Rethinking Networks In Telecommunications

00:00 — Burr Sutter
... do you remember your first wireless phone? Was it a size of a shoebox, or perhaps something small and sleek? We have seen the hardware in our hands change dramatically over time, but what about the unseen parts of network infrastructure? Well, I wanted to explore the evolution of telecommunications and how that has fundamentally impacted all of us as humans on this planet and how we connect with each other. So I went out and talked to Sandeep Sharma, Vice President of Tech Mahindra, a multi-billion dollar tech company, and I want to find out from his perspective, given his career in telecommunications, how this industry has evolved over time.

00:36 — Sandeep Sharma
When from hard line we moved to 2G, it is a great revolution. Everyone is fascinated about that we can make a call from a wireless network, but that was the only expectation from on the 2G. When we moved from 2G to 2.5G or 3G, then the expectation started to increase. Okay, there has to be data as well so that we can exchange the texts, we can exchange the pictures and all. Until 3G, it was more evolution from voice to data. When we moved from 3G to 4G, the expectation was, okay, no, we need a lot of data. And not only data, we need to have an IoT type of devices where every device needs to be connected. And in this whole evolution, the expectation from the technology from a consumer perspective keeps on increasing.

01:27 — Burr Sutter
Well what we all expect, what we all want, is that instant access to the world's knowledge base at Wikipedia, our bank accounts, and of course some real time streaming video of the world's best dance moves. We're going to talk about some user expectations and how the technology has evolved to meet them, as our conversation continues. (01:48): I'm Burr Sutter and this is Code Comments, an original podcast from Red Hat. Coming back to Sandeep, he told me more about this evolution as he witnessed it unfold, and how frequency bands factor into particular use cases.

02:04 — Sandeep Sharma
5G, it supports exorbitant speed in terms of gigabits of speed, a different amount of spectrum as well. A spectrum being a wireless spectrum which is always a scarce source. It was limited to few frequency bands until 4G. In 5G, there's N number of bands, and the bands you can think of low band, high band, mid band, and every band has its own advantages depending on specific use cases. So in this evolution, lot of big things happen. Technology evolved. Now today's technology, the 5G, is more towards the virtualization, more towards satisfying a large number of use cases, connecting industries, connected devices, and came out as a, I would say, full box that satisfy the basic connectivity to a high amount of data requirements for high SD streaming and whatever you can think of in terms of connectivity today. But it is still limited to the vision of use cases that we have today. The evolution has not stopped at 5G. It'll keep on evolving, and I'm looking forward to the next view of technology as well.

03:12 — Burr Sutter
I have also seen this evolutionary path that you're talking about here from for the last 25, 30 years or so, just as a consumer, just as a person who wanted that mobile phone and wanted that wireless freedom to roam around in my car or around the park or around the block and speak to my friends and family. (03:29): So, I think most people listening to us right now are like, "What was it like before we had wireless, and that 2G capability? And what was it like before we had data on our phones?", so we had smarter phones and I think that is so important. So, I love what you talked about there in that history. There's elements here that you mentioned that I think are super critical. You mentioned the concept of use cases, and I want to definitely double click on that a little bit more. Let's talk a little bit about the architecture of these networks. I think as I come from an application development background, traditional enterprise software development, building, web-based systems, client server systems, mainframe and what I call CUI systems, character user interfaces back in the day, what is it about the architecture that's so important here? And then, there's some laws of physics that also apply. You mentioned the concept of spectrum and having those bands, so can you talk to us more about what architecture looks like from this perspective?

