PARTNER FEATURE: In the communications industry, network optimization is an indispensable link for operators since network construction. Continuous optimization and adjustment is the work with the longest cycle, the highest cost, and the most complicated operation. With the large-scale construction of 5G network and the introduction of new technologies, the network structure becomes more and more complicated, and various problems are more serious. Traditional manual optimization methods have become more and more difficult, with lower accuracy and efficiency, while higher and higher investment costs. Therefore, major operators and equipment vendors are actively exploring how to change their ideas to improve network optimization efficiency, and reduce optimization costs.

To solve the problems of traditional optimization methods, ZTE proposes the intelligent optimization service solution.

ZTE’s intelligent optimization service solution is based on self-developed VMAX-AI system, which can automatically collect various types of historical and real-time network data integrated with NMS, and achieve the automatic, intelligent and closed-loop network optimizations for operators with a series of in-depth learning algorithm models, which are continuously revised in practice. Such as k-means, GMM and graph theory, etc..

Firstly, traditional network optimization relies on the experience of network optimization experts to continuously adjust network parameters, compare and analyze front-end drive tests results and back-end NMS KPIs. An optimization cycle usually takes 1-2 months, and the efficiency is very low. The intelligent optimization service solution uses the special AI model algorithm to implement efficient iterative calculation, and can quickly obtain the optimal cell parameters within one week or even one day. It avoids the dependence on experts and the tedious manual adjustment, and increases the optimization efficiency by nearly 10 times. Taking the intelligent co-channel interference optimization and RCA-based subscriber rate optimization of a telecom operator for example, it takes less than one week to automatically output the antenna adjustment scheme, which increases the rate of edge subscribers by 22.9%, and the average subscriber rate below 5% in the region by 15.84%.

Secondly, due to the limitations such as human resources for traditional network optimization, it is difficult to do the optimization in complicated network scenarios. The parameter thresholds are relatively simple and the granularity of optimization is relatively coarse. In addition, it is hysteresis to solve issues by statistics and analysis with NMS or DT data performed manually or through some simple tools. The intelligent optimization service solution can realize cluster-level or cell-level dynamic threshold adjustment in accordance with various complicated actual scenarios, avoiding one-knife cutting parameter threshold configuration. In addition, through deep learning and training of historical data, the system can effectively predict the trend of KPIs in the future periods, prevent KPIs deterioration, solve problems which are difficult to find or deal with in traditional optimization, and improve optimization accuracy. In the pilot project of 5G KPI intelligent deterioration detection and root cause location carried out in an operator`s network, the threshold dynamic and detection accuracy of the abnormal detection algorithm are fully verified. The root cause location accuracy of the abnormal point reaches almost 100%. In the intelligent load unbalanced optimization project for another operator, the AI algorithm is used for prediction, dimension reduction and balancing optimization. The proportion of unbalanced sectors was reduced from 26.67% to 8.33%.

In addition, the intelligent optimization service solution is based on ZTE self-developed VMAX-AI system, which can be flexibly deployed with most simplified on-site server deployment or remote cluster deployment according to the on-site hardware configuration or customer requirements, compared with competitors or other tool systems. In a mobile operator project, only one server is required for 6000 cells to complete all AI function calculation and parameter delivery. The installation and debugging are much easier.

ZTE`s intelligent optimization service solution can implement automatic sample data collection, intelligent modeling analysis and automatic tuning parameters distribution to NMS. Almost no manual intervention is required during the whole process. While improving the efficiency, it avoids a large number of manpower stacking investment as traditional optimization, assists the operators in reducing the (HR) cost of Human Resources effectively. Since the solution has been promoted, many domestic and overseas operators have been greatly interested in it, and have issued field trials or optimized delivery requests. In recent pilot implementation in many regions of domestic and overseas, the optimization effect were unanimously recognized and praised.

ZTE is committed to continuously improving the accuracy and predictability of the algorithms of intelligent optimization service solution. The visualized interface and usability of the product operations is also in deployment. The solution will also enables operators to operate without any support, optimizes their costs and benefits.