Comment by throw0101b
2 years ago
> The simple answer is that although it's early days for DECT-2020 NR, it promises to fill a genuine 'gap' in the wireless IoT market for massive machine-type communication. An area where failure is not an option and could put at risk automation processes, critical infrastructure, livelihoods, if not lives themselves.
With regards to reliability, Wifi 8 seems have been dubbed "Ultra High Reliability" (UHR), as that will be its area of focus:
* https://en.wikipedia.org/wiki/IEEE_802.11bn
> This amendment defines modifications to both the IEEE Std 802.11 physical layer (PHY) and the IEEE Std 802.11 Medium Access Control (MAC). The amendment adds an Ultra High Reliability capability to a Wireless Local Area Network (WLAN). The Ultra High Reliability capability is defined for both an isolated Basic Service Set (BSS) and overlapping BSSs as:
> *At least one mode of operation capable of increasing throughput by 25%, as measured at the MAC data service Access Point, in at least one Signal to Interference and Noise Ratio (SINR) level (Rate-vs Range), compared to the Extremely High Throughput MAC/PHY operation, and
> *At least one mode of operation capable of reducing latency by 25% for the 95th percentile of the latency distribution compared to the Extremely High Throughput MAC/PHY operation and
> *At least one mode of operation capable of reducing MAC Protocol Data Unit (MPDU) loss by 25% compared to the Extremely High Throughput MAC/PHY operation for a given scenario, especially for transitions between BSSs.
* https://grouper.ieee.org/groups/802/11/Reports/tgbn_update.h...
* https://www.ieee802.org/11/Reports/802.11_Timelines.htm#TGbn
> An area where failure is not an option and could put at risk automation processes, critical infrastructure, livelihoods, if not lives themselves.
Which is exactly 5G's sales pitch, which is designed for low latency and high reliability aimed at critical applications like factory automation, remote surgery, self-driving cars, etc. And there is currently a push for 5G private networks.
So it remains to be seen if this gets any traction.
The problem is that the 5G hype isnt telling the whole story. Yes, you certainly could drive high thruput across the radio access network, at low latency, and yes, thats exactly what you would need to do things like self driving cars. The problem is that the radio is just one tiny part of the whole communication chain, and for safety-critical things, the entire chain must be equally fast and robust.
That means, for example, that the chip inside the car/robot must detect failures in the transmission path incredibly quickly and switch to a secondary channel, that the radio controller can detect when the network (packet core) it thought it was talking to goes away, and recover, and that the packet core itself can detect failures in its components and fail over or restart. It has to do all this in the time it takes for a warehouse robot to crush a worker, or a car to hit a bollard. Did I mention that todays packet cores are built from kubernetes and prayers? This degree of safety simply isnt happening anytime soon.
Perversely, it might actually be safer to deploy an entirely private network under the control of an enterprise and take your lumps there as best you can, than rely on an operator's network being able to do what you want all the time.
The radio is just a small part of 5G. The whole system is designed for these use cases.
So yes, this requires upgrades to fully compliant packet cores, which is expensive and not really happening because there is actually no business case at the moment.
Anything needing high reliability is plugged in. Actual remote surgery involves an on site backup surgeon, and redundant private wired links.