Enhancing User Perception with uSmartNet: Achieving a Leap in User Experience

With the rise of the metaverse, big data and generative AI are gradually penetrating into diverse sectors, creating new application fields. Meanwhile, the global digital economy is developing rapidly, and the construction of 5G networks is speeding up. Communication networks are becoming more complex, and the number of applications is increasing. People have raised higher requirements for network quality, and user experience has become the key to the success of communication network businesses.

A conventional KPI defined based on an event trigger counter cannot visually reflect actual user experience of a network. As a result, the focus of network quality improvement is gradually shifting from the conventional target of improving network KPIs to the core target of improving user experience. For operators, there is an urgent need to establish a network indicator identification and optimization system based on user perception, and ensure that it provides efficient and accurate functions in daily networks.

Rapidly identifying and solving user perception problems, and achieving predictability and prevention of such problems, is an ongoing goal pursued by the technical team of ZTE Global Services in the digital and intelligent era. Through the delivery of a large number of commercial network projects around the world in recent years, ZTE has accumulated comprehensive service capabilities in platforms, personnel, knowledge, and processes. The company has developed an end-to-end user perception identification, improvement, and prediction solution based on uSmartNet, with rate and latency perception as its core objectives.

USmartNet is an autonomous network platform that integrates big data and AI technologies, providing end-to-end coverage for communication networks from single domain to cross-domains (Fig. 1). The platform provides insights into data from multiple domains including users, terminals, services, networks, and operation,  supporting the digital and intelligent transformation of operators. In terms of ensuring user experience, uSmartNet uses a comprehensive KPI+KQI+QoE approach to establish a user perception evaluation system to accurately restore service perception. Leveraging the perception analysis and processing capabilities of big data, wireless networks, bearer networks, and core networks, it builds an end-to-end user perception analysis and optimization system to demarcate and locate end-to-end quality problems, analyze user perception and network complaints, predict perception problems, and handle the problems in a closed loop quickly.

uSmartNet Enables Intelligent Identification and Location of User Perception Issues

Identifying user perception problems in the network in a timely manner is key to ensuring user experience. uSmartNet can collect the control-plane and user-plane data from the wireless, bearer, and core networks, and clean, correlate, and integrate the data to form a high-quality perception database. To solve the increasing problem of service identification rates for new applications, AI capabilities are introduced into the probe system at the data collection layer. Through machine learning, new applications can be identified automatically, boosting the overall service identification rate of the system to over 90%. This provides a reliable basis for analyzing perception problems.

With the support of complete, reliable, and high-quality data, the platform quantifies the user experience for various service categories like network browsing, video streaming, gaming, OTT video, and OTT voice. Using appropriate mathematical models, it comprehensively evaluates the overall user experience of data services. By means of cascading, time sequence, space domain, aggregation, and failure reasons, the platform demarcates problems with low perception scores to specific network domains.

To locate user perception problems more quickly and accurately, the platform relies on AI capabilities and expert experience to continually explore and expand its ability to delve deeper into the root causes of the problems. On the wireless side, it analyzes and locates the root causes of cell faults, coverage issues, interference, and capacity constraints based on wireless MR, CDT, alarms, and performance data, and provides corresponding solutions. On the bearer side, it locates problems to specific links based on the delay, jitter, and packet loss measurement data of each link, and more accurately identifies and locates the problems that affect user perception in accordance with the single-domain capabilities of radio, bearer, and core networks.

Expertise Combined with uSmartNet's Perception Analysis Capability: Efficiently Resolving User Perception Issues

To solve the perception problems that affect user experience, such as rate, delay, voice quality, and video quality, ZTE has developed comprehensive and practical functions that facilitate the improvement of end-to-end user perception. In terms of voice services, functions like intelligent pre-scheduling, intelligent resource allocation, and adaptive handover optimization effectively improve MOS and latency indicators. For data services, functions such as intelligent pre-scheduling for web browsing, intelligent AMC, adaptive CFI optimization, intelligent downlink scheduling, user location-based load balancing, and service-based inter-frequency handover all  contribute to a notable improvement in perception-related elements like data rates and latency.

