5G The Use Cases to Start With

When it comes to 5G, the most important thing that pops up into the mind is "The Use-Cases", it has to offer. Besides all the hype, the industrialists (especially the Telecom Vendors) & technology critics are putting into this space, 5G will be real by next year(2019). However, it is important to classify the Practical use-cases from the Potential use-cases, 5G is going to offer in its initial couple of years of launch.

The first use-case any CSP, will be interested to implement will be eMBB and FWA. Below are some factors, which make this to be a viable choice over others:

1. Market Drivers:
There is a lot of talk about 5G not being the only about faster download. However, higher speed will still be the prime factor for this upgrade. CSPs are still having the majority of revenue from the retail users. and as a matter of fact the term "Next-G" directly translates to higher speed to most of those users. They really don't care about the potential and technology boundaries of any generation, but only the outcome and experience, they are getting out of it, which in this case is 'Faster Downloads'.
There are evidences from the market research about the consumer's need and Return on Investment, which shows that more than 75% of the revenue of 5G depends on eMBB & FWA. Which make them the first choice over the other 2 use cases namely, URLLC and mMTC.


2. Race to Win:
CSP leaders in North America and Europe always want to be ahead of the  world to be named as 'The First' in their work-space. As a matter of fact, Verizon and ATT are planning to launch the FWA in more than a dozen cities, starting next year. Apart from this, the Telecom vendors are also focusing in this direction. Samsung, Ericsson and Nokia are also offering their upgraded EPC, suitable for NSA-LTE-Assisted 5G to meet the North American CSPs needs to start with 5G.


3. Time to Market:
If 5G to offer a latency of sub-milliseconds, then there is a lot of work on Vendor's plates to be done. A typical LTE-EPC can offer anywhere between 50ms to 500ms of latency, but in most practical cases, it usually above 100ms, which is not even the starting point of mMTC & URLLC requirements of 5G. In order to achieve  this, a complete restructuring of the network and redesigning of individual blocks is required. In other words, the current EPC is simply not possible to be upgraded into 5G-Core.
Hence, it becomes prominent to incorporate what can be delivered faster. So, the only use-case left with is eMBB and FWA, which really don't require any fundamental upgrade in EPC, apart from opening up bigger optical data pipes to offer more throughput.


4. Convenience of Non-Mobility:
Restructuring or redesiging the Radio network is even more complex than EPC, as it requires complex algorithms for best radio resources and converging multiple RATs to provide reliable connections, while user is on the move. Further more, Radio network in 5G also requires the dynamic slicing assisted by SDN and NFV. Considering this complexity, the Mobility in 5G will still be limited during the initial couple of years. Hence it makes the FWA as the most eligible use-case where only demand is a higher data rate. Verizon and ATT are also working on this direction, where first thing they are going to offer is FWA with higher throughput (~300Mbps to 1Gbps)


5. EPC Transition:
EPC has been in play for more than a decade, even main packet core nodes like SGSG, GGSN, SGW, PGW has been there in the network since the inception of mobile-internet (SGSN, GGSN). A crude way to explain it is that SGSN and GGSN were simply upgraded to support higher throughput and advanced protocols like GTP-V2, Diameter etc, but their functions have not been changed fundamentally. Apart from the higher throughput, there has never been talks about reliability and low latency. Still LTE has significantly improved the latency upto hundreds of milliseconds, but it still not usable for URLLC and mMTC. 5G-Core on the other hand, requires a complete redesigning of fundamental blocks on Packet core, where not only the throughput, but also the latency and reliability is a prime design factor. The new protocol stacks like HTTP/2 and REST will take a lot of time to be available in production grade. Hence the transition of EPC to 5G-Core will take more than a couple of years. This makes it almost impossible to address any use case other than faster bandwidth.


6. Adoption of 5G by Vertical Industry:
Any use-case is not valuable, until it has the consumption! 5G is targeting the vertical segments under mMTC and URLLC umbrella, but are they (Verticals) ready to use this? Vertical industries are using wired infrastructure for decades. It took years to them to reach the automation scale, they have right now. There is not only the cost which is involved for them for this transition, but also an insecurity of unproven technology. This will take years  to get them half of the confidence, which they currently have on they current automation space work.