04:17 — Sandeep Sharma
Sure. But I think, and you raised a good point, and even the experience that you're talking about yourself of application development and all the stuff, that's making more sense to telecom networks today as well. So then 4G, we are talking about more, I would say that in 3G, we're talking about more about telecom networks which are deployed in a traditional way. We have a tightly coupled architecture which satisfied by a set number of vendors, but from 4G onwards there's a concept of virtualization comes in. So architecture is getting more, I would say, diversified. The hardware functions which are there, they're started getting virtualized. Then, the concepts of cloud moving the workloads from the telecom world to the cloud is coming into picture. And when 5G comes in between 4G and 5G, it was more transitioning that happened from towards open RAN or the cloud RAN. And this transition towards open RAN and cloud RAN is important from multiple perspectives because since the expectation from the industry is high from the user perspective, every user wants to use the telecom network on its own way. (05:27): For example, someone in an industry which is using a manufacturing as a main business, they want to use the network so that their manufacturing efficiencies can be achieved. As a college-going student, they want to use all the educational videos run in three dimensions, everything XR, VR that you may think of can be realized in a wireless environment. But to realize all this from a single network, you have to do a lot of, I would say, innovations. You need to bring flexibility to the network, and allow the users to use the networks on the desired way. And in the current state of network in the open RAN coming in, open RAN is basically utilizing the, I would say, the openness of the architecture where you can take couple hardware software from different vendors, and compile it in a single stack to create a best of breed solution, satisfying the requirements that operators are expecting from their network. And that's where the application development is getting more important in case of cloud RAN network and open RAN network. (06:32): Reason being, now the use case or the expectation... When you say a use case, use case, the expectation, the end user's expectation or the network, it's a use case. So, that use case will be realized just by using a specific applications, and I'll talk about later when different evaluations that are happening in terms of open RAN, you can realize a network slice or the network function just with a software developed, which is customized for this specific requirement. That's how the transition have been from a traditional telco, which is talking only about voice and data, to a network which is virtualized, which is application driven, and the network can be tuned to a specific requirements with a set of applications, which that too again in an open end environment. So it gives a lot of flexibility to the users, to the operators, to deploy it in a selective way. And at the same time, users to use it in the most efficient way.

07:29 — Burr Sutter
So it's not just the network has changed: the spectrum has changed, right? How many bits we can push over the wireless spectrum, that has changed, and that's that 2G, 3G, 4G, 5G that we've been talking about. But the fact that you guys are essentially now an application platform to meet the custom use cases of certain industries is super interesting. Can you walk more through those use cases, those industries, and some examples of what people are trying to achieve out there?

07:55 — Sandeep Sharma
So, from the end user perspective, for example, different age groups have a different expectation. When you talk about gaming, gaming users will require a different quality of a network or different capacity of a network. You talk about Metaverse, which is an amalgamation of XR, VR, and AR, everything that you can talk about, it's Metaverse. Different use cases are there. Everyone wants to use the network on its own way, but how a single network will satisfy this? Then, there's a concept of slicing comes in, which use case or its use against which type of capacity is defined by the slice. So, slicing is important concepts that came with 5G, and it's very well utilized in the open RAN cloud RAN environment. So, the slicing allows the network to be sliced instantaneously to meet the end user demand. For example, if a slice is damaged to support a, I would say, Metaverse type of application and certain users demand for that, then it's the responsibility of that slice to ensure that end-to-end cloud quality of services, I would say, is ensured. (09:06): Now coming to industries, if you talk about healthcare industries, the healthcare industries, everything is very time sensitive. So, every action that is there has to have an equal reaction at an instant, that to a level of millisecond if you're monitoring a patient remotely and if you are attempting to control something remotely. And that's where the 5G low latency networks comes into picture. (09:33): The technology promises to have a latency of the order of few milliseconds to the length of, I would say, one millisecond. That is the industry's specific use case, and even few of the automation industry is there. The decision has to be faster if you want to automate a production line. Again, the time is critical. So, all these automation in the industry that you want to achieve robotics and everything, the 5G use case is there with a low latency. (09:58): Now, coming to the operator perspective: operator also want to achieve an efficiency in terms of operation efficiency, in terms of observability of the data, so that they can do analytics, use AI/ML on the network. So, once you are clarifying the infrastructure, you are making it virtualized, you are making functions interact in the software world. You have a single depository of data, which can be analyzed, which could be used as a single point repository to take critical decisions, maybe in terms of predictive analytics, maybe in terms of taking critical decisions for the future investment in the network, or maybe solving the problems on the end consumer. So, every dimension has its own use case, and that we have to slice or dice the network accordingly.

10:46 — Burr Sutter
If you go back further in time and back further in your career, could you even imagine where we are today back when we only had 2G and wireless technology that took advantage of that?

10:55 — Sandeep Sharma
I would say it's a total transformation, that point of time, 2G, no one thought of supporting cloud. No one ever thought of that, okay, sometimes telecom networks will be virtualized. At that time, the expectation from the network was not that big as well, to be honest. Getting a connectivity was, I would say, a privilege. But now, getting a high speed connectivity is still the basic necessity in today's environment. (11:21): And the efficiencies, and it doesn't mean that okay, we are not aware, we are not informed at 2G point of time, but at that point of time we have not learned anything. But from 2G to 5G, we learned what was good in the previous generation, what was not working in previous generation, and we actually evolved the network with other technology with that approach. So, it is a learned network that we have today, and this learning will keep on continuing, and since I have lived through all these transformations, it doesn't make me feel that we are transitioning, but when you discuss today, when I go into the past, okay, then I realize, okay we transformed so much. This technologies allowed the boundaries of time, boundaries of countries to be diminished as we progress.

