Since 5G made its initial appearance this year, much of the emphasis has been on phones. Most, if not all of the data I have been privy to talks about the performance of these 5G devices, even dissects them to explain how they achieve 5G performance. That is all well and good. However, much less attention is being paid to the infrastructure – the network, at least, publicly.
5G phones are, pretty much, a no-brainer and the least of the 5G ecosystem’s hardware worries (will Apple’s iPhones use Qualcomm or Intel modems, or design their own?).
In fact, more than 80 percent of 5G challenges are in updating and creating networks to seamlessly support the variety of services and capabilities that have been hyped. Challenges such as getting the various vectors (frequencies, modulation schemes, internetworks, hardware compatibility, etc.) to all work seamlessly will be a challenge 5G will face for at least the next decade.
The 5G network infrastructure will not have the luxury of coverage holes as 4G currently has. The reason for that is 5G will connect everything – the Internet of Everything/Everyone (IoX), autonomous vehicles, streaming media, smart “X”, the edge, and more. For all of the components to work efficiently and effectively requires 24/7/365 ubiquitous interconnect.
While that is achievable in populous areas (with some exceptions such as underground and in-building), it is not that easy in areas where there is little civilization. Parts of Texas and California, as well as large swathes the West come to mind. And, this is much more prevalent in other countries such as Canada, Africa, the Middle East, Russia, China – the list goes on).
One might argue that uninhabited or sparsely inhabited areas are not on the radar screen for early 5G coverage and the real need is in densified areas. With 4G, that was an acceptable argument. However, the 5G vision has been hyped as connectivity for everyone and everything, anywhere, so that argument is not acceptable, especially with the IoX, and autonomous vehicles (both of which will need to be connected regardless of where they are).
So, what are the major challenges? Let us drill down on that a bit.
Perhaps the most overwhelming challenge is developing a fully virtualized core. The current ecosystem of dedicated and single-purpose functions (NIC, RAN, switch, etc.) is akin to the old analog phone systems, incapable of provisioning new services or offering an agile platform capable of dynamic adaptation to conditions and demands.
The future is virtualization. Termed Network Function Virtualization (NFV), it is the platform that will enable all of the goodies 5G promises. However, it will require a new core. Without it, technologies and platforms — such as QoS, dynamic spectrum sharing (DSS), network slicing, edge intelligence, and more — cannot happen.
This new core will require a major changeout (or add-in) of hardware because current systems are a hodge-podge of propriety mixed solutions consisting of ARM, PowerPC, MIPS, DSP, and proprietary chips. There is virtually no architectural standardization.
However, there is another critical component, without which, all the virtualization in the world will be moot. That is AI. Tomorrow’s networks will be much too complicated to administer manually. They must be self-managing (optimizing, organizing, healing, etc.) and autonomous with minimal intervention by humans. That cannot be accomplished without AI.
Once all of this technology is ready for prime time, there is a third challenge – deployment. Installing this in existing sites is not the issue. The segment of 5G below 3 GHz will not be that much of a problem. But above 6 GHz will.
Millimeter wave (mmWave) is where the action is for applications that require wide bandwidths and low latency. Some of this can, of course, be achieved at lower frequencies, but that space is limited. MmWave is what promises to enable much of the IoX, autonomous vehicles and various segments of smart “X”.
And, herein lies a big challenge. How to get that coverage. Because 5G mmWave platforms will, generally, be low power. Add to that the propagation characteristics of mmWave and, well, we all know that will require dense radiating elements – antennas everywhere.
That will be the toughest part of this “big challenge” – finding the real estate and resources (power, back/fronthaul) for them.
So, while much of the attention is on handsets, let us hope that the industry is working under the sheets to resolve the real problems – the network.