7. Every Vertical is Unique:
Vertical industries are having completely different requirements than others. For instance, Connected cars, don't require a very high bandwidth, but they require >1Mn connections/Squire-Km and a wider coverage along with a robust contingency mechanism to handle sudden rise in traffic. On the other hand, Industry automation, requires a very reliable (99.99999%) and sub-milliseconds latency (<1ms). This simply makes it clear that one network is not enough to cater everything in mMTC and URLLC space. The use of multiple and converged Access and Core-network will be only way to achieve this. But this all will take years to get matured.


8. Network Slicing:
Slicing of Core network is not new in mobile telecom space. However it was implemented with a very crude way till 4G. The main slicing criteria were include APN and Location, which has been there for more than 2 decades now. But granular level of network slicing will become necessary in 5G more than ever before. Network Slicing in 5G is not only selecting a PGW and SGW pair based on an APN, but it covers end to end network functions, including Access-Network, User-Plance-Functions, Control-Plane-Functions and even Application-Functions. We are looking at a complex network topology, where any part or set of parts of Network-Functions can be offered as a Network-Slice. There network slices are designed/configured to serve a specific set of use-cases.
However to manage this complex and forever changing network topologies, the conventional OAM mechanism is simply not capable enough. The orchestration of NFs and Network-Slices, assisted by completely automated framework, primarily assisted by NFV and SDN, will need a lot of time to be available in Telecom space. Even to this date, NFV in Telecom is less than 10%, which is commercially implemented. This will take more than just a couple of years to offer the Network-Slicing services, which is a primary tool in 5G to cater multiple services and multiple networks.


Conclusion:
Considering above reasons, it is very clear that all Vendors and CSPs will continue to focus on eMBB and FWA for next couple of years. But there is also some use cases in URLLC, which can be taken into account as they move forward. For instance remote gaming, AR and VR demands 10-15ms of latency, but not being a mission critical applications, they can be the perfect use-case candidates  after FWA. Once 5G is proven to completely take over the Remote Gaming and AR/VR driven infotainment, then CSPs can convince the other virtical.

Still moving ahead to 5G with eMBB and FWA only is still a bold and necessary move to start adopting the transformations, 5G has to offer.

5G Myths Facts Use Cases Solutions Realization

What is 5G? The first thing come into mind:

1. All about faster downloads

2.  Killer of Fixed Line Networks

3. The one North America is using since 2017

However, the reality is:

5G: Myths Vs Reality

 

5G is  not just another 'G' in telecom evolution and there are facts to support this, which includes unprecedented growth in high speed data demand & connected devices, completely different and unique use cases, Ubiquitous network and inception of new network functions.

The demand of high speed data is not new for Telecom consumers. However, 5G is not all about a higher speed; it has to be something more than just a bigger data-pipe. That is the only thing that can survive this transition and sustain the next generation mobile technology on the course of future benchmarks.

Below is a picture, which depicts the journey of Telecom industry, since its inception till date.

Telecom Convergence Journey with Horizontal Industries

The noticeable points here are:

1. Mobile Internet opened up a new doors of Multimedia offerings for Telcos
2. High quality content demands draw attention of Computer and Software industry to enter into Telecom Domain
3. By the time LTE became matured, the new ICT industry has taken a shape converging all 3 main Horizontal Industries (Telecom, Computer, Entertainment & Multimedia) into a single ICT industry
4. Today, as 4G has reached its limit and Telecom has converged all the possible Horizontal industries, there is nothing left for a new generation of Telecom to offer.

In other words, there is not really a need of a new generation of Mobile Telecom or 5G for that matter, as there is no new use case left to cater in any of the horizontal industry, because if its only to offer more speed, they can widen their fiber optic infra and densify their coverage in 4G itself.

This concern of sustainability is actually the prime thing, which 3GPP, ETSI and all Telecom Vendors CSPs were having in their minds, when they were developing the 5G specifications. That's what makes them open up beyond thee Horizontal industries, 'THE VERTICAL INDUSTRIES':

Entering the Verticals

But there is are problems with each of these vertical industries use cases:
1. Each industry has their unique use cases
2. Each use case demand is different than others
3. One network is not enough to cater all these use cases

For instance:
1. Connected Cars, they don't require a very high bandwidth, as they are only transmitting 10s of bits per second; instead they require a wider coverage, >100Mn Connections per Squire Km, a different identity system, battery life of a decade or more.
2. Industrial Automation is fine with a limited coverage and not very high number of connected devices, but they require extreme reliability and very low latency (sub-milliseconds)
3. Immersive gaming (AR/VR) requires very high internet speed, not only in downlink, but also in uplink with a latency >10ms.

