminals, and in order to meet the growing demand for mobile data services, telecom operators, schools and companies laid large numbers of large-scale wireless LANs on a global scale. The large-scale public wireless LANs have the following obvious characteristics:1) wide coverage;2) mobile terminals oriented;3) large number of users, widely distributed;4) Data services etc. In addition to the problems faced by the traditional WLANs, the large-scale WLANs are face with more problems, such as scarcer resources, overlaped network coverage, higher user traffic, more transmission errors and lack of fairness etc. These problems seriously hampered the development of large-scale WLANs. In order to promote large-scale wireless LANs, this paper proposed a set of methods of discovering, analyzing and solving problems for WLAN optimization in large-scale wireless LANs. In this paper, the error source diagnostics, traffic characteristics analysis, and network performance optimization are the key technologies to solve the problems encountered by large-scale wireless LANs.Firstly, analyzing of error causes and locating the error sources are needed in the performance optimization for large-scale wireless LANs. As a result of the characteristics of large-scale WLANs, problems such as low signal-to-noise ratio, collision, and small-scale fading have seriously impeded the performance of WLANs. Thus, analyzing and locating these three types of errors are required, and errors caused by low SNR and collisions can be avoided by appropriate methods, while there is no effective way to prevent small-scale fading, therefore different transmission errors must be distinguished. This paper presents a cross-layer analysis method based on the combination of received channel power sampling at the physical layer and information extracation at the medium access control layer. The proposed method analyzed the causes of error frames by recording samples of received channel power at the physical layer on a small time scale and employed the particle filter-based joint likelihood ratio method to detect changes in the received channel power and do model matching within the time domain. At the same time, it determines the source and the destination addresses of the error frames by decoding packet physical addresses at the MAC layer and then locates the error source.Secondly, the study of traffic characteristics, modeling and forecasting in large-scale WLANs, in connection with variation of traffic, can make us design more effective network modele and optimization methods. In this paper, we studied the traffic data in several large-scale wireless LANs with more than150access points and discovered that Granger causality existed between traffics at different access points. The Granger causality illustrates that the historical traffic of access points that have causality help predict the future of target access points with better accuracy than when considering information from the past of target access point alone. Bayesian network, also known as causal model, can reflect the characteristics of Granger causality and we used a Bayesian network to model the causal relationship between access points and adopted a Gaussian mixture model, as well as a weighted combination of several normal distribution functions in order to approximate the joint probability distribution in Bayesian networks, In order to verify the accuracy of the proposed method and compare the accuracies of the two methods, we, respectively, imported the data generated by a famous model and the traffic data in two large-scale wireless LANs, having hundreds of access points, into the proposed method and the Wavelet Analysis and Autoregressive Integrated Moving Average (WARIMA) model which is considered as the best method for prediction of traffic as far.Finally, after studying the error source diagnostics and traffic characteristics in large-scale wireless LANs, this paper presented the user traffic based proportional fairness model, using the power control and frequency selection methods to optimize and maximize performance of WLAN, while ensuring that allocation of each user’s bandwidth was fair. For network modeling, as considered the uplink and downlink traffic simultaneously, built a more realistic user traffic based proportional fairness model based on previous research findings of the flow characteristics of wireless LAN. This paper mathematically proved the theoretical upper limit of this model and the conditions to achieve this limit. In this paper, power control and frequency selection methods were both used to improve the traditional proportional fairness optimization algorithm. Based on the actual traffic data described above, various simulation experiments were taken to verify the method is more efficient in comparing with a variety of related optimization methods for optimizing throughput, latency, and bandwidth allocation of large-scale wireless LANs.This paper provided research reference for basic methods for systematic optimization of wireless LANs performance according to the characteristics and problems of large-scale WLANs by studying these three key technologies mentioned above. The proposed methods can fully use the barren radio spectrum resources, and improve network performance without changing the existing IEEE802.11protocols.