AI-Powered Video Perception Solution Enhances O&M Efficiency

With the rapid development of broadband and video coding technologies, big video IPTV & OTT services have become fundamental broadband offerings for telecom operators. As these services experience exponential growth in commercial use, the O&M complexity of service systems also escalates significantly. Traditional O&M mode focuses on manual analysis of service statistics reports and alarms, which is inefficient and time-consuming. However, it fails to adapt to the evolving requirements of big video services, such as monitoring performance indicators, node failures, and fault delimiting and locating. Therefore, a more intelligent O&M mode is imperative to improve operational efficiency and ensure the stability, reliability, and quality of big video systems.

AI technologies are introduced into the O&M of big video services, greatly improving their automatic and intelligent capabilities and providing the optimal technical solution for addressing current O&M challenges. On one hand, AI technologies enable the deep integration and unified presentation of various performance and network indicators. On the other hand, AI algorithms can be used to analyze massive and multi-dimensional monitoring data, enhancing the ability for fast fault detection and location.

Responding to the imperative need for enhancing the user experience of video services, ZTE integrates AI technologies to develop the BigVideo ezMon intelligent O&M platform (Fig. 1). It introduces a content distribution network (CDN)-based indicator system, including performance indicators, channel quality monitoring, intelligent scheduling, and log aggregation analysis, to meet the requirements for intelligent O&M in big video networks. Leveraging a big data platform and industry-leading Prometheus and Etcd architectures, ZTE’s BigVideo ezMon platform boasts high-performance computing and storage capabilities, professional data processing and analysis capabilities, accurate service quality perception, real-time monitoring and alarm analysis, log aggregation analysis, end-to-end network diagnosis, and intelligent and rapid fault location.

Performance Monitoring for Visualized Indicators

The BigVideo ezMon intelligent O&M platform can collect and monitor KPIs within seconds, offering visualized indicators for various services across multiple dimensions, encompassing the entire network, areas, nodes, and devices (Fig. 2). Users can quickly customize large-screen monitoring indicators and display modes (such as trend charts, histograms, and tables) as required. Through the correlation analysis of alarms and performance indicators, the platform can promptly identify the location of system alarms and their impact on services and users.

• Service indicators: In accordance with service monitoring templates across dimensions of the entire network, areas, nodes, and devices, the platform provides various performance indicators such as total number of concurrent users, service success rate, data backhaul success rate, bandwidth usage, delay of first service packet, request hit rate, and download rate.
• Hardware indicators: The platform provides unified hardware monitoring indicators such as CPU load, memory usage, storage usage, NIC IO rate, hard disk IO rate, and NIC packet loss rate.
• Network indicators: The platform provides switch monitoring indicators such as interface traffic and packet loss.
• Health overview: Through the correlation analysis of alarms and performance indicators, the platform intuitively presents the health overview of various service systems according to dimensions such as the entire network, areas, and nodes.

Channel Monitoring for QoS Assurance

The BigVideo ezMon intelligent O&M platform provides OTT channel monitoring capabilities to obtain real-time QoS data for each monitored channel, quickly identify abnormal services, locate channel faults, and determine their impact scope.

• Service topology monitoring: The platform monitors channel service data in real time, aggregates data by node and device, and displays channel QoS on nodes and devices in real time.
• Channel topology monitoring: The platform draws the backhauling link based on the actual service of the channel, visually analyzes the service indicators of the nodes and devices on the backhauling link, and quickly identifies abnormal services.
• Channel quality analysis: The platform supports the analysis of channel service indicators and impact analysis on areas, nodes, and devices. It visually displays the service change trends within a specific time period, and rapidly locates the impact scope of channel faults.

Intelligent Scheduling for Load Balancing

The BigVideo ezMon intelligent O&M platform provides CDN intelligent scheduling decision-making capabilities. It uses AI algorithms to analyze service data in various scenarios such as intra-area, inter-area, live broadcast hotspots, and VOD hotspots to obtain scheduling policies. These policies are then delivered through automatic interfaces to achieve more accurate and timely intelligent scheduling.

• Intra-area load scheduling policy: Based on the load of each node within an area, the platform uses a multi-dimensional dynamic algorithm to provide scheduling suggestions for user groups, node service priorities, and weights, thus implementing the intra-area scheduling policy for multi-node load balancing.
• Inter-area load scheduling policy: Based on the actual load and uplink/downlink load capabilities of each area, the platform uses a multi-dimensional dynamic algorithm to provide scheduling suggestions for user groups, node service priorities, and weights, thereby implementing the inter-area scheduling policy for load balancing.
• Live hotspot content prediction and scheduling: The platform analyzes and predicts hotspot live broadcast channels based on the regional dimensions. It then provides distribution policies for these channels, aligning with node service capabilities to facilitate rapid channel creation and rollback scheduling. In addition, it supports configuration of hot channels for special events, ensuring high-priority scheduling for live broadcast hot channels.
• Hot VOD content analysis and scheduling: The platform analyzes hot VOD content regionally and provides a distribution policy based on node service capabilities. It can also facilitate rapid VOD creation, duplication, rollback, and cross-area propagation to improve hit rates.

Log Aggregation Analysis for Enhanced Fault Location Efficiency

The BigVideo ezMon intelligent O&M platform remotely collects service and system logs from each device, aggregates and analyzes them, and rapidly retrieves end-to-end service access information. It supports multiple query and filtering conditions, such as IP address and time range, enabling retrieval of process logs from each service module to identify potential exceptions in the entire service process. This enhances the efficiency of fault delimiting and locating.

In a big video project in Turkey, routine O&M involves over 300 servers and more than 30 services, resulting in a heavy workload. Traditional O&M methods could not effectively cover all devices and services within the existing network, leading to passive management and difficulty in identifying operational risks in advance. Addressing these challenges, ZTE deployed the BigVideo ezMon intelligent O&M system. This system enabled real-time large-screen monitoring of network-wide device services through a single portal, facilitating early detection of potential network risks and accurate fault location. This implementation of active O&M significantly reduced the workload of on-site maintenance personnel and improved O&M efficiency by more than 50%. With abundant O&M functions and user-friendly operations, the BigVideo ezMon intelligent O&M system provides simplified O&M experience for front-line personnel and visualizes network quality for customers, winning recognition from both front-line personnel and customers.

As the big video service market continues to expand, operators face challenges such as high user experience expectations and the need for improved O&M efficiency at lower costs. Therefore, intelligent network O&M has attracted much attention. ZTE’s BigVideo ezMon intelligent O&M platform solution offers distinct advantages in improving network quality and maintenance efficiency. It is poised to be widely used in future network construction and optimization endeavors, helping customers implement intelligent network O&M.