12:06 — Burr Sutter
And when it comes to 5G, and you talked about those slices, and of course making sure that spectrum is available to a certain set of use cases, or you have virtualization, if you will, that basically lets you to work on a certain use case, one of those use cases that we're seeing or hearing about a lot these days is that self-driving car and managing a fleet of vehicles moving down, let's say, a fast moving highway. Can you talk more about what your vision is from that perspective? What could happen in our future when it comes to self-driving cars and how would this network be leveraged for that?

12:34 — Sandeep Sharma
Having a self-driving car, it's not that easy to be honest. A self-driving car requires a lot of sensors, a lot of cameras, and it generates a lot of amount of data. And that data has to be used and consumed at a central source, so that it can take a decision faster. Maybe at the local distances as well. (12:54): That required a high amount of connectivity, I would say dense connectivity. That too and the decision making power has to be so quick that the decision of applying brakes, if some object comes out and the vehicle is in the front, all those things has to be in a fraction of a second. That's where the latency comes into picture. And it'll not only limit to here, and the geolocation perspective, and the technologies evolved so much that all the geolocation specific applications can run pretty smartly on the 5G, and 5G is still evolving, more things will come into picture. But the location precision in everything, it is quite precise in terms of 5G and other technologies that exist today. So, this is from the self-driving perspective, and device to device communication, with everything communication. We call it V2X. It's like, full set of industry that has evolved so far now, and that will keep a promise of that was made when 4G was here that almost everything that you can think of will be connected. And having a driverless car, our smart cars, V2X is just a start.

14:14 — Burr Sutter
So, tagging onto the idea of smart cars needing ultra low latency, and perhaps like many of you, you might have learned about traditional IT networking eons ago. Well, with that in mind, I asked Sandeep more about his early days in networking and about latency, redundancy, and even error correction.

14:30 — Sandeep Sharma
For sure. The error correction mechanism has improved. Not only error correction mechanism, in terms of how the ACKs and NAKs are being handled in 5G, it is evolved to a significant level. And on top of it, the modulation schemes that we talk about. For example if you go back to 2G, it was only then soft QPSK or BPSK at that point of time. Then, we worked to a certain level of QAM, maybe 16 QAM at the max. (14:57): And now when we talk about 5G, you can think about taking a 1024-QAM. That's like driving exorbitant data on the wireless platform, and not all the spectrum will be able to support it. Predominantly it'll come towards the spectrum which is close to the users, with high data adaptation comes in, and depending on standards and the normal ecosystem, the standards also have a defined set of values in terms of error correction, which type of environment needs a better correction, hard modulation schemes, all these things will derive basically the capability that can be achieved from the network. But in summary, from 2G to 5G, there's a significant improvement has happened from the standards perspective, and because of this improvement, that's a reason we talk about different use cases today.

15:49 — Burr Sutter
I like that you mentioned standards there, and there's actually one I've heard of or at least an area of technology I've heard of and I'd love to have your thoughts on, that is O-RAN, for open radio access networks. Can you tell us more about what you see happening in that space and how is it impacting your world?

