SAMURAI

Smart 5G Core And MUltiRAn Integration


Description 

5G networks will meet the different requirements of new services and applications, such as IoT, virtual/augmented reality, autonomous cars, and precision agrobusiness. To deal with this diversity, multiple modes of operation, provided by different wireless access technologies, have been defined.

In addition, 5G networks are being developed under an intense softwarization process, characterized by the use of cloud, virtualization and programmability. This process is significant in access networks and even more notable in the 5G core. Given the many challenges, there are still several open issues, such as the integration of non-3GPP IoT access network technologies to a 5G core.

The SAMURAI project thus proposes to research, deploy and extend 5G systems, developing the software needed to demonstrate the integration of wireless access technologies into the 5G core. Additionally, the project will address issues related to the adoption of Artificial Intelligence/Machine Learning (AI/ML) as a critical component in the evolution of 5G networks. Although standardization institutions are advancing in defining a framework, there are still several gaps for the full use of AI/ML in 5G.

To overcome some of the most relevant gaps, the SAMURAI project will determine AI/ML algorithms and techniques suitable for problems in access and core networks, such as link adaptation to channel conditions, beam selection in millimeter waves and functionality positioning, in addition to advancing the state-of-the-art in data collection and use. Bringing together academic institutions and RNP, the project enables the assessment of AI/ML in a nationwide network, allowing the development of solutions that have predictable behavior and can be effectively adopted in production systems that leverage priority use cases in Brazil, such as online education enabled by long-distance networks.

Project goals 

The SAMURAI project aims to leverage the use of AI/ML techniques in different wireless access networks, as well as in the 5G core, which are found at different stages of development by the research groups in this project. The access networks considered are: millimeter wave networks, long-distance access networks and IoT access networks, as illustrated in Figure 1. When application requirements include high transmission rates, the use of millimeter wave bands is interesting, due to the large bandwidths. In this case, the use of intelligent beam steering, based on AI/ML techniques, is fundamental.

Figure 1. Access and core networks addressed in the SAMURAI project

In addition, the SAMURAI project aims to develop a scalable 5G core that allows the integration of 3GPP and non-3GPP technologies, and that facilitates the use of AI/ML techniques in the various layers. The project also aims at the development and evaluation of AI/ML techniques applied to the various components of the 5G network. The expected results will result in contributions to the field of mobile networks, ranging from the availability of a 5G core with AI/ML, to solutions for the integration and optimization of wireless access networks employing AI/ML techniques.

Development

The 5G system is based on cloud computing concepts, whose core is service-oriented with native support for network resource slicing, virtualization and mobile edge computing [3GPP-TS-24.502]. Aspects of 5G architecture allow virtualized deployment of 5G networks which can be fully distributed, redundant and scalable. Furthermore, the various instances of network functions can be present individually or together with services and microservices, representing the fullness of this new generation of mobile networks.

As described by 3GPP Release 15, the main functions are: Session Management Function (SMF), Mobility Management Function (AMF), User Plan Function (UPF), Application Function (AF), Policies (PCF), Unified Data Management (UDM), Authentication Server Function (AUSF), Network Exposure Function (NEF), Network Function Repository (NRF), Network Slice Selection Function (NSSF) , Non-3GPP Interconnect Function (N3IWF) and Network Data Analysis Function (NWDAF).

These functions are critical to support the requirements of new 5G services, both in the context of the data and control planes. Thus, these functions will be used in the SAMURAI project to integrate the different access networks through the concept of network slicing, as can be seen in Figure 2.


Figure 2. 5G Core and SAMURAI Platform

In addition, one of the main objectives of the project is the implementation of the 5G core in RNP's software-defined infrastructure, called IDS@RNP. This infrastructure consists of a central cloud and a physically distributed edge cloud, both interconnected by an SDN network overlaid on the backbone, as well as by a tunneling service native to the equipment of this backbone. At the end of the implementation, it will be possible to interconnect the four access networks arranged in the RNP client institutions involved in this project (UFPA, UNISINOS, UFG and Inatel), as illustrated in Figure 3.

Figure 3. Integration of the SAMURAI project with IDS@RNP

Publications

News

Post doctoral fellowship positions

There are several postdoc openings at Brazilian universities within the scope of the SAMURAI project.


Extra information about the postdoc positions can be found at https://nextcloud.lasseufpa.org/s/g5s3sGZBp74XpAS


Members

Acknowledgement

Samurai Project has received funding from the Brazilian Ministry of Science, Technology and Innovation (MCTI) through FAPESP.