Page 183 - ITU-T Focus Group IMT-2020 Deliverables

ITU-T Focus Group IMT-2020 Deliverables 3

9.3 Scenario 3: LTE in a slice

The softwarization of the 5G wireline network must of course permit multiple simultaneous RATs with
multiple simultaneous COREs and multiple simultaneous MEC functions. The 5G wireline must provide the
required isolation / connectivity and computational resources to guarantee correct QOS/QOE to each of
these 5G slices.
While slicing is intended to support the very different air interface requirements of different use cases in 5G,
there are a number of practical considerations with respect to migration from 4G to 4.5G to 5G and also the
requirement to support 4.xG flavours long after the deployment of 5G.
We expect therefore that a single 5G wireline infrastructure will be adopted well before 5G RATs/COREs are
actually deployed and that there will be initially one or two LTE slices running in this 5G wireline infrastructure
prior to additional 5G slices being enabled; The alternative which would require parallel 4G and 5G sets of
antennas, fronthaul, CRAN, DRAN, DC, backhaul and IDC interconnect resources is a vastly more expensive
and therefore undesirable option. While a common infrastructure will save CAPEX we also need to save OPEX
during any migration. The infrastructure’s orchestration software should therefore permit resources to
slowly be moved from the LTE slices to the 5G slices and then of course between the 5G slices as required.
If we break down the infrastructure network into its components (Figure 9.3-1) we can begin to analyze each
for their respective softwarization requirements implied by simultaneous support of both 4G and 5G.

5G concept of an end to end “Slice”(Network View)
CRAN CRAN RAT Back Core
Antennas Fronthaul fabric CPU/S/W Numerology Haul CPU/S/W

U = + + + + + +

S
If U is the set of all resource sets { Antennas, Fronthaul , ..} then
Slice S i is a set of resource subsets U taken from resource sets { Antenna, Fronthaul }
There are a nearly infinite number of possible slices | S * |
U
CRAN CRAN RAT Back Core
Antennas Fronthaul fabric CPU/MEM Numerology Haul CPU/MEM

S = + + + + + +
i
10
S 10 10 1

Figure 9.3-1 – End to end view of a slice through the 5G wireline infrastructure.
Antennas – ideally the antenna should support 4G and 5G. This means that software must be able to query
the capability of the antenna. If the antenna is MIMO then the question of if some segments can be used for
4G while others are used for 5G would lead to additional requirements on capability query. If an antenna can
be physically aimed or aligned then software would need to be able to control this hardware especially in the
case aim is different from 4G to 5G. Minimally the control system would need to know the position of the
3
antenna in some convenient R and also its current aim etc.
Fronthaul – if the fronthaul is not switchable there is not much for software to control. Some form of testing
and software reporting would be required to know if the fronthaul used for 4G can be re-used for 5G. If the
fronthaul is switchable then software must be able to connect a previously 4G antenna to its 5G RAT. If 5G
adopts a packet fronthaul then support for digitized analog over that packet fronthaul would be required to
support 4G and control of any intermediate switches for QOS guarantees to minimize jitter
(buffering/retiming) requirements would also be required.

177