[MUSIC] I think we saw a different type of RAN configuration, like dRAN, cRAN, vRAN. Does my 5G phone need to know which one it's talking to? >> Just like many other configurations regarding network, this fortunately is fully transparent to your phone. The way to understand that is, let's go back to our ATM analogy that we discussed a while ago. Your ATM card functions exactly the same whether you are inserting it into into an ATM that is close to the bank or whether that ATM is far away from the bank. Your ATM card doesn't need to know what kind of ATM it is. As long as it belongs to the same bank, your ATM card will function exactly as expected. Just like that, your phone does not need to know what kind of configuration the RAN is operating in, as long as it transmits the necessary signals which every RAN will. Your phone doesn't need to worry about it, all that is seamlessly and rather beautifully abstracted out from your phone. Only the network needs to be worried about it. So to summarize, no, your phone doesn't need to concern itself with what kind of configuration the network is operating with. >> I see, and you mentioned there are so many advantages of vRAN. If it's so awesome, why didn't we have it earlier for previous technologies, let's say? >> The answer to that goes more towards the advent of virtualization itself rather than the limitation of LT as a technology. LT was designed back in 2006, 2007, that era. And around that time, virtualization as such wasn't exactly as prevalent as it is today. Sure, we did hear about cloud computing back then, but it wasn't as integral a part of our lives as it is today. And that is the reason why the system designers chose to stick with the traditional model of designing and deploying the RAN. Rather than bringing in this additional factor of virtualization which might have added complexity. But like I said, there is no technological limitation on LT that makes it somewhat difficult or impossible in that there definitely are certain virtualization solutions that are applicable to LT RANs as well. But because of the way it was designed, the virtualization paradigm in LT is quite limited. Nearly all major LT deployments these days are in favor of traditional RAN. There is another factor that may have influenced this decision the way I think about it. Virtualization is beneficial when you deploy your network at scale. So when you look at cell densification, that 5G will possibly have to deploy, then instead of having just let's say 10 LT cells in a given region, you might need 15, 20 5G cells. And when you scale your deployment, that is when virtualization begins to make more and more sense from a commercial standpoint, amortization of resources essentially. Because LT cell count was somewhat limited, it is quite possible that virtualization may not have made as much commercial sense for LT as it definitely does for 5G. So those are the couple of reasons for which I think virtualization may not have gotten much traction with LT. Although I repeat, there was no fundamental technological limitation that absolutely prevented virtualization from happening. >> Got it. For virtualization RAN, we're moving towards that way, can it support all of the use cases that the traditional RAN that used to do? Or any use cases under certain scenario we still prefer to use a traditional? Or maybe the other way around, that this use case that we really, really want to use virtualization. >> Another good question. So I heard three parts to that. That is, is vRAN capable of supporting all the use cases that your traditional RAN can? In a nutshell, the answer is yes. As I mentioned that vRAN and traditional RAN, as far as their wireless functionalities are concerned, they are practically identical. So any use case that you can satisfy with traditional RAN can also be satisfied with virtual RAN. The second part of your question, that is whether there may be any use cases where we were still somehow prefer traditional RAN? There may be certain use cases, and to understand that, let's go back to the discussion we had about the cellphone tower and the intelligence that governs the behavior of the cellphone tower. Let's say that your wireless channel is so highly variable, let's say you are moving at freeway speeds, so your wireless channel is changing every few milliseconds. Now, the cellphone tower needs to depend on that proverbial intelligence in order to make decisions about the subsequent operation of the cellphone tower in light of the highly variable wireless channel. Now, in an extreme case, let's say that intelligence is located several hundred miles away. And it takes several milliseconds for the cellphone tower to reach that intelligence, provide the necessary information to that intelligence, and get a response back in order to take corrective or preventative action. If it takes the whole procedure, let's say ten milliseconds, and the channel is changing within five milliseconds, it wouldn't make sense to rely on a virtualized intelligence. So this is one use case that I can think of where your traditional RAN might still have some advantage over virtualized RAN. Although that advantage is minimized by bifurcating the intelligence that we have been talking about into multiple layers, so to speak. And still have some of the necessary layers locally at the cellphone tower. And only relocate some of the other layers which are not involved in making such a split second real time decisions in the intelligence that you need to make real time decisions that can still reside with your cellphone tower even in a virtualized paradigm. And that might mitigate the possible disadvantage that virtual RAN might have in this case. The third part of the question as to whether there are any use cases where were virtual RAN were preferable to traditional RAN? Absolutely, there are many, many used cases. Once again, let's jump back to cell densification. Let's say you have ten LT cells or ten LT base stations or towers in a given region. And for 5G, you are going to densify that network and deploy, let's say 20 cells. Your traditional RAN would require that 100% of that intelligence we have been talking about be present at each of those 20 cellphone towers. Once again, while that sounds very easy and manageable on paper, when the rubber actually hits the road, in practice, that is going to be very expensive to have that piece of software running on every single base station. And that is one paramount use case wherein a virtualized RAN would have significantly more advantages over a traditional RAN. The way it manifests in reality is, consider a private network that is being deployed in an industry. Conditions in a typical manufacturing plant are changing very dynamically. One day you might have a big machinery over here, the other day that machinery might move to another place. It means that the conditions in which the network has to operate, they change frequently as well. And that might entail that you have to physically move your small cell base stations from one point to another. And if it were a traditional RAN, doing so would be much more difficult. Whereas in a virtualized RAN, doing so is much easier. Because no matter where your cellphone tower is, your intelligence, or at least part of it, always resides somewhere in the back end in the cloud, so to speak. Which does not care where your precise base station is located on that particular day. So cell identification, private networks, industrial factories with 5G networks. This is one prominent use case where virtualized RAN will definitely have tangible and provable advantages over traditional RAN. And why exactly that is the case, those are some of the aspects that we are going to touch upon when we will get into the details of private networks and industrial IoT in the next module.
