Home5GWhy ultra-low 5G latencies are not assured, and Europe is worst placed

Why ultra-low 5G latencies are not assured, and Europe is worst placed

The physics of time-division duplexing in radio communications means 5G could, in certain frequency bands, struggle to achieve the ultra low latencies it promises. Worse, in some markets, regulatory forces have compounded the challenge with backwards rules about 5G radio operations.

As a consequence, radio engineers face an uphill battle to liberate the transformative power of 5G, and enable operators to address lucrative vertical sectors for the first time with their ultra-reliable, low-latency (URLLC) new tech.

This was the thrust of a useful 3GPP presentation at URLLC 2018 in London last week, which skipped through the latest achievements and newest challenges with the 5G New Radio (NR) standard.

Channelling Blue Peter, BT’s director of technology standards, Kevin Holley, brandished a couple of toilet-roll tubes in order to describe the difference between full duplexing techniques, and get to the bottom of the 5G latency dilemma.

After all, 5G latencies will not just depend on antenna types and antenna tilts, and whether or not communications run in millimetre-wave spectrum. The difference between time-division (TDD) and frequency division (FDD) duplexing matters, too, he said.

“These terms get thrown around, but one of the things people in the industry haven’t really understood is the difference between them,” said Holley.

The toilet rolls came out, and he positioned one at his mouth and one at his ear. This is FDD, he explained, with an uplink channel for talking and a downlink channel for listening. “You can do both at the same time; you can talk and listen at the same time.”

Holley removed one tube. With TDD, you have but a single channel, he explained – and you talk, and then you listen. “There’s a gap – a physics between the talking and listening,” he explained. The uplink and downlink packets have pre-allocated time slots in the frame structure, too.

Two-way TDD communications follow a pre-set pattern, which introduces latency.

Operators employ a mechanism called hybrid automatic repeat request (HARQ) to verify packet radio communications have arrived. This imposes a transmit-and-receive pattern. The frame structure for TDD in LTE networks defines a four millisecond delay between certain downlink and uplink communications (between slot three and slot seven in the below graphic).

As a result, latency is capped in LTE networks. “You can’t do latency lower than four milliseconds, even if your processing is zero,” said Holley.

The challenge for “ultra-reliable, low-latency” (URLLC) 5G is how it plays in TDD bands, notably the 3.5 GHz band, defined by the ITU as part of its IMT-2020 vision as spectrum between 3.3 GHz and 3.8 GHz. In Europe, the 3.5 GHz band is the principle allocation for delivering 5G communications.

3GPP has sought to overcome these limitations in 5G NR by breaking up the LTE frame structure. Release 15 divides the old millisecond TDD time-slots into “mini slots”, explained Holley. “Rather than insisting you talk and then listen, you can listen, and then do a quick-talk while you’re still listening.”

There remains an issue with interference for 3GPP to sort out in Release 16, but the assumption is 5G operators will be able to apply this “mini-slot” technique in TDD spectrum to get latency down, “very, very low”. There is cliffhanger, however, in the UK at least.

Ofcom, regulating operators like Holley’s employer in the UK, has already defined frame structures for certain spectrum in the 3.5 GHz band, during the auction of 150MHz of frequency north of the 3.4 GHz-mark in June. These bear more than a passing resemblance to the old LTE TDD patterning.

Which means, ultimately, 5G could also be limited to four-millisecond latencies in most cases in Europe. Disaster, as Holley suggested. “You’ve got to ask yourself, how are we going to deliver low latency if we have to follow what Ofcom’s asked us to do?”

UK operators have some thinking to do. They must hope, as well, that Ofcom does better next time around, when it sets the operating conditions for radio spectrum at 3.6 GHz to 3.8 GHz – “so we can deliver something more in the URLLC vein”.

cyber threats
Previous post
Cyber threats to industrial control systems
Next post
Western IoT brands lose 15% share in 12 months as China ramps up NB-IoT focus