white paper Communications Service Providers Service Assurance Wipro* Real-Time 5G User Experience Analytics Uses Intel Network Edge Virtualization SDK The need for service level agreements and consistent network performance on nextgeneration 5G services will make user experience analytics critical. Wipro leverages artificial intelligence and analytics deployed in a multi-access edge compute (MEC) cloud to interpret and react to changes in network conditions Table of Contents Introduction...1 MNOs Need Real-Time User Experience Systems for 5G...1 Real-Time Analysis Without Being Obtrusive...2 Wipro Edge Analytics Combines RF and UE Metrics...2 Intel NEV SDK Provides MEC Services...3 Conclusion...4 About Wipro...4 About Intel Network Builders...4 Table of Abbreviations...5 Introduction As the mobile network evolves from 4G to 5G, the industry is setting an end-user expectation for improved network experience and assured service levels. Mobile network operators (MNOs) feel they can deliver on this because 5G technologies are expected to have higher throughput, lower latency, and predictable service performance compared to 4G/LTE networks. The emerging concept of network slices which are multiple virtual network channels running on one physical media promises assured levels of throughput. User experience (UX) analytics differ from standard network management mechanisms in that they collect a much broader set of data that contributes more directly to the performance that a mobile user is experiencing. Whereas traditional network management uncovers important issues with network congestion and network latency, UX analytics are more focused on metrics such as the type of content delivered, the user s mobile device, user location, and other related data. These new service expectations are driving the need for mechanisms in the network to measure and assure that the committed service levels are met. This white paper explains how Wipro,* a member of the Intel Network Builders Edge Ecosystem, has developed an edge analytics application running on multi-access edge compute (MEC) servers built using Intel Network Edge Virtualization (NEV) SDK to create a set of analytic functions and artificial intelligence/machine learning (AI/ML) methods to ensure that the user experience level of a multimedia session is met as per the service level agreement (SLA). MNOs Need Real-Time User Experience Systems for 5G Unlike previous generations of cellular service that were focused on cellular phones and other mobile devices, 5G introduces a number of new wireless technologies and use cases. These range from millimeter wave fixed wireless access services operating at gigabit-per-second and higher speeds to slow-speed IoT applications to traditional cellular service, called 5G New Radio (NR). As a part of these technologies, 5G standards bodies are introducing a significant amount of user experience data that can be used in UX analytics. Processing this data in the core or a public cloud requires it to be transported to the core network before it can be processed. This increases the traffic on the transport network and can lead to congestion on that network. It also introduces latency that can slow the network s AI-powered self-correction mechanisms. Another aspect of 5G networks that could impact end user experience is network traffic profiles that must share the same physical network but have potentially conflicting network requirements. Some applications, such as large file transfers,
transport a lot of data in flows of large packets. These are known as elephant data flows because they can consume switch resources for a significant period of time. Mouse data flows, such as machine-to-machine communications, have small packets that need low latency. If too many mouse flows, for example IoT control signals, get caught behind elephant flows, such as a virtual machine migration, latency increases for the mouse flow applications. UX analytics can optimize the system throughput through the use of smart algorithms and network slicing to reduce the performance penalty from the system throughput perspective, while ensuring the end user experience of these kind of contrasting requirements. as they require a technician for any changes or moves throughout the life cycle of the network. The recent MEC standardization of open APIs allows development of a virtualized probe to get easy access to radio KPIs and associate it with the user application with minimal overhead impact. The European Telecommunications Standards Institute (ETSI)* recently launched the radio network information (RNI) application programming interface (API). When the RNI API is combined with the user equipment (UE) identity API, it is possible to get the radio network KPIs on a per-user basis. Availability of this information enables creative applications using network analytics to enhance the user experience. Real-Time Analysis Without Being Obtrusive User experience analysis must deliver the means to measure the service performance level in real time. But doing so directly could be a challenge as it will consume network bandwidth and impact user performance which can be intrusive to the end user. Hence there is a need to have a means to measure the user experience level passively (or transparently). Most analysis done today utilizes operational support systems/business support systems (OSS/BSS) that focuses on measuring the performance level of the network as post-facto information based on a set of network key performance indicators (KPIs) and service quality indicators (SQIs). In addition, some MNOs have implemented a customer experience management solution using an external probe in the core network. A key piece of these efforts is obtaining data from the radio frequency (RF) part of the network that measures the impact of the RF on user experience. Getting insight from the radio network layer requires a probe that has been implemented on the radio access network node. Implementing appliancebased network