16:04 — Sandeep Sharma
Yeah, so open RAN people always gets confused. If people are going to say, people is like end consumers when they think of open RAN, it's actually a transforming... It's a new technology. So, open RAN first of all, let all of us understand: for ORAN, it's a transformation in this philosophy. So, ORAN allows you to deploy the network and using all the things that are there in the [inaudible 00:16:30] technology. For example, if you want to deploy 5G and if you deploy 5G in a traditional way or 5G in an open RAN way, the technology that allows from the 3G perspective, that's a standards body for evolving technology, very much there in the open RAN, but open RAN from the ORAN Alliance standards perspective is giving us flexibility, because with the ORAN coming in or the cloudification coming in, there's a desegregation that is happening. You have decoupled the hardware and the software. (16:56): If you talk about the traditional way, the hardware software were tightly coupled: everything comes from a single vendor. So, all the innovations that can happen in the network is tightly coupled with the way that single vendor evolving their software roadmap or hardware roadmap. But with the open RAN coming in, there are a lot of the standards that are defined in terms of interfaces between a hardware and software, and there's a dependency removed. So, software can come from any vendor, hardware comes from any vendor, and mostly hardware is a [inaudible 00:17:29] hardware. There's a general purpose hardware that we keep on using in terms of servers in the IT world, and that flexibility is giving the operators the option to deploy it in a flexible way, in an agile way, and define their own roadmap for the network. (17:44): Finding out which fits best for the use cases that are thinking of network, it also gives them a decision making power in what type of investment they should do with a certain vision, and that investment is secure for a certain amount of time that they have a control, rather than giving control to a single entity. So, that type of flexibility and power is actually there with the open RAN coming in. And not only flexibility, I would say. If you think of ORAN, I would say is end by end metrics. So, it is flexibility not only in terms of how to deploy network, flexibility from the supply side as well. Now you’re with a lot of vendors to select. And since most of the things are virtualized, and you have a central depository of data, you can evolve in an innovative way which innovation fits for your own network, rather than replicating and based on some recommendations which is coming from a single entity. (18:38): And since everything is data driven, you can use AI/ML in a more efficient way. And there's this other component of evolution of ORAN, we call it radio intelligent controller, or RIC. RIC is a beautiful concept, that RIC framework is actually allowing to realize all these use cases that we just talked about by just using an application. We call it RRIC in case of non-real time RIC or [inaudible 00:19:04] in case of near real time RIC. So, these applications are open source applications and will evolve with that RIC, that any entity can create an application with a certain logic that I want, I have this expectation from this network. For example, if there's an emergency or an emergency situation, how the network should behave, how the resources should be provided to the end user. An excerpt will be created and it will be noted in that RIC, and some framework will take decisions based on that. (19:33): So, there'd be N number of use cases that can be enabled in the network by just using those excerpts and RReps. That's a beauty that's there. The lot many things are done, and I think in a year or two, we'll see the commercial deployment of happening. Since we are talking about not much, you know, I would say innovation and I would say AI/ML things, it'll improve the TCO for a longer term. It'll achieve the automations, it'll bring the open interfaces, and the cloudification. All those efficiencies will eventually lower the TCO in terms of CapEx, Opex, and the like.

20:09 — Burr Sutter
I certainly see how that openness that you described around open RAN has fundamentally changed the game. You already talked about that throughout our session today. The fact that the ecosystem has opened up with virtualization and the choice of multiple hardware providers, and of course unleashing all this agility and innovation in the space. I see how that has helped the telecom industry tremendously. Would you say that has been such a fundamental game changer and impacted your career path, your personal interaction in the industry as well as the consumers as a whole?

20:39 — Sandeep Sharma
Yeah, that's a fundamental shift, to be honest. And the bigger thing that I have seen while I transition from traditional technologies to the open-end technologies, that is lot of learning is there. And this learning is actually to be a fundamental for not only for an engineer or a person who is who's in the industry. It's actually a learning for a larger audience as well, so that they can understand the beauty that this openness and innovation is bringing in. (21:10): And bigger challenge that I see from my personal experience perspective, because all these nouns, terms, for example AI/ML, cloudification, virtualization, these terms were not there traditionally in any of the then-networks. So when people started talking about, hey, does an organization share, that also comes in, okay? These are new terms. We have never thought of it. We have a long experience of...in this industry, so these things might not be relevant. (21:38): And some fears, quite, you know, these are quite complex terms, whether we should be able to realize the benefit out of it or not. That learning curve has to be there, is there, and I learned a lot while evolving towards open RAN part of my career. And I have a strong belief that open RAN is there, it's very much here, and it's going to be the fundamental of all the technologies that are evolving. Even the talks are there: when we talk about 6G, it's going to be there by 2027 or 2030 timeframe. The standards started talking itself [inaudible 00:22:14] itself, and there's a separate group called [inaudible 00:22:18] Alliance, we call it [inaudible 00:22:20], that talks about how the open RAN can be transitioned towards a 6G framework. And the basic fundamental is the same: it's the have the openness in the overall architecture so that the end users can better get benefit, the operators can get a benefit, and the innovation should always be an important pillar with the flexibility that it brings in.

22:41 — Burr Sutter
Well I'd love to ask you more about this. We could actually talk forever I think about some of these topics because I'm certainly very interested in these different use cases of how this technology rolls out, but I would love to know what your key advice is for someone looking to implement a next generation wireless network for whatever their use case may be. What would you tell them? What area should they explore? What technologies should they think about and what are the big gotchas? What would you say some of the big gotchas are in that space?

