Enabling the Enterprise — Private 5G

This will be an ongoing blog/primer — — I will add a section over time.

My intention is to help the enterprise in the process of moving to a “Private Cellular network on 5G”, therefore this primer is a humble attempt to explain the options and general considerations in making that decision. I will try to keep it as simple as possible.

I will not get into the debate here on why you need a “private cellular network”, but I will definitely discuss what can it really offer, and you are free to compare it with any other access networks out there — WIFI, Fiber, etc.

This primer will look at 5G Standalone (5G-SA) architecture from the context of Private network. It will cover all the architectural components at a high level, the need for edge, the requirements on devices, security, and the flexibility with slicing, and quality of service.

One other important point — this will be carrier centric deployment, though it is possible to bypass the carrier — read on.

Scenario

Just to set the stage, let’s discuss a scenario. We have an opportunity to deploy a 5G Private cellular network in a manufacturing facility. The facility is about, 1 million sq ft in size, has about 1000 people on the floor, and some 2000 machines that need to be connected, out of which 20 of them are high-efficiency critical machine tools.

Given this need, the private network needs to:

1. Cover a large coverage area,
2. Support personnel devices, and IT traffic
3. Support critical machine traffic. We can assume these machines are already equipped with connectivity modules for cellular wireless (5G).

We have to design a Cellular Private network setup that will accommodate.

1. Segmentation of traffic — a separate network segment for Personnel traffic and machine traffic
2. Security between devices and applications
3. Guaranteed reliability — adherence to a Quality of service for all the communication
4. The network needs to support that by giving ubiquitous coverage, high reliability, and security established between the device and application.

But the most important of all, we need to identify the actual use cases which will run on this network.

5GC Architecture for Private Network

Let us shift our discussion to understand the architectural components of this private 5G network. A simple representation of this setup is below.

The private network is the backbone of connectivity towards the applications. Applications can be IT applications used by the personnel or specialized applications used by the personnel and by the machines themselves.

For now, I have constructed the 5GC architecture as follows :

The architecture is divided into two parts — There are components of this architecture that may reside in the Carrier data center, while other components will reside in the manufacturing facility, which we will call “On-Prem” deployment.

5GC architecture shown above has many co-operating components called “virtual network functions” (vnfs). 5GC architecture as defined by 3GPP employs a Service-Based Architecture (SBA). Note: I have kept the interfaces between the components to be minimum for better understanding.

SBA is an architectural pattern where each service is “contained” therefore modular software component. The modularity allows it to be flexible, and its life cycle management is easy. Microservices, containerization, API's play a vital role in this setup. The best part is that these components may be multi-vendor but in compliance with the standards set by 3GPP (in 5G architecture context), therefore gives the carrier to source and integrate best of breed components.

A quick description of each component is below :

AMF — Access and Mobility management function — Manages Radio access and mobility management of the users. AMF determines what radio resources to be assigned to the device and manage all signaling for mobility. Also does Access authentication, aids in network slicing, and mobility management.

SMF — Session management function — Manages the data sessions for all the user connections. It’s responsible for IP address allocation to the UE, Session management of specific data flow, the Quality of service associated with that data flow, and session establishment and tear-down, selection of user plane(basically user data flows), policy enforcement on that data flow.

AUSF — Authentication, and Security management function — Manages all the security keys for access, subscriber authentication during registration, or re-registration.

UDM — User Data management function — Data user profile of all registered users.

PCF — Policy Control function — Very important function. This determines all the policies that are assigned to the device and the services. This establishes the quality of service parameters — (5QI values) per session type established and to be maintained by the SMF.

NSSF — Network Slice Selection function — — Establishes a dedicated slice — a data network exclusive to specific traffic in the setup. PCF, NSSF, and SMF will be covered in more detail in this blog.

NRF — Network Repository function — As the name suggests, it is a repository/list of all Network resource functions which are co-operating in this service.

NEF — Network Exposure function — Support function which collates network events and exposes those events to subscribing functions within the network. These events may include all network and device information, session information, and all the data which is deemed exposable for other services to consume. A lot can be done with this information by the application layer.

UPF — User Plane function — — The important function, which basically handles all user data from the device on to the applications residing within the network or outside the network (Cloud). Packet handling, routing, and inter and Intra -Radio Access technologies (RAT) handovers (NR to LTE )

RAN — Radio Access Network — As the name suggests, radio infrastructure consisting of Baseband, radio units deployed across the location to handle users.

The Use case discussion

Apologies for the diversion here, but this section is important. The architecture really depends upon the nature of the use case.

1.What kinds of actual use cases will this network use? — Ex: Personnel accessing Email applications (noncritical use case), OR a soldering robot being remotely controlled by an application (Critical use case).

Critical use cases are those use cases that can have a major operational consequence, meaning slight degradation of service can have large impact on the quality, and performance of a manufacturing task. It relies upon the capabilities of the supporting network like latency, reliability, availability.

Latency management with 5G — Read about this concept — https://harpsolo.medium.com/latency-how-does-it-improve-with-5g-c699d7beffb8

2. What mobility characteristic will the use cases need? — is it an indoor-only use case or is there roaming between indoor and outdoor (macro cellular network) roaming or a combination of both.

3. Is the use case using a licensed or Unlicensed spectrum? — This is very important.

Licensed versus Unlicensed Spectrum

As an enterprise, we have two choices. We can either use a licensed spectrum or an unlicensed spectrum. With licensed spectrum, we can operate a private network under the licensee of the spectrum owner which is typically a Carrier. Unlicensed on the other hand is offered by FCC on a shared usage basis.

Just to make it easy,

Licensed spectrum

This ranges from 4Mhz — to — millimeter-wave range depending upon carriers — There is a great deal of flexibility from bandwidth, latency, and backed by Carrier’s operating SLA for reliability and availability.

Unlicensed spectrum

This is shared between various parties operating on that spectrum band and some federal services. Federal services have the priority in midst of the operation, therefore there are chances of disruption. CBRS -Citizens broadband radio service spectrum in the US falls in this category. This is very important consideration when the use case is a critical use case, and demands are high on reliability and availability from a connectivity perspective, and any interruption may not be acceptable. The cost of operating this spectrum is low. Enterprise may still need to pay fees for reservation of this spectrum.

So, use case performance will rely heavily on spectrum choices. Indoor mobility and outdoor mobility or a combination of both will be better served with licensed spectrum option, similarly critical low latency transfer, high bandwidth support will be better handled with licensed spectrum (depending upon the spectrum band). Regular IT traffic (non-critical) may be best suited for unlicensed.

Dimensioning the 5GC Architecture for On-Prem Deployment

The type and scale of the use cases will drive the dimensioning and the appropriate deployment model. Remember, we have the following main requirements to handle,

1. Coverage — 1 million sq ft of floor space
2. 3200 device ( personnel and machines), and we will add a use case which needs,
3. Indoor mobility and outdoor mobility

The questions that came to my mind while thinking through the design are documented in the same schematic below.

Feel free to glance through those and we will cover them one by one in my next post.

To be continued …….