probes has been expensive and inflexible, Wipro Edge Analytics Combines RF and UE Metrics Wipro developed an edge analytics solution based on the RNI API principles to collect RAN L2/L3 KPIs via a software probe application running as a MEC service. Network analytics are performed on these captured KPIs with an aim to extract information that will improve the user experience. For example, the collected KPIs can be analyzed using a machine learning technique to detect the user experience issues associated with user session like a video streaming session. Once the user experience issue is detected, another machine learning algorithm can be used to understand the problems that may exist with an associated network segment and determine the right action to improve the user experience for example, a handover decision. If the decision is allowed to be implemented in a closed loop manner with the resource and service management functions, the situation can be auto-corrected for a better user experience, which can be measured with the mean opinion score (MOS value). More details on measurement of multimedia session quality could be found in ITU standard, P.910, J.147, J.246, J.247, J.341, J.342, etc. Analytics Engine RAN L2/L3 stack MEC App Service Software Probe VM MEC Platform MEC App (Analytics) Intel NEV SDK VM Figure 1. Wipro Edge Analytics solution architecture.¹ 2
Figure 1 depicts the Wipro solution architecture. It uses the Intel Network Edge Virtualization (NEV) SDK, which provides an environment to quickly deploy analytics on edge computing servers. In the solution, a software probe collects KPIs from different protocol layers from the radio access network (RAN) L2/L3 node, including MAC, RLC, PDCP, GTPU, and S1U. As it is collecting the data, the software probe uses deep packet inspection to classify the content. These KPIs are exposed to the MEC platform via MEC appservices and can be subscribed to by making API calls. The MEC platform then shares this information with the subscribed MEC applications and henceforth to the central analytics engine. These KPIs have real-time, low-level network and RAN information, and harvesting them offers a significant amount of information that can be used to improve the quality of experience. The analytics engine uses these KPIs to apply Wipro-developed machine learning algorithms that are tuned with the training data sequence for detection of the network performance level. Video quality is a good example of how the analytics solution can improve service quality. Video quality is typically measured by the mean opinion score (MOS) metric. MOS value is traditionally computed using an active probe to collect performance data or video subjective assessment by experts. However, using a software-based passive probe that can collect and report the network KPIs to an analytics engine that uses machine learning methods to detect the MOS value from the KPIs makes it possible to determine the quality issues in real time. The machine learning models are previously trained with the labeled KPI samples to enable real-time detection of the MOS values from the reported KPIs. The traditional quality monitoring using an active probing app running in the end user device poses the challenge on battery management and extra CPU load. Once this is applied to the detection of MOS for an inservice video session, the associated KPI values provide vital information on the possible actions that are available to improve the video quality. The insights that are made by the analytics engine can be made available to a network orchestrator for possible remediation, and if the orchestrator is authorized to effect the recommended action, it will be possible to tune the network parameters automatically to improve the quality of the video session. Intel NEV SDK Provides MEC Services MEC is a European Telecommunications Standards Institute (ETSI) initiative to open up the radio access interface for services and applications in the radio access network (RAN) such as edge analytics. This location in the network edge offers an environment with low latency, high bandwidth, and direct access to radio information, such as subscriber location. Due to the proximity of the mobile edge server, responsiveness to applications and services is increased, improving the quality of experience. NEV SDK MODULE DESCRIPTION MEC reference libraries Wind River Titanium Server* components Wind River OpenStack* Data Plane Development Kit (DPDK) Intel application development tools Network Traffic Services (NTS) Network Edge Services (NES) Network Information Services (NIS)1 NES API1 Accelerated Virtual Switch (AVS) Open Virtualization Profile (OVP) OpenStack virtualization management Open source Intel System Studio 2015 Also available in 90-days license. Processes the packets between the enb and the EPC over the S1 interface. Including IP identification, routing, and GTP decapsulation and encapsulation. This shields the application layer from the lower transport layer 4G protocols used to transfer data on the S1 interface. Responsible for interfacing with the VM, and service registry. Extracts and stores radio network information, and provides it to the upper layer applications via the NES. Interfaces with the MEC applications to provide services. Provides communications between applications running on different virtual machines (VMs) and enables very fast packet throughput. An embedded virtualization solution that combines a real-time open source kernel with a proven Wind River carrier-grade Linux* distribution. This integrated platform delivers a low cost solution that is capable of supporting the high throughput, low latency, and deterministic requirements for 4G network traffic. Removes the need for the application developer to set up the virtualization environment. A set of libraries and drivers that optimizes packet processing on Intel architecture. Tools that enable the developer to accelerate product development, optimize performance, and debug code. Table 1. Intel NEV SDK components 3