23:10 — Sandeep Sharma
If someone is planning to deploy a network, which is a latest network in terms of technology, in terms of latest ecosystem, then they need to look at the expectation, they need to list all the expectation of the network. Listing the expectations means what's the priority? Is it end user experience, or it's investment is a priority, or X number of things? Based on those business expectation of the network, because now more and more networks are business driven, there has to be business objective that will satisfy. And business objectives can be huge, it can be small, it can be large. For example, some business objectives can be as small as, "I just want a basic connectivity, I just want to connect the utilities," and business objectives can be as either, "I want to have a network which is Metaverse capable in the future." It means very huge amount of capacity and the latency sensitivity. So, to have those objectives written down, deliberated, and not only from the today's perspective, look from the perspective, what you expect as you evolve your network in the next five years. (24:16): And then formulate a strategy. Strategy in terms of framework, how you want to evolve this network, and this framework will have different stages as I discussed earlier. First, you have to discover, discover a sense of what things exist today, what can be evolved, which standards have a better visibility for the future, and which standards will bring flexibility so that you can select your technology or your investment accordingly. Then, you define the architecture. What architecture will best suited for your network? Is it a traditional way or is it a virtualized way? Most definitely, since you are looking for more use cases, no one will deploy for a single use case today. There has to be more use cases. Once you have a definition done, you just design it for different use cases, and develop a stack. It can be horizontal stack or a vertical stack of the blueprint, and this stack will not include only the hardware. (25:10): Since we are talking about more in virtualized network, it'll be a metrics, again, of end by end. We talk about which vendor hardware, which vendor software, cloud from which provider, whether that cloud will be a horizontal cloud that will take care of both my RAN and the phone network, or it has to be a different cloud that I need to deploy. And select the best which is available. (25:31): And then comes the deploy, which options help them to deploy the network in a fast efficient way, so that your time to market, your time to revenue, is faster so that you remain competitive in the market? And first deployment is done, you need to look for how efficiently the network can be bandaged. Not only efficiently but intelligently, how the data can be used to take intelligent decisions in terms of proactive correction, proactive investment strategies, and the solution of the customer complaints. (26:05): And that with the cost efficient way. And while you're evolving towards operation stage, you need to keep on iterating and evaluating the technology evolution parts, which all things in the current standards as well that they need to look at how the network to be evolved. For example, if you talk about open RAN, at certain point of time, you need to think of incorporating RIC in the network. Once RIC comes in, you need to think of different sources where the applications can be developed to realize the different use cases on the single network. And as the 5G is also evolving, you are at QAM 16 today, what can be expected out of release 19. To that extent, you should have a visibility understanding. So, logic has to be end to end because your today's investment choices, your today's vendor choices, will define how your networks will, I would say, remain competitive in the next 5, 10 years.

27:05 — Burr Sutter
I can tell you as someone coming from a software background, traditional enterprise IT background like myself, the person building web and enterprise-style applications, the world you live in, the telco world does look very different, and you have to think heavily about the investments you guys make when it comes to laying down the physical infrastructure out in the actual physical world, whether those be 5G antennas, and how they connect to MECs, and exactly what the distance is between antennas, and how all that data gets hauled back up to the central office. I think it's amazing what you guys have to do on a regular basis with that kind of infrastructure. Hopefully you also feel quite amazed yourself when it comes to the building out of this technology and how it impacts the world we live in.

27:49 — Sandeep Sharma
Oh, definitely. Because whatever projection that you do today, you never know how the traffic will shape up. Suddenly, an application comes in and that chokes the network, like anything. And these different OTTs coming in. So you need to factor in all those, I would say, unforeseen changes that might happen in the user patterns from the different sources, maybe from industries, maybe from OTT applications, maybe end users, and taking conscious decisions.

28:17 — Burr Sutter
I want to thank Sandeep Sharma for being our guest today. I am personally fascinated by how telecommunications and connectivity are reshaping our world. And exploring his deep industry expertise and perspective was absolutely fantastic. You can find the full transcript of this conversation and links to further reading about the Tech Mahindra and Red Hat partnership at redhat.com/codecommentspodcast. This episode was produced by Brent Simoneaux and Caroline Craighead. Our sound designer is Christian Prohom. Our audio team includes Leigh Day, Stephanie Wonderlick, Mike Esser, Johan Philippine, Kim Huang, Nick Burns, Aaron Williamson, Karen King, Jared Oates, Rachel Ertel, Devin Pope, Matias Faundez, Mike Compton, Ocean Matthews, Alex Traboulsi, and Victoria Lawton. I'm Burr Sutter, and this is Code Comments, an original podcast from Red Hat. Thank you so much for your time today and I hope you join us again next time.