Large-Scale and High-Dimensional Cell Outage Detection and Compensation Techniques in Self-Organizing Networks
The objective of this research is on the self-healing in 5G Self-Organizing Network (SON). The key concept of the self-healing is to automatically detect the failure or degradation of base stations, and to automatically compensate it to mitigate its effect on the overall mobile communication networks and to maintain the service quality. That is, the self-healing is to provide the mobile communication networks with the rapid recovery capability. The self-healing can be divided into two sub-problems, including the cell outage detection and the cell outage compensation. The cell outage detection is to automatically detect whether there exist some failures or degradation in the base stations, such that users could not obtain mobile services, or the obtained mobile services do not fulfill their requirements. The cell outage compensation is to automatically adjust the parameters and configurations of each neighboring base stations to recover or to compensate the coverage of the outage cell, and to provide services to users in the outage area.
Self-healing in 5G SON is with great challenge. The deployment of future 5G mobile communication networks is expected to be heterogeneous and ultra-dense. The mobile communication environments are very complicated. They include the multipath transmission, fading, shadowing, interference, and so on. Users’ mobility and usage pattern also vary. In such environments, the mobile data would be large-scale and highly dimensional. Traditional small-scale and low-dimensional anomaly detection methods would be unsuitable. Moreover, operational mobile communication networks should be normal almost all the time. Cell Outage would be seldom. Therefore, the normal data and anomaly data would be imbalanced. Finally, the decision of one base station would affect the performance and measurement of the other base stations. Therefore, the parameter adjustment of one base station might not get precise and real-time feedback. Moreover, inter-cell interference would also increase in the cell boundaries. Finally, users usually do not want to provide their location information to mobile service providers due to privacy concern. All these factors make self-healing in 5G become very challenging.
In this research, we investigate the deep learning and the reinforcement learning techniques to deal with these problems. First, we propose the autoencoder-based cell outage detection method to solve the cell outage detection problem in 5G Self-Organizing Networks. The proposed method trains the autoencoder neural network by the measurement reports from mobile stations and compares the reconstruction error of a new measurement report with the decision threshold to predict whether there exists a cell outage. Simulation results validate the effectiveness of the proposed method. We further investigate the combination of cell outage detection and convolutional neural networks and propose the convolutional autoencoder-based cell outage detection method. Based on this improved method, the number of parameters in the model could be reduced, and the overfitting problem could be avoided. Simulation results show that the improved method could achieve lower reconstruction error and thus better classify the normal data with the outage data.