5G spec process starts for real at 3GPP meeting in Busan

Today is the last day of a five day meeting being held in Busan, South Korea.

The meeting, of the 3GPP’s RAN Working Group one and officially numbered 3GPPRAN1#84, has seen around 400 delegates gather together to develop specifications of next generation radio network and air interface technologies.

Host of the meeting, Samsung Networks, described the meeting as the first at which work to determine specifications of 5G radio and air interface standards happened in earnest.

A release from Samsung said, “This meeting [is] the starting point of 3GPP’s in-depth research on 5G candidate technologies.”

So what have 3GPP’s members been discussing this week? Well, as you can see (below) things get pretty technical pretty quickly, but on a high level the RAN group have been looking at some fundamentals of how to design an air interface that can meet different demands in terms of latency and throughput performance, at a wide range of spectrum bands, to a wide range of devices.

That means determining the underlying components of a new radio access technology such as waveform, channel coding schemes, frame structure and length, and numerology.

There were 1,306 submissions made prior to the meeting, covering the gamut of the nuts and bolts of new structures, use cases, modulation schemes,  modelling evaluations and so on. 1,306.

It’s nearly all new, and that should put a question mark over the claims from some quarters to be able to “accelerate” 5G, or even in some extreme instances to have 5G services by next year. Really, to make that claim will be a stretch.

It’s also interesting that the proposal from the operator member China Unicom put as a key requirement that both CAPEX and OPEX can be reduced as part of the design principles.

In more depth

Amongst those 1,306 submissions prior to the event for discussion or decisions, here are are just a very few specifically related to 5G, or naming 5G:

Basic Principle for the 5G New Radio access techology – This piece, from Alcatel-Lucent Shanghai Bell and Nokia, addresses high level design principles on waveform, numerology and beam forming. It proposed UF-OFDMA with windowing sas a solution to impove spectral efficiency, multiplexing of different numerologies and supporting unsynchronized transmissions mainly below 6GHz. For higher frequencies (30-90GHz) the proposal is for the “zero-tail DFT-spread OFDM”. This latter waveform has favourable peak to average power ratios that it said can mitigate more difficult propagation scenarios.

On numerology principles the proposal was for “a set of different numerologies options to support different spectrum and deployment requirements”. It’s interesting to compare this with the proposal outlined below from Huawei, which proposes f-OFDM and also proposes the co-existence of different numerologies.

The paper also addressed the frame structure, stating that the subframe length should be 0.125ms, with the number of symbols in the subframe dependent on the used numerology.

The proposal said that single and multi-user MIMO is going to be “essential” in 5G, and it called for flexible support for different numbers of TX and RX antennas at the base station and user device, with support for different TX and RX architectures.

Overview of the 5G frame structure – a revision from Huawei and HiSilicon of the potential numerology and frame structure for the 5G RAT. The principle aim here is to design a structure that can meet diverse requirements in terms of devices, spectrum bands (including unlicensed bands), deployment scenarios and services.

The proposal is to enable flexible resource multiplexing between air interfaces in 5G by designing a single unified air interface framework with multiple numerologies to allow the coexistence of 5G RAT and LTE in a single continuous block of spectrum and also offers the forward compatibility for further enhancements (e.g. new vertical applications).

The proposal argues that that can be done using the f-OFDM scheme to mitigate inter-subband interference if different numerologies are used between subbands. Here’s an example that shows several frame structure configurations for diverse service requirements, such as eMBB unicast with normal CP, MBMS service with extended CP and mMTC with the smaller subcarrier spacing of 3.75 kHz (e.g. taking NB-IoT numerology as an option).

flexible frame structure for 5G

The proposal therefore is the the new radio frame structure should be designed to support a flexible structure on parameters like subcarrier spacing, cyclic prefix length, TTI length, subframe duration and number of OFDM symbols in a subframe. More specifically,

  • The sub-carrier spacing varies with the frequency of the spectrum and/or the maximum UE speed to minimize the impact of Doppler shift and phase noise.
  • The CP length may vary with deployments like outdoor or indoor due to different delay spread requirements, may vary with spectrums like low or high frequency bands, may vary with services like broadcast or unicast, and may depend on whether beam forming techniques are used.
  • TTI length can be flexible with service types, and can be flexible for downlink, uplink and sidelink. For example, TTI length can vary with the latency requirement of the service. TTI length can be set independently for downlink and uplink, for the alignment of coverage and/or different traffic requirements in downlink and uplink.
  • For TDD, downlink-to-uplink switch-point is needed, thus flexible frame structure for TDD should also support flexible guard period (GP) configuration for different requirements, including relays.