The role of NFV and SDN in the evolution towards 5G

Usually, the concepts of Software-Defined Network (SDN) and virtual Network Function (vNF) are being misunderstood and mistaken for each other. However, the definitions for both terms are acutely different. Software-Defined Networking is “The physical separation of the network control plane from the forwarding plane, and where a control plane controls several devices”

(OPENNETWORKFOUNDATION , 2018), whereas Network functions virtualization (NFV) is “the concept of replacing dedicated network appliances - such as routers and firewalls - with software running on commercial off-the-shelf (COTS) servers” (ADVA OPTICAL NETWORKING, 2018). The both paradigms have a similar common goal, which is to transform the way communication service providers (CSPs) architect networks and deliver network services. Network operations are transformed as network function software, that is dynamically instantiated in various locations in the network as needed, without requiring the installation of new equipment.

As stated previously, the Long-Term Evolution (LTE) is constituted of two entities: UMTS-RAN and EPC, namely, the 3G Universal Mobile Telecommunication Service - Radio Access Network and the Evolved Packet Core. The EPC contains the HSS (Home Subscriber Server), which is in fact a database for the users; the PGW (Packet Data Gateway) that routes the data communication via IP network and interfaces SGW (Service Gateway) with the Internet. The SGW is responsible for routing the traffic of calls and other services from the eNB to the PGW, which is dictated by the MME (Mobile Management Entity). The dedicated hardware for the LTE network serves the purpose of the functionality it is dedicated for. Adding additional hardware in the core network adds to the complexity of the system, as well as the cost. When it comes to accommodation of larger number of users, the peak times are handled by duplicating the entities.

Such process will lead to incremental complexity of the core network and linear increase of costs for providing the adequate hardware equipment. As the number of users is predicted to increase with the emergence of the next-generation networks, the hardware-based 4G LTE equipment is no longer apposite to correspond to the requirements, because the expenditures will remain constant while the revenues per subscriber will gradually decrease. This introduces the necessity for the Software-Defined Networking (SDN) paradigm and Network Function Virtualization (NFV). The proposed methodologies offer various scenario improvements, while enabling combinations that are usually not feasible with the existing hardware only. In fact, moving the EPC to the cloud helps achieve cost reduction for the benefit of enlarging revenues margin (TAWBEH, Ali et al., 2017).

A crucial requirement for telecom network infrastructure is the compatibility with cloud or computing architecture as a flexible and cost-effective service platform. With the fact that the NFV and SDN are enabling the hardware equipment emulation into the cloud, technical issues from a networking standpoint emerge. The desirable outcome for introducing peculiar 5G network slices, should encompass automated and scalable management of cloud-based NFV infrastructure; as well as the possibility for improvement of the particular performance of the current infrastructure, in terms of latency, throughput and reasonable applicability of the model parallelly to the given scenario.

2.5.1. Mobility meets virtualization

In a 5G world, the capacity and latency are the most critical units that need to be taken into consideration.

One of the appliances that the technology will be based on, is the Distributed Cloud. A Distributed Cloud represents arranging a data center at the edge domain, like central office or a base station. All services that are included in this domain are taking a virtual form, which will enable ease of access, reduction of latency, exponential reduction of hardware cost for implementation etc. Taking all these instruments into consideration will contemplate a logical need for orchestration machinery as well as analytics and monitoring solutions that are of frivolous manner. Evidently, this machinery shall be comprised of location and even personality-based AI objects, that will automatically evaluate, monitor, troubleshoot, organize and manage the infrastructure. This network intelligence is a complex assortment of Software Defined Networking (SDN), Network Virtualization Function (NFV), Artificial Intelligence and machine learning combined with immutable infrastructure (UDDENFELDT, Jan, 2017).

