Converged backhaul and fronthaul considerations. Jouni Korhonen Broadcom Ltd. 10/26-28/2016 IEEE TF

Transcription

1 Converged backhaul and fronthaul considerations Jouni Korhonen Broadcom Ltd. 10/26-28/2016 IEEE TF

2 Compliance with IEEE Standards Policies and Procedures Subclause of the IEEE-SA Standards Board Bylaws states, "While participating in IEEE standards development activities, all participants...shall act in accordance with all applicable laws (nation-based and international), the IEEE Code of Ethics, and with IEEE Standards policies and procedures." The contributor acknowledges and accepts that this contribution is subject to The IEEE Standards copyright policy as stated in the IEEE-SA Standards Board Bylaws, section 7, and the IEEE- SA Standards Board Operations Manual, section 6.1, The IEEE Standards patent policy as stated in the IEEE-SA Standards Board Bylaws, section 6, and the IEEE-SA Standards Board Operations Manual, section 6.3, 2

3 IEEE TF NGFI Bomin Li Practical approach to converged FH/BH network architecture and functional partitioning Date: Author(s): Name Affiliation Phone [optional] [optional] Jouni Korhonen Broadcom Ltd dcom.com 3

4 Outline Architecture proposal for converged fronthaul and backhaul network for 4.5/5G RAN. Proposal for NGFI interfaces based on different functional splits. 4

5 Objective (same from Aug meeting) Evolutionary path from 3/4G to 5G RAN. Identify the essential features from 4.5/5G RAN transport circuit & equipment realization point of view: Flexibility vs Bandwidth/time-synchronization/complexity/cost. Propose an architecture and functional splits to 4.5/5G RAN that: Allow E2E packet & Ethernet solutions. Allow converged fronthaul and backhaul network deployments. Scale up to 5G numbers keeping align with optics evolution. Aim at transport level interoperability. 5

6 Proposal Functional splits should aim for simplicity: Identify the most common and important functions that are easy to design 5G ready. Adopt the three interfaces proposed in this contribution as a baseline: NGFI1 lower layer splits ; high volume of nodes, lower bandwidth per device but tight synchronization demands. NGFI2 lower layer splits ; aggregation, converged front- and backhaul, high aggregated traffic volumes and tight synchronization demands. NGFI3 higher layer splits ; with full service provider functions. High aggerated badwidth per node. 6

7 Functional splits and radio features.. Higher layer splits Lower layer splits User processing Cell processing RRC PDCP RLC Low- RLC MAC Low- MAC PHY Low-PHY RF Data Option 1 Option 2 Option 3 Option 4 Option 5 Option 6 Option 7 Option 8 RRC Data PDCP DC (1A like) RLC Low- RLC MAC Low- MAC PHY Low-PHY RF DC (3C like) e.g., CU: ARQ + buffering + reorder, DU: seg/reassm. Centr. ARQ,.. CU: coord. sched., ICIC,.. DU: HARQ,.. (adapted from 3GPP RAN3 #93 material these are just examples, do not take them too strictly..) DL CoMP, CA, PartialUL CoMP,.. BetterUL CoMP... UL CoMP,.. Insert Title here Insert Date here 7

8 Functional splits and impact to transport RRC PDCP RLC Low- RLC MAC Low- MAC PHY Low-PHY RF Data Option 1 Option 2 Option 3 Option 4 Option 5 Option 6 Option 7 Option 8 RRC PDCP RLC Low- RLC MAC Low- MAC PHY Low-PHY RF Data End-2-end max latency Bandwidth scales.. ~100ms >20ms >10ms ~0.2ms (~TTI) <40us (~0.2ms TTI) Tens of Gigabits(*) Tens++ of Gigabits Terabits (*) Assumption: 5G radio with 1GHz air spectrum and M-MIMO, 0.2ms TTI.. (adapted from 3GPP RAN3 #93 material) Insert Title here Insert Date here 8

9 Mapping to a high level architecture Evolved Aggregator Enterprise vran/mec DC vran/bbus/ MEC DC Intra- or Internet Old BTS Caches etc Aggregation Metro S1/NG3 Legend: IP/Eth/MPLS Backhaul Fronthaul(p2p connections) Packet-based fronthaul vran-vran X2-like midhaul 3C-like split midhaul 9

10 Mapping to a high level architecture Evolved Aggregator Enterprise vran/mec DC vran/bbus/ MEC DC Intra- or Internet High Split L2/L3 Intra- or Internet S1/NG3 Old Aggregation Metro BTS Caches etc Legend: IP/Eth/MPLS Backhaul Fronthaul(p2p connections) Packet-based fronthaul vran-vran X2-like midhaul 3C-like split midhaul 10

