Documente Academic
Documente Profesional
Documente Cultură
Autonomous Vehicle Intelligent Transport Smart Education Remote surgery Task automation Smart Utility Mgmt Mission-critical PTT Augmented Reality /
Online Predictive Systems Smart Agriculture Wearables Predictive (Water, Gas Metering, Mission-critical video Virtual Reality
Maintenance UAV-based Maintenance power, pollution …) (high upload) Gaming
Smart Grid Remote consultations
Infotainment surveillance / Telemedicine Mission-critical control Smart Traffic Mgmt Mission-critical
Smart airport (traffic routing, sensors (drones,
Platnooning Smart metering
parking, monitoring..) smoke detector,
Fleet Management
security camera)
Railway Signalling
Network Programmable
New Radio (NR) Transport Management
Elements Networks
• Flexible numerology – allows multiplexing of services with qualify and latency requirements and also • Service Based Architecture (SBA) – stateless, open, flexible and realization using VNFs, enabling
large SCS for mmWave movement and scaling of AFs dynamically
• Spectrum - Allocation of higher frequency bands – ensures additional spectrum and wide bandwidth • CUPS: Separation of control and user-plane
availability, ensuring support for very high data-rates • SDN: Improved QOS model for packet flow and policies. Helps defining service chaining for SDN
• Adaptable air-interface - scalable sub-carrier spacing, variable slot-lengths, scalable TTI, minimize based data flow
control overhead, short PUCCH (for latency) and long PUCCH (for coverage), advanced channel coding • NFV: Orchestration and Virtualization (NFV) – de-couple logical function from hardware
techniques, flexible HARQ
• Slicing – logical end-2-end networks tailed to customer needs
• Self-contained slot structure (TDD) to reduce latency – adaptable UL/DL switching, data/ACK in
same slot, SRS in every slot etc • MEC Support for low-latency services and offloading of data at EDGE. It provides Computing
resources, Caching, Low latency and less traffic through core to meet the requirements for use-cases
• Ultra-lean design to enhance network energy performance – minimizing always-on signals, reduce
• Exposure Functions, APIs, Common API Framework – to enable external interworking with 3GPP
periodicity of PSS/SSS/PBCH
• Shortened TTI and processing– reduces latency • New 3GPP accesses: wire line-wireless convergence, satellite access. Also allows subscribe to events and
have analysing and optimizing network performance and behaviour wrt services being offered
• Support of carrier aggregation of upto 16 carriers,
• Beam-centric design enabling usage of beamforming and massive number of antennas to improve
performance
Copyright © 2018 Aricent. All rights reserved. 3
5G Transport Architecture Terminology
RRU DU CU/MEC
eCPRI F1 NG CN
Aricent FH-TSS
4G - CPRI
Aricent FH-TSS UNI Aricent FH-TSS UNI Aricent FH-TSS UNI
1914.3 RoE
eCPRI
eCPRI
Fronthaul
UNI
Time Server
Latency
~100-500us Ref: T-TUT-HOME-2018-MSW-E
~100us 1.5 to 10ms
UP latency
eMBB – 4ms, URLLC – 0.5ms
Coverage
FH 1-20km, typically p2p MH 20-40km, p2p or p2mp BH upto ~200km, p2mp or mp2mp
✓ Fronthaul – Network between RRU/RU (Remote Unit) and DU (Distributed Unit) – can be CPRI or eCPRI or IEEE 1914.3
✓ Midhaul – Network between DU and CU (Centralized Unit) – “F interface”
✓ Backhaul – Network between CU and 5G NGC (and EPC)
RRH
Ethernet
BASE BAND BASE BAND BASE BAND
DU
fiber F1
fiber BASE BAND BASE BAND BASE BAND
BASE BAND
CONTROL UNIT CONTROL UNIT
CONTROL UNIT
TRANSPORT UNIT TRANSPORT UNIT
TRANSPORT UNIT CU
vBBU
• RAN -split options helps to reduce the fronthaul requirements and also allow flexible and scalable HW implementations
Data-rate
Latency
7a: 10.1-22.2Gbps
7b: 37.8-86.1Gbps 157.3Gbps
DL data 4Gbps 4016Mbps 4000Mbps 4000Mbps 4133Mbps 7c: 10.1-22.2Gbps
Data
Data
7a: 16.6-21.6Gbps c
UL data 3Gbps 3024Mbps 3000Mbps 3000Mbps 5640Mbps 157.3Gbps
7b: 53.8-86.1Gbps
7c: 53.8-86.1Gbps
Latency 10ms 1-10ms ~100us ~100ms 250us
Scenario - 100MHz and 256QAM UL/DL, MIMO layers – 8 UL/DL, Number of antenna ports – 32,IQ – (2*7-16)) UL/DL
Latency requirement becomes stringent and data-rates also increase as we move to option-7/option-8
Option -2 Option -7.x Option -8 • Most commonly used RAN-Split options are Option-2, Option-6,
Option-7.x and Option-8
CU CU CU
• Option-8 is equivalent to Small-cell type of realization
RRC RRM RRC RRM RRC RRM
CPRI eCPRI
• CPRI is pre-dominantly used in 4G fronthaul. Max data rate supported in CPRI v7.0 is 24.33 • eCPRI is used as fronthaul between CU/DU and RU for 5G network, using packet based fronthaul
Gbps (rate 10) transport network
• For typical LTE scenario of 20MHz, 2x2 DL MIMO, the fronthaul data rate is ~1.96Gbps • Enables flexible functional decomposition while limiting the complexity of the eRE - Supports for
• The IQ data of different AxCs are multiplexed by TDM scheme onto an electrical or optical Ethernet interface types – 10G, 25G, 40G and 100G
transmission line, and link is always “ON” with Constant bit-rate data • More traffic dependent rather than antenna dependent, and and Ethernet can handle this with
• Specified for point-to-point topology and is more antenna dependent (rather than traffic support of statistical multiplexing
dependent) • Enables realization of SDN/NFV based fronthaul
Copyright © 2018 Aricent. All rights reserved. 8
5G Fronthaul Transport Requirements
Handling of traffic and for different service types (or slices) with varied priorities (include packet prioritization) and quality of service
over a unified network
Flexibility to scale the bandwidth based on user plane traffic
Cost / Performance Trade-off by selecting proper FH/BH suitable for the network deployment
p(D) = offered traffic can be transported without queueing with a probability Ref: 5G transport network requirements for the next generation fronthaul interface
• CU/DU Hardware
Migration towards Software-defined “Open” Network philosophy • 5G is accelerating the adoption of commodity HW,
disaggregated solution within Telco’s.
Monolithic, Custom build Solutions Flexible, COTS based Solutions • Limited SOC options available for 5G RAN realization,
CPU based (x86 or ARM) solution with FPGA used
Closed and Proprietary interfaces Open and modular interfaces and flows
• ASIC based solutions for baseband will come into picture
Built around available network Network defined by Services once the solutions are verified
Pre-tested & HW Platform Independent • Built around usage of open-source and open interfaces in overall
Feature Rich & Cost Effective
solutioning along with 3GPP standards
INSTANCE
INSTANCE
INSTANCE
INSTANCE
INSTANCE
INSTANCE
INSTANCE
INSTANCE
• Fully programmable, FPGA based and enables disaggregated “cloud”
based architecture
• Optimization at Platform Level and replaces standard NICs HYPERVISOR HYPERVISOR
• Can be programmed and scaled on-demand resulting in a real “elastic” cloud TRADITIONAL NIC SMART NIC
Programmable SmartNICs help in accelerating critical BB and security workloads and migrating acceleration
services
Domain Specific
Architecture