UniCare CEA: Creating Superb User Experience

LTE networks have advanced faster than expected in terms of handling mobile internet. The share of revenue from mobile data and internet increased from 17% in 2013 to 23.5% in 2014. Of all mobile users, 45.3% are broadband users, which is 12.6% more than in 2013.

According to the above statistics, the number of data users is exploding, and data services are rapidly replacing traditional voice services as major sources of revenue for operators. To respond to challenges created by data services, an increasing number of operators have shifted the focus of their O&M from network performance to user experience.

Key Phases of O&M Transformation

ZTE’s UniCare customer experience assurance (UniCare CEA) is a high-end service that guarantees rapid improvement of user experience when operators transform their O&M model. This service has two phases: establishing a user experience evaluation system and improving end to end user experience.

Phase 1: Establishing a User Experience Evaluation System

The method commonly used for building a user experience evaluation system is to install probes and analysis tools on network interfaces to collect raw data, inspect the service and extract a key QoS field through deep packet inspection (DPI), and work out single-user, single-service QoS indicators based on the service and field.

The results of network trials conducted by operators show that user experience evaluation is not accurate enough because of

●    poor network coverage. Users would consider the network to be poor if their mobile terminals do not display full signal bars. This kind of experience is not included in the current evaluation system.

●    inadequate service access. The current user experience evaluation system does not indicate whether access to a RAN, CN or service is successful or not, nor does it indicate how long it takes to gain access.

●    insufficient service usage. Problems occur during service usage, for example, websites cannot be opened, it takes a long time to open a web page or download a file, and video streams cannot be played smoothly. Currently, most services cannot be inspected. More than 10% of total traffic is uninspected, which means that service experience of some users is not evaluated. Non-subdivision of already inspected web browsing services like Netease newsreader app and Ifeng news app, or even non-subdivision of inspected sub-websites like headline news and sports news, would lead to an inaccurately evaluated user experience. The extraction and algorithms of indicators for some core services and applications is not accurate. This would also affect user experience evaluation.

Phase 2: Improving End to End User Experience

Once a mature user experience evaluation system has been built, operators can use the indicators provided by the system to verify user experience by VVIP, VIP, VAP, and roaming users or by regions. If user experience is poor or deteriorates, user experience needs to be analyzed and improved from end to end. Three steps are involved:

●    delimit problems. Clarify the NEs and corresponding departments that are responsible for degradation of user experience. The Terminal Department should take charge of terminal problems, and the Network Optimization Department or Wireless Maintenance Department should be in charge of radio problems.

●    locate problems. Relevant departments analyze problems and determine the causes.

●    solve problems. Work out ways to solve the problems. For example, optimize or expand the network to address coverage, capacity, and interference problems; troubleshoot problems, or provide differentiated services like PCC to improve user experience.

However, an end-to-end user experience analysis and improvement mechanism has not yet been fully developed. There are many reasons for this, but the primary reason is lack of experience and mature skills, which leads to a waste of human resources and inaccurate delimitation and location of problems.

UniCare CEA: More Accurate and Efficient

Drawing on 30 years of experience in telecom equipment R&D, O&M and network optimization, ZTE has rolled out its UniCare CEA solution that can help operators transform O&M more stably and rapidly. UniCare CEA has the following distinct features.

More Accurate User Experience Evaluation System

A network cannot receive service requests from a UE because of poor reverse coverage. The app probes installed on mobile terminals can collect terminal signaling and behaviors and use them to supplement relevant user experience data that cannot be obtained on the network side. The accuracy of a user experience evaluation system can be increased by 1–3%.

The MR/CDT data on the radio side can be used to supplement relevant user experience data in network coverage, RRC setup, and RAB establishment. System accuracy can be increased by 2–3%.

The DPI function can inspect more than 95% of traffic and sub-divide it into specific apps or trails that a user follows. For example, each time a user opens a website on the Netease newsreader app can be inspected.

Deep user experience indicators such as page display success rate/delay, and the number of video interruptions as well as shallow user experience indicators such as DNS success rate/delay, TCP handshake success rate/delay, website response success rate/delay, and download rate can also increase system accuracy.

More Intelligent Problem Delimitation and Location

UniCare CEA provides automatic end-to-end problem delimitation. Its core tool ZXVMAX can apply big data-based machine learning technologies to the telecom field and use the clustering algorithm to accurately and automatically delimit user experience problems such as investment processing and service quality problems. These problems can be delimited to a cell, MME, XGW, SP, or mobile terminal.

UniCare CEA can be used to automatically locate RAN problems. Problems on the RAN side associated with poor/overshoot coverage, capacity, parameter configuration, and network fault can be automatically located by correlating MR/CDT data, network configurations, and parameter configurations and using the automatic matching algorithm in the rule base.

Stronger Data Processing

ZXVMAX has been developed on ZTE’s enhanced big data platform. It consists of Hadoop, Impala, Spark and other core components (Fig. 1). ZXVMAX provides correlation analysis of huge data through clustered data storage and computing, and can greatly improve data analysis and query efficiency.