A Research On Cell Outage Detection In Wireless Communication

With the explosion of mobile equipment and the demands for user quality of service(QoS),the challenge of performance optimization and network maintenance is upon us.Besides,the deployment of Long-Term Evolution(LTE)networks over traditional networks raises challenge of handling the complex scenario of heterogeneous networks.Self-Organizing Networks(SON)have been drawing considerable attention recently to solve these problems,cell outage detection is one of the main tasks of self-healing in SON.Cell outage detection in wireless communication includes different kinds of fault causes and key performance indicators,the detection and diagnosis of base stations in homogeneous networks have been researched a lot.However,as a result of poor performance of traditional approaches in HetNets and the structural restriction of separated data and control layer,more effective outage detection approaches should be further researched.A SON solution for cell outage detection using a cooperative prediction approach is proposed in this thesis,which is designed for small cells in dense HetNets.The triggering phase based on collaborative filtering and the detecting phase based on grey model make full use of the spatial and temporal correlation respectively.In the triggering phase,each base station runs the triggering algorithm with the reported RSRP statistics from its associated users and reports the results to the corresponding neighbor cells to monitor their states.In the outage detecting phase,an triggered cell informs its neighbor cells to run the detection algorithm and report the results back for final decision.The triggered cell with higher abnormal results rate than the pre-defined threshold is decided to be in outage.The simulation results demonstrate that the proposed scheme is able to achieve higher detection accuracy even when there are few or none users in the problematic cell at the sacrifice of computation in neighbor cells.Meanwhile,raw data sets are not required to be collected for the centric computation and the computation is carried out in the serving base stations,which outperforms the existing statistic and data-mining based schemes designed for macro,micro or pico cells in communication overhead.It is necessary to set up threshold parameters in the cooperative prediction approach,which calls for sufficient experimentation and manual intervention in different network scenarios.In addition,it is difficult to explore the correlation of features and achieve accurate outage detection based on both spatial and temporal user data with a linear classification and prediction model.In order to reduce the influence of uncertain parameters and achieve better classification of non-linearly separable user data,this thesis further researches on the radical basis function neural network for the application in the outage detection problems.The decision tree base learners for feature selection and artificial bee colony algorithm combined with mutation for global optimization of parameters are applied to improve the performance of the neural network classifier.The cooperative approach relying on the operation of neighbor cells is able to achieve decent performance in dense small cells with sparse users,improving the detection accuracy and reduce the transmission of user data.The simulation results demonstrate that our approach is able to achieve reasonable detection results even with sparse users.Meanwhile,the detection performance is significantly improved with higher density of base stations.