QoE Modeling

A Hot Topic

Quality of experience (QoE) is not a new concept. Every effort an operator or supplier makes in their business has the goal of improving customer satisfaction and QoE. Why, then, is QoE still an issue in network management? Why has it not been adequately addressed?

QoE is gaining attention for two reasons: IP transformation and pipeline telecom operators. In the TDM age, bandwidth was occupied exclusively, voice and SMS were controlled by operators, and quality of service (QoS) could be monitored by signaling at the control plane. Everything was controlled by operators. IP-based multiuser multiplexing and best-effort service has greatly reduced network costs for operators but has also created unstable QoS. Voice and video in the future will most likely be accessed through the browser, so traditional operators will become mere pipe providers. China Mobile’s self-operated services account for only five percent of its data traffic. Servers and terminal applications for Internet services are not controlled by operators; however, operators are eager to know what services are running through their pipes and how QoS is implemented. Usually, customers lodge complaints to the operator hotline rather than to an Internet company such as Taobao or Sina.  

An easy and accurate way of assessing QoE has not yet been found. Standards organizations and operators are conducting research into QoE issues and standards and have published research findings.

What is QoE Modeling?

The TeleManagement Forum (TMF) shows the relationship between QoE, QoS, key quality indicator (KQI), and key performance indicator (KPI) (Fig. 1). QoE can be affected by technical or non-technical factors.  QoS is represented by KQI and is what end-users experience directly. KPI represents the performance of sources (network equipment) and is invisible to users.

Customers experience services, so QoS best reflects QoE. Different services have different standards of measurement. The TMF has defined KQIs for VoIP and IPTV. KQIs for IPTV include voice quality, image quality, synchronization of voice and image, delay, and response speed. Each KQI is affected by different network nodes.

The European Telecommunication Standards Institute (ETSI) has also introduced standards for different services. ETSI TS 102 250 series defines QoS and trigger points for FTP, mobile broadcast, ping, POC, streaming, telephony, video telephony, HTTP, email, SMS, and MMS. A successful HTTP session begins with the first SYN command and ends with receiving of the last data packet with content. However, trigger points for the quality of voice and video streaming are yet to be defined.

The ETSI has also divided QoS into layers (Fig. 2). QoS includes network availability and accessibility, which are independent of services, service accessibility, integrity, and retainability. QoS also includes parameters for different services. Service access can be subdivided into three stages: network access, IP service access, and Internet access. Different stages have a different impact on QoS.

To determine the QoE, it is necessary to obtain QoS statistics. There are many Internet applications and thousands of Internet protocols that are constantly changing. It is very expensive to exploit and manage every transaction on the network, and an early description of QoS in ITU-T Recommendation G.1010 was very general.

ITU-T Recommendation G.1010 defined three key factors that affect QoE: delay, delay variation, and information loss (including packet loss and information loss caused by codec). Table 1 shows the tolerance of typical services to delay and information loss. QoS for each type of service can be given according to delay, delay variation, and information loss.

To measure QoE, some equipment vendors and standards organizations map KPIs of the network and services to KQIs and then map KQIs to QoE.  Because there are many KPIs and KQIs, the mapping relationship is a data matrix with parameters that can be determined by intelligent computer fitting of data from customer questionnaires.

How is QoE Modeling Applied?

Monitoring QoE is complicated. To give an objective QoE assessment and locate elements within the network or terminals that affect QoE, three-dimensional monitoring needs to be done.

ZTE provides service monitoring solutions for equipment at each layer of the network. With these solutions, operators can collect and analyze user traffic while monitoring KPIs for both equipment and services.

ZTE’s customer experience management center (CEMC) is an end-to-end service-monitoring and QoE assessment system. CEMC provides abundant interfaces that collect NE data from terminals, RAN, bearer, core network, and services.  Data collection is based on different levels of granularity: NE-level, user-level, and service-level. CEMC also has abundant adaptive interfaces that are capable of collecting data of a third-party vendor and in various formats. CEMC assesses end-to-end network accessibility, availability of bearer pipes, and key QoS indices using deep packet inspection. It monitors QoE at user or NE granularity, allowing a balance between investment into network operation and data accuracy.