Good network quality is the basis for guaranteeing user experience. Based on the concept of self-discovery and self-optimization, ZTE continually advances the autonomous capability of network maintenance to enhance network quality. In a latency perception optimization project for an operator, ZTE performance experts leveraged uSmartNet’s big data perception analysis capability to comprehensively analyze the reasons behind variations in user-perceived latency and performed multi-dimensional analysis on the latency indicators for key services like games and videos that customers are concerned about. This approach enabled ZTE to quickly demarcate and locate problems, thereby enhancing network quality in the areas of weaknesses. From the perspectives of time domain, area domain, service, and NE, ZTE formulated a cross-domain end-to-end optimization strategy covering wireless, microwave, IPRAN, core network, IPCORE, international ISP links, and CDNs, and devised detailed latency enhancement plans for each product category. Through measures such as poor-quality service identification and rectification, traffic balance, international link optimization, CDN cross-traffic optimization, transmission link optimization, and air interface scheduling optimization, the user’s experience about latency was greatly improved. This propelled the operator to the top position among many operators in China.

Excellent network service quality can enhance user satisfaction and create brand value. The insights into market competition provided by third-party authoritative communication network testing and analysis organizations play an important role in the competitiveness of service quality that operators focus on. Organizations such as Tutela, Ookla, and Opensignal collaborate with widely-used market apps and secure necessary permissions. When a user opens an app authorized by a crowdsourcing enterprise, the app will track the base station connected to the mobile phone, measure the signal strength, and send this data anonymously back to the dedicated server deployed globally. The crowdsourcing platform can then use this data to analyze the network quality of different operators in the same area and generate a report on network perception quality analysis.

Though crowdsourcing data contains a large amount of third-party network data, it is limited to user terminal data and lacks network device data. As a result, it cannot be directly used to demarcate and locate perception problems. Operator only knows the results, but they cannot identify root causes or provide tailored solutions. In response, ZTE comprehensively explores the impact of users, terminals, NEs, and servers on crowdsourcing test results from multiple dimensions of big data. This approach identifies weaknesses and provides robust support for operators to enhance network evaluation.

In the uSmartNet solution, the "Digital-Intelligent Brain" explores the hidden relationships between crowdsourcing measurements and network performance, uncovering critical network performance and parameters. It deduces optimal parameter configuration models by using algorithms like random forest and decision tree to demarcate and locate related network problems. It automatically calculates network factors that affect perception, formulates solutions, and provides direct recommendations for adjusting parameters at the cell level. Through the connection with wireless EMS, it issues commands to adjust parameters, achieving a fully automatic closed-loop of the entire process. This application has demonstrated outstanding performance in practical use and has been featured in the 2022 TMF release of the "Autonomous Network White Paper V4.0" thanks to its successful implementation in an Indonesian telecom operator.

uSmartNet Empowers Precise User Satisfaction Prediction

At present, the trend in network operations is gradually changing from passive response to proactive guarantee. When user complaints arise because of perception issues, it means that network problems have escalated to a point where their impact cannot be ignored, often requiring last-minute solutions. To proactively identify, locate, and handle user perception problems, it is necessary to shift the focus towards “preventive treatment” measures. To this end, ZTE has built a customer satisfaction scoring system (CSAT) that combines multi-dimensional indicators such as key quality indicators (KQI), coverage quality, and customer complaints. This system also incorporates AI capabilities to predict future satisfaction trends in different regions, ensuring an average accuracy rate of over 90% when forecasting the trends of various indicators over the next week. Maintenance personnel can then optimize areas with low satisfaction in advance, actively ensuring user experience.

After CSAT was deployed on the uSmartNet platform of an operator in a country, it automatically detected potential drops in satisfaction within a specific region and issued early warnings. After troubleshooting, it is confirmed that the interference problem existed in this region, with a discernible upward trend. Maintenance personnel optimized the performance, solved the interference problem, resulting in significant improvements in various indicators for that region. Video download rates increased by 3%, regional data traffic grew by 7%, and user satisfaction notably rose. This proactive approach prevented a large number of complaints arising from poor user perception. This system has been embedded in the routine optimization process of operator network maintenance. During the quarterly optimization task planning, the maintenance team devises optimization strategies and solutions based on the satisfaction scores of the top N areas identified by the CSAT function as having poor satisfaction, along with the areas with low satisfaction predicted by AI. The CSAT function is also used for closed-loop verification of optimization results.

Throughout the entire process of network construction and operation, ZTE prioritizes enhancing user perception. It tailors network performance enhancement solutions to match the unique characteristics and requirements of operators. Over the past decade, ZTE has driven the development of high-quality networks. It has now achieved a multitude of leading network quality benchmarks in over 100 countries and regions, providing reliable user perception guarantee for operators worldwide. Drawing upon its rich experience in global network operations as well as its collaborative capabilities driven by digitalization and intelligence, ZTE will continue to help operators accelerate their network automation and intelligence. This effort aims to reduce operational costs, increase efficiency, expand revenue streams, and empower digital transformation of the industry.