White Paper Wipro* Real-Time 5G User Experience Analytics Uses Intel Network Edge Virtualization SDK The Intel Network Edge Virtualization (NEV) SDK provides a network functions virtualization (NFV) platform targeted for multi-access edge computing applications and services. The Intel NEV SDK takes advantage of an Intel Xeon processorbased server that is configurable with real-time virtualization software and Intel s edge computing reference libraries for directing radio traffic information to the virtual machines based on policy settings. The Intel NEV SDK comes with a full tool suite for testing and profiling the applications. MEC servers utilizing Intel NEV SDK support new 5G standards that facilitate edge analytics. In the recent 5G system specifications, 3GPP* has introduced a couple of new functionalities that serve as enablers for edge computing that are essential for integrated MEC deployments in 5G networks. These enablers include local routing and traffic steering, support of local area data network, and session and service continuity to enable UE and application mobility. In addition, there is the ability of applications to influence user plane function (UPF) (re)selection and traffic routing directly via the policy control function (PCF) or indirectly via the network exposure function (NEF). Conclusion Ensuring a great user experience will be very important in a 5G era. To help MNOs achieve great 5G customer service experience, Wipro has developed a MEC application using Intel NEV SDK platform to enhance user experience using edge analytics and machine learning algorithms. MNOs can be expected to want future detection and correction of user experience issues to be measured in MOS score of less than a second. Another example of great user experience could be to assure an MOS score greater than 3.5 at all times. The MEC platform enables new possibilities in offloading network data and allowing local processing to react and change network components before these components impact users. Learn More: Please contact https://www.wipro.com/contactwipro/ to get in touch with Wipro for further details. About Wipro Wipro Limited (NYSE: WIT, BSE: 507685, NSE: WIPRO) is a global information technology, consulting, and business process services company. It harnesses the power of cognitive computing, hyper-automation, robotics, cloud, analytics, and emerging technologies to help our clients adapt to the digital world and make them successful. A company recognized globally for its comprehensive portfolio of services, strong commitment to sustainability and good corporate citizenship, it has have over 160,000 dedicated employees serving clients across six continents. Together, its employees and clients discover ideas and connect the dots to build a better and a bold new future. About Intel Network Builders Intel Network Builders is an ecosystem of infrastructure, software, and technology vendors coming together with communications service providers and end users to accelerate the adoption of solutions based on network functions virtualization (NFV) and software defined networking (SDN) in telecommunications and data center networks. The Intel Network Builders Edge Ecosystem is a new initiative gathering ecosystem partners with a focus on accelerating network edge solutions. As an integral part of the broader Intel Network Builders program, this initiative aims to facilitate partners access to tested and optimized solutions for network edge and cloud environments. Learn more at http://networkbuilders.intel. com/networkedgeecosystem
TABLE OF ABBREVIATIONS 3GPP AI/ML API ETSI GTPU KPI 3rd Generation Partnership Project Artificial intelligence/machine learning Application programming interface European Telecommunications Standards Institute GPRS tunneling protocol Key performance indicators L2/L3 Layer 2/ layer 3 LADN MAC ME MEC MNO MOS NEF NEV Local area data network Media access control Mobile edge Multi-access edge compute Mobile network operators Mean opinion score Network exposure function Network edge virtualization NR OSS/BSS PCF PDCP RAN RF RLC RNI S1U SDK SLA SQI UE UX UPF 5G New Radio Operational support systems/business support systems Policy control function Packet data convergence protocol Radio access network Radio frequency Radio link control Radio network information RAN to core user interface Software development kit Service level agreement Service quality indicator User equipment User experience User plane function ¹ Figure provided courtesy of Wipro. Intel technologies features and benefits depend on system configuration and may require enabled hardware, software, or service activation. Performance varies depending on system configuration. No computer system can be absolutely secure. Check with your system manufacturer or retailer or learn more at intel.com. Optimization Notice: Intel s compilers may or may not optimize to the same degree for non-intel microprocessors for optimizations that are not unique to Intel microprocessors. These optimizations include SSE2, SSE3, and SSSE3 instruction sets and other optimizations. Intel does not guarantee the availability, functionality, or effectiveness of any optimization on microprocessors not manufactured by Intel. Microprocessor-dependent optimizations in this product are intended for use with Intel microprocessors. Certain optimizations not specific to Intel microarchitecture are reserved for Intel microprocessors. Please refer to the applicable product User and Reference Guides for more information regarding the specific instruction sets covered by this notice. Notice Revision #20110804 No license (express or implied, by estoppel or otherwise) to any intellectual property rights is granted by this document. Intel Corporation. Intel, the Intel logo, and Xeon are trademarks of Intel Corporation or its subsidiaries in the U.S. and/or other countries. * Other names and brands may be claimed as the property of others. 1118/DO/H09/PDF Please Recycle 338461-001US 5