Adopting 5G into Smart Factory: How to Design and Test the Network

Today 90 percent of the manufacturing plants don’t have network to implement industry 4.0. The best option for them is using 5G private network to connect the entire factory.

Curated by Vinay Prabhakar Minj 

The limitations of current connectivity technologies pose challenges for adoption in Smart factories.

• Wi-Fi networks may not be able to meet the latency and mobility requirements of the Smart Factory use cases. As an example, Wi-Fi struggles when robots need to move quickly or stream data in real-time. Wi-Fi does not support on-demand scalability.
• Wireline LAN is too expensive to maintain, inflexible and difficult to ensure the guaranteed quality of service required.
• LTE cannot meet the latency and throughput requirements for critical use cases.

These are some of the current challenges.

When designing next-gen networks, the things that probably would create opportunities are based on the above-defined challenges.

With today’s market, a lot of challenges are around broadband speeds, especially video and remote monitoring. Some start-ups are working on vision-based systems which can track the manufacturing process by just observing without any feeds. It can actually generate a process flow based on what it sees in the recorded video. This requires very high broadband speeds.

Another task is to store data locally and process it to generate edge analytics. Ultra-reliable low-latency communication (uRLLC) is also required for AR/VR and high reliability robotics control.

Design considerations for 5G
Network “slices”: This is going to enable the kind of throughput and latency that will be required and ensures a guaranteed connectivity pipe with the required latencies and throughput. If you are doing CCTV offload (i.e. heavy bandwidth required to feed data to the backend), then you need fat slice. A thin slice can be used for a smart meter data collection use case, as it is not going to generate too much data on a continuous basis. This flexibility is being brought in as part of what we called network slicing.

Scalable & elastic network for on-demand services: Imagine enjoying a live event, such as IPL match, from different locations with the same experience as sitting in the stadium. This will be possible with the scalable & elastic network that will happen with 5G.

Global roaming: It is also possible with 4G. But with e-sim coming in, it is going to be enhanced in terms of power. You won’t be required to bring products which are pre-built with sims. Instead you can have an electronic sim that can be activated depending on the country where you are selling the products. Whether it is for a robot or a car, you will be able to have the connection that can be established based on the place where it is going to be operationalised rather than based on the place where it was originally manufactured.

5G Private Network: This is going to be cheaper than wireline and faster than 4G. Through this, all the warehouses, plants and dealerships will all get connected with high-speed bandwidth.

Battery life: It is the most relevant thing for IoT because you have an opportunity to keep a “network sleep cycle”. The biggest challenge that people are ignoring today is the longevity of the sensors. How to replace the battery for billions of sensors? 5G will offer the option to extend the battery life by putting sensors on sleep mode whenever data sharing is not required. This is the significant benefit that you will start seeing with 5G network.

Massive IoT: Which will support high number of devices for high-density deployments.

Mobile Edge Computing: Relates to content caching and localisation at edge. This will lead to better user experiences with respect to AR/VR videos and gaming.

Network Exposure for Applications: This is another new feature that is required for 5G. With this feature, IoT applications like patient monitoring or fleet management can query and influence traffic latency/through-put.

Standards-based IoT devices: This refers to “Off the shelf” devices conforming to 3GPP standards for IoT applications across the spectrum.

These are the design considerations for 5G and many of these are specific to 5G while some will be an enhancement to what we already have.

5G design key strategies 

• Partnerships with technology powerhouses.
• Trial labs to verify and optimise product and network design.
• Verticalized Industry Consortiums to optimize implementation of respective vertical. 5G would get more verticalized industry applications to be more relevant, such as hospital applications for doing robotic live surgeries or for Industry 4.0.
• Open source adoption without risking security. While 5G would enable a lot more open source adoption, it is important to make sure that you don’t compromise on security.
• Network analytics with ML utilities for self-optimisation and self-healing (real-life decision making). The inherent features that is going to come with 5G would be around network analytics and machine learning. So, the combination of edge compute, 5G and machine learning would enable factories to make real-time decisions. Today 99 percent of factories are not enabled for real-time decision making. It won’t be just about collecting data, you will be processing it and making decisions based on it. It would help in self-healing and also provide opportunities for self-optimisation.

Test considerations for 5G

• Device certification – One needs to look at how to get devices certified by ensuring compliance with standards and carrier requirements.
• Network validation – This is important for interoperability and roaming. One should ensure a secure and reliable data pipe for interoperability roaming.
• 5G IoT application test – Use cases should be tested for verticals such as healthcare, UAVs, asset tracking and future factory infrastructures. It is also essential to verify end-to-end latency, burst, survival time, jitter, etc. These are done through baseline test pack for each vertical. So, you need to create test pack for each vertical which can be tested against the use case requirement.
• Performance and capacity testing to validate different IoT traffic models – It includes deterministic periodic communication, deterministic aperiodic communication and non-deterministic communication. All three conditions need to be tested when validating for performance.

Adopting 5G into smart factory
Below is a roadmap of how 5G factory environment will evolve in the coming years.

5G network innovation lab: In the beginning you may have only a sandbox environment. Here you can test various conditions, set it up as innovation and start bringing private LTE and then begin looking at use cases.

Private LTE based on LSA band – Then in 9 -12 months, challenges such as lower OPEX, high security and no CSP dependency can be addressed. Use cases will involve digital logistics and RFID, digital plant and AR/VR.

Edge cloud and network slices – Then in 12 – 18 months, as the maturity of 5G will also evolve, then customer experience will get enhanced and optimal network slices will get created. This will address the challenges of lower latency, local caching and QoS. Use cases will be enhanced to include field connectivity 4.0, virtual reality on safety training and predictive maintenance, personalised nutrition for employees, etc. A lot of other edge-based applications can also be brought here.

Standards-based 5G network – Then in 18 – 24 months, challenges of full 5G spectrum availability will be addressed. Use cases will include mixed reality enabled maintenance management, collision avoidance system for AGVs, connected sensors for utility management, dispenser maintenance optimisation, telemedicine and digital plants.

Disrupting with 5G – connected AGVs (Automated Guided Vehicles)
This is about connecting your factories, warehouses and making them fully interoperable. Here, you will find a private cloud, a campus edge and connect vehicle-to-vehicle as well as vehicle-to-infrastructure. The vehicles here could be AGVs which will transport various goods from one place to another or transport goods from plants to warehouses.

If you want to apply industry 4.o today, the cost is going to be definitely high to start with. Moreover, today 90 percent of the manufacturing plants don’t have network to implement industry 4.0. They won’t be able to really connect the whole factory completely. For them the best choice is that instead of going with the wireline lane to connect the entire factory, they should opt for private networks.

About the author
The above article is an extract from a speech presented by Karthikeyan Natarajan, Global Head, Engineering, IoT & Industry 4.0, Tech Mahindra, at IOTSHOW.IN 2019. He has more than 27 years of professional experience in product development, globalization and digital transformation strategies.