Looking at the above use cases, its clear that each use case demands a unique characteristics of network service(s) & 5G is all about fulfilling those. Hence, 3GPP has categorized, categorized all 5G use cases under 3 category:

3 Umbrellas of Vertical Industry Use Cases

1. eMBB (enhanced Mobile Broadband): Gigabit Mobile Internet
2. mMTC (massive Machine Type Communication): Millions of connection per Sq. Km.
3. URLLC (Ultra Reliable Low Latency Communication): Remote Healthcare, Industrial Automation

Below is a double click view on the above use cases and their unique network demands, along with the possible solution(s) for each of such use cases, all driving along under the umbrella of above 3 main categories:

Granular Use Cases

Just to make this reading light, I will be taking 3 different use cases from each of 3 umbrella with its possible solution in 5G:

1. Gigabit Mobile Internet

To provide 100s of MbPS or more data throughput, it all brings us down to the bottleneck of the Mobile network, which is the Radio Network. There are 2 solutions proposed in 5G for such use cases of very high data throughput demand:

(i) Use of mm(millimeter) waves: The good thing about mm waves are that they can offer much higher spectrum (upto 300mhz of channel bandwidth), which makes it possible to pack more data into the carrier

(ii) Use of Beam-forming & Beam-Tracking: Since mm waves requires a very tiny spot to focus on, it makes easier to point the carrier waves to a particular target UE, rather than radiating it all the way wide open.

(iii) Use of Massive MIMO: By using as high as 64X64 MIMO makes it possible to increase the spectral efficiency and data rates by a factor of 10 than using 8X8 MIMO mostly in LTE.

Use of mm waves makes Massive MIMO possible, since the antenna size is very small in case of mm waves, which makes it easier to fabricate 10s of such antennas easily on a usual UE.

Massive MIMO and Beam Forming

However, there are certain drawbacks of using mm waves; by having such a small wavelength, mm waves are very prone to the atmosphere conditions and can be blocked easily by a mere presence of fog or mild shower. Which certainly makes it impossible to use mm waves for wider outdoors.

2. Connected Cars

Connected cars requires a wider outdoor coverage, Millions of Connections per sq. Km and a longer battery life. To provide solution for such characteristics, the use of multiple access networks (especially mid range waves, WiFi and possibly Satellite Communication in certain areas), along with less complex data processing algorithms & use of On-demand network scaling (to cater to sudden increase in no. of connections) will be required. Below are 3 solutions to cater such requirements:

Wide Coverage for mMTC

Non-IP Data Delivery

(i) GERAN, UTRAN, WiFi, Satellite Access Network to provide wider coverage

(ii) NIDD (Non-IP Data Delivery): Use of Ethernet PDU Sessions/ Flow, reducing the complexity, hence  increasing the battery life.

(iii) NFV for auto-scaling and auto-healing of network functions for providing on-demand network capacity

3. Remote Healthcare

To support such a critical use case, it is obvious to have an extremely reliable connection and a sub-millisecond of latency. Distance between the application-server and the Access-Network may not be a conceptual reason of increased latency, but it really a contributing factor in latency. Hence, it make sense to keep the application-server as close to the RAN as possible, this is what makes the use of MEC (Mobile Edge Computing). However, Edge Compute doesn't always talk about to put everything on the edge, instead it is to be decided by considering 2 factors; Latency Demand of Application, Cost of bringing the NFs to edge. It is well known that bringing everything to the edge is always leads to higher cost, as edge-compute can only serve a specific set of consumers and is usually non-reusable for other use cases.

Below picture outlines the trade-offs and options of Edge-Compute for latency sensitive Use-cases:

Hence to summarize it all, we had 3 areas of use cases, with different approaches for each off them and it is definitely not possible by a single network. So, 5G is a network of multiple network options.

Mobile Edge Computing Options

Then the question arise on how to manage such a dynamic demand of use cases and their possible solution in run time. That is where the concept of Network Slicing comes in. I will be writing another write-up on Network Slicing for this. But here is the final take on 5G, What actually it realize into:

5G: Summary View