Network Aware Functions
Network Aware functions
Do you want to learn more about the Network Awareness Function?
Check on each of our Network Applications for more information!
You can also check also our Network Application Catalogue.
Network Awareness in 5G-IANA: What Is It and Why It Matters
Network awareness is a major advancement in 5G, making it intelligent and more responsive than previous telecom generations. It allows 5G to detect, understand, and adapt to changing network conditions, optimizing communication and resource use.
In the 5G-IANA project, this capability is built into the core of the 5G system based on four critical functions discussed next. Importantly, this built-in network awareness is designed with both backwards and forwards compatibility in mind, i.e. it can work with different telecom or networking technologies, including WiFi, older ones like LTE or any other ones in the future. This makes 5G-IANA one of the first efforts offering a network- and generation-independent solution for managing and monitoring networks based on native network awareness features across both old and new communication technologies.
The (i) Network Monitoring Application tracks and analyzes all network activity in real-time, helping network managers to understand traffic patterns and performance, enabling them to make informed decisions and optimize the network proactively, reducing the risk of unexpected issues. The application is composed of two atomic parts: (a) the Network Monitoring Component, which collects data from different sources like traffic mirrors and firewall logs; and (b) the Network Monitoring Driver (a Golang application) that gets real-time network status information using its REST APIs.
The (ii) Monitoring Virtual Network Function (VNF) further enhances network awareness by gathering data from various sources, including IoT devices, and making it available for analysis. It allows users to customize what data is collected and how often, providing valuable insights into network health and performance. Further to this, a more advanced approach (Network Status Monitoring) involves Distributed Machine Learning (DML) to achieve Predictive Quality of Service (PQoS). With help by DML, one may improve the accuracy and efficiency of network performance predictions by decentralizing the training process across various network and application functions, particularly within Vehicle On-Board Units (OBUs).
Linked to and enriched by the DML-enhanced monitoring above, the (iii) PQoS application leverages input predictions to forecast network behavior so as to ensure that the network can meet the high demands of modern services such as autonomous driving or real-time video streaming.
Last, the (iv) Active Network Monitoring module further complements the above by actively measuring and storing network metrics like bandwidth, making this data readily available for ongoing analysis and refinement of the PQoS application.
The Network Monitoring Network Application contributes to network awareness by analyzing and recording all network traffic, and then providing a single point of access to the stored data
The application is composed of two Atomic Components:
– The Network Monitoring component, built upon a dockerized version of ntopng, offers 360° network visibility by gathering traffic data from various sources such as traffic mirrors, NetFlow exporters, SNMP devices, and Firewall logs.
– The Network Monitoring Driver, a robust Golang application, retrieves real-time network status information from ntopng, using its REST APIs.
This integration ensures smooth extraction of critical network metrics, contributing to a deeper understanding of network utilization, traffic patterns, and overall performance. Through automated data retrieval, Golang efficiency, and Kafka broker communication, the system enhances network awareness by providing consistently updated data, optimal performance, and real-time streaming capabilities. Moreover, its configurability and flexibility allow users to retrieve monitoring parameters from diverse network architectures, enabling a proactive approach to network management and optimization.
- Description of the functionality
Monitoring VNF is in general a component used for receiving all kinds of data via https interface, to store these data and to expose them to interested consumers (e.g., other components) via https and mysql, depending on the certain situation, i.e., the type of data. The Monitoring VNF also includes its client-side sub-component which actually serves for collecting data (e.g., from IoT devices) and forwarding them to Monitoring VNF. One of such “collecting” devices is so-called qMON Agent which serves for network performance testing and monitoring, thus providing network-aware data to the Monitoring VNF. Additionally, the components may also provide values of multiple 5G radio parameters, however, it depends on the HW device the qMON Agent is deployed to. There are various configuration options available for the customization of providing network-aware data, e.g., a customer may select what kind of data will be collected and how frequently the collection should take place.
- The potential that this function brings to third-parties, not limited only to a specific UC example / cross-UC example, but in more detail (more practical examples)
The component enables insight into the network in terms of its performance which can be observed (i.e., tested and monitored) from multiple viewpoints. As already mentioned in the “Description of the functionality”, the Monitoring VNF includes qMON Agent which serves as a key sub-component for testing and monitoring: a (performance) test run between two qMON Agents (e.g., speed-test, latency test, etc.), while some monitoring cases requires one qMON Agent only (e.g., monitoring accessibility of a server (DNS, web), etc.). It is then up to the customer/user to define which paths and/or instances within the network are of its interest in order to deploy qMON Agents accordingly. Similarly, qMON Agent can report on 5G radio parameters as perceived by 5G device (UE), however, this functionality highly depends on the HW device the qMON Agent is deployed to (certain 5G modems may not be supported yet).
- Anything else you consider useful, please let us know so that we harmonize the descriptions.
The Monitoring VNF component has already developed interfaces enabling an easy integration with Analytics VNF component (based on Grafana) which provides various GUIs for observing data and/or performing (statistical) analyses.
- Description of the functionality
The Predictive Quality of Service (PQoS) application aims to predict application-level network metrics critical for services with stringent network demands. It utilizes an FML-trained LSTM model to predict the network’s spatio-temporal behavior. PQoS analyzes the last 100 data points to predictions. It focuses on application-level throughput and round-trip time (RTT) metrics.
- The potential that this function brings to third-parties, not limited only to a specific UC example / cross-UC example, but in more detail (more practical examples)
The LSTM model within the components can be substituted with an alternative model tailored to the specific service using PQoS. This flexibility allows users to customize PQoS features, such as past observations, and metrics to be predicted, according to their service requirements. This modular approach enhances the adaptability and versatility of PQoS across different applications. Moreover, the I/O APIs facilitate seamless integration into any environment, ensuring straightforward deployment and compatibility across diverse systems.
- Anything else you consider useful, please let us know so that we harmonize the descriptions.
This component is provided as open-source
- Description of the functionality
The Active Network Monitoring Module is a client-server solution designed to provide comprehensive active measurements for evaluating various network metrics, including maximum available bandwidth between two nodes, by using a python wrapper around the IPerf library. This module records and stores the collected data in a database, ensuring reliable and persistent data management. Additionally, the recorded metrics are accessible through an HTTP endpoint, providing easy retrieval and analysis of the network performance data, either of the full network monitoring session or for the last update.
- The potential that this function brings to third-parties, not limited only to a specific UC example / cross-UC example, but in more detail (more practical examples)
This component is a user-friendly application designed for measuring network bandwidth between two nodes. It operates on a peer-to-peer basis, ensuring direct communication and accuracy in data transfer rates.
- Minimum Granularity of 1 sec.
- Functionality similar to running ‘iperf’ sesión. Users familiar with ‘iperf’ will find this tool similar in functionality.
- Anything else you consider useful, please let us know so that we harmonize the descriptions.
This component is provided as open-source