To establish these modules, a precise automation method is required. Immutable infrastructure provides stability, efficiency, and fidelity to applications through automation and the use of successful patterns from

of fidelity in humans is the constant destruction and replacement of subcomponents. It triggers the immune system, which destroys cells to maintain health and it motivates the growth system, which allows different subsystems to mature over time through destruction and replacement. The individual human being maintains a sense of self and intention, while the underlying components are constantly replaced. Systems managed using II patterns are analogous (BERNSTEIN, Ben, 2015). The reimbursements of immutable infrastructure are manifold if applied appropriately to an application and have effusively automated deployment and recovery methods for any infrastructure:

• Simplifying operations. With fully-automated deployment methods, it is possible to replace old components with new versions to guarantee that systems are never far in time from their initial “known-good” state. Maintaining a fleet of instances becomes considerably simpler with II since there is no need to track the changes that occur with mutable maintenance methods.

• Continuous deployments, fewer failures. With II, it is known what is running and how it behaves, deploying updates can become mundane and continuous, with fewer failures transpiring in production. All change is tracked by the source control and Continuous Integration/Continuous Deployment processes.

• Reduces errors and threats. Services are built atop a complex stack of hardware and software, and events usually take wrong occurrence over time. By automating replacement instead of maintaining instances, instances are regularly and repeatedly regenerated. This reduces configuration drift, vulnerability surface, and level of effort to keep Service Level Agreements. Many of the maintenance fire drills in mutable systems are taken care of naturally.

• Complete cloud rebooting. With Immutable Infrastructure, the running components are familiar, and with fully automated recovery methods for the services in place, cloud reboots of the underlying instances should be handled gracefully and with minimal, if any, application interruption.

The concept of immutable infrastructure is an emerging IT strategy enabled using Docker and containers (BRYZEK, Michael, 2014). Docker can empower the 5G networking components to behave just like the organs in the human body, where a single malfunctioning organ can be replaced with adjacent one from a donor that has a corresponding genetic sequence at the 6th chromosome (the Human Leukocyte Antigen).

The advanced idea that is researched is beyond the replacement of the organs, where an intelligent system will decide to automatically perform the replacement of the modules, adjusting network performance, create a temporary solution for a peculiar problem. For example, one of the main modules deducted for the successful 5G operation is the Air Interface that is the SDN / NFV module of the 5G network, to facilitate successful service delivery to the end users (INFLUXDATA, 2017). The Air Interface can be easily incorporated into containers, and furthermore in form of a microservice architecture (that opens even additional potentials). From this point, the prospects are interminable. At a soccer match, tens of thousands of viewers can record the event with presumably HD or even 4K imaging devices. Due to the popularity of the social networks, many of the experiences tend to be shared with the acquaintances. The only possible way is using the mobile infrastructure at that point. A current LTE network supports [real] 10s of Mbit/s traffic speed, which will allow the user to send the video content on Facebook or upload on YouTube.

Simultaneous uploads from most of the viewers will bottleneck the neighboring base stations with the GBs of content intended for sharing. In a 5G scenario, using the immutable infrastructure, the Air Interface containers can be delivered to a Distributed Cloud in the vicinity of the soccer match. This action can be performed automatically using genetic algorithms for prediction and identification of bigger demands for the network, where the NFV module will receive instructions to integrate itself within the Distributed Cloud and replicate to enable load-balancing and high availability for other services that require resources at the same moment (interconnected train sensors, self-driving car sensors, IoT devices, wearable gadgets etc.).

This way, the end-to-end service delivery would be uninterrupted, perfected, while lowest possible latency

is ensured, altogether approximately less than 1ms. This can scale up to a situation where surgeons can perform remote surgeries over long distances using automated hardware and robotics, that will get as much resources as required due to priority, all in the same area where the soccer match is taking place.

2.5.2. SDN and NFV solutions, network overlay and underlay

To successfully connect remote workloads, a networking solution that can manage L3 and L2 operations is required, namely routing and switching. The actual devices that can perform these operations are hardware routers and switches. For the particular requirements of the mobile network infrastructure, the hardware usage feasibility is limited for a simple reason, which is virtualization of the networking function.

Specifically, a physical router cannot be virtualized without usage of a software-defined network. For the deployment of the mobile network and connecting the eNB base station to the virtualized Evolved Packet Core, the most appropriated SDN options are selected and consequently described. Each of these technologies have their own distinct features that can contribute to establishing a secure, trustable and fault-tolerant next-generation virtualized mobile infrastructure.