11 Time-synchronization accuracy requirements Evolved Aggregator High Split Old BTS Caches etc Aggregation Metro S1/NG3 Up to <12.5 ns phase/time, ~50(?)ppb (*) < 1.5 µs phase/time (*) Simplified view; sync requirement depend on used radio features and location of common masters, etc. 11

12 Latency requirements Evolved Aggregator High Split Old BTS Caches etc Aggregation Metro up to hundre(s) of microsecond(s) S1/NG3 tens of microseconds tens of milliseconds 12

13 Bandwidth requirements Evolved Aggregator High Split Old BTS Caches etc Aggregation Metro S1/NG3 Terabits.. Tens of Gigabytes or to hundreds of Gigabits(due aggregation) tens to hundres of Gigabits 13

14 Network domains per node requirements Lot of nodes Aggregation nodes 10/25G links. SyncE, preemption, (sub)ns accurate timestamping for 1588, 802.1AS,.. Eth/IP/MPLS. Evolved Aggregator (10/)25/50/100G links. (sub) ns accurate timestamping for 1588 and 802.1AS. Preemption & SyncE beneficial. Eth/IP/MPLS. Enterprise vran/mec DC vran/bbus/ MEC DC Typical service provider equipment. 100G links, Eth/IP/MPLS. Existing backhaul requirements apply. Intra- or Internet High Split L2/L3 Data center switching.. Intra- or Internet S1/NG3 Old Aggregation Metro BTS Caches etc 14

15 About interoperability targets 1/3 What are the assumptions of interoperability? Purely at the transport level? Vendor A evolved RRH Vendor A BBU Vendor A Split S1/NG3 Vendor B evolced RRH Vendor B Split (if any) Vendor B BBU 15

16 About interoperability targets 2/3 Or promoting some common split(s) that would ensure interoperability beyond transport level? Can be very(!) hard to get any agreement on.. A fixed split is dangerous regarding future proofness.. Vendor A evolved RRH Vendor A BBU Common Split S1/NG3 Vendor B evolced RRH Vendor B BBU 16

17 About interoperability targets 3/3 Accept the fact that splits are moving and evolving entities. Think a split as a side card in a networking device or an additional hop in the network.. Interoperability still remains at the transport level. Vendor A evolved RRH Vendor A BBU Common Split Aggregator Vendor B evolced RRH Vendor A split engine Vendor B split engine Vendor B BBU S1/NG3 17

18 Proposal high level architecture Radio Aggregation Units Small number of (flexible) splits Evolved Aggregator Enterprise vran/mec DC vran/bbus/ MEC DC Intra- or Internet High Split L2/L3 Intra- or Internet S1/NG3 Old Aggregation Metro Transition from 2/3/4G to 5G BTS Caches etc Xhaul support? Legend: IP/Eth/MPLS Backhaul Fronthaul(p2p connections) Packet-based fronthaul vran-vran X2-like midhaul 3C-like split midhaul 18

19 Proposal transport interfaces A *lot* of nodes, low power, simple split-l1 functions, ideally no state,.. Many nodes, mappings to RoE, advanced split-l1 functions, processing element interface,.. high bandwidth, a lot of queues, Dual Connectivity type functions,.. Fewer nodes, high bandwidth, service provider funtions NGFI1 NGFI2 NGFI3 <- Interfaces 19

20 Interface Summary NGFI1 A lot of nodes with ~10-25G links. Tight network sync requirements up to 12.5ns time alignments.. but rather homogeneous traffic profiles. End-2-end latency tens of microseconds. Network aggregared bandwidth up to Terabytes. Mainly Ethernet & MPLS over fiber. NGFI2 Many nodes up to 100G links; up to close terabit scale. Tight network sync requirements up to 12.5ns time alignments.. Heregeneous traffic profiles (converged network enabling features needed from nodes). End-2-end latency tens of microseconds. Network aggregared bandwidth in tens to hundres of Gigabytes. Ethernet/IP/MPLS over fiber. NGFI3 Fewer nodes; terabit scale; 100G links. Network sync requirements in backhaul class. Heregeneous traffic profiles. End-2-end latency measured in scales of millisecond. Network aggregared bandwidth in hundres of Gigabytes. Ethernet/IP/MPLS over fiber; service providers features required. Insert Title here Insert Date here 20

21 Motion # Agree as a baseline the high level architecture and NGFI interfaces described in slides 18 and 19 of tf1_1610_korhonen_converged_1.pdf. Mover: Jouni Korhonen Seconder: Yes: No: Abstain: (technical motion needs >= 2/3)