open ran - otwarta architektura sieci 5g€¦ · lean design, tqm, agile, kanban, customer...
TRANSCRIPT
1
Open RAN - otwarta architektura sieci 5G
Sławomir Pietrzyk and Andrzej Miłkowski
PLNOG 23 Conference, Kraków 24th of September 2019
2
IS-Wireless is an advanced wireless communications company
Mission:
To become global provider of software-defined 4G and 5G mobile networks
About
3
Team: 40+ skilled telecom professionals
Protocol EngineersUMTS, LTE, 5G NR, WiFi, IoT, PLC
R&D Engineers10 PhDs, SDN. NFV, RRM, Cloud, MANO, Edge, Slicing, PHY, OFDM(A), MIMO
Software DevelopersC, C++, Java, PythonRTS, QT
Team LeadersLean design, TQM, Agile, Kanban, Customer Development
Office300m2 in Piaseczno HQSmall office in ŁódźSmall office in Warsaw(Warsaw University of Technology)
4
Why now: technology curves and company activities
2006
R&D on 4G (LTE)
20222010 2014 2018 2020
R&D on 5G
Implementation of 5G mobile networks
R&D on 6G
Implementation of 4G models
Sales of 4G models
Sales of 5G mobile networks
5G opportunity window4G opportunity window
Implementation of proprietary networks
• IS-Wireless is present on the 4G market since 2006 what gives competitive advantage in the form of: team, know-how, customers, partners
• IS-Wireless is present in R&D on 5G since 2012 what gives sound understanding and confidence in systems’ evolution
R&D Projects
Product Development and Commercial Projects
7
Speakers
Sławomir PietrzykFounder/CEO
At IS-Wireless resposible for defininig the long-term technical vision. Expert in
wireless technologies and passionate entrepreneur. PhD in wireless from Delft
University of Technology, pioneer work on OFDMA.
Andrzej MiłkowskiProduct Manager
At IS-Wireless responsible for managing SD-RAN/Core/Mano products. Passionate about mobile technology and IT, pioneer
of mobile communications in Poland participating in creation Era, Heyah and
Play.
10
Expectations
Higher system capacity: 1000 x capacity / km2
Energy savings and cost reductions
Reduced latency: < 1 ms
Higher data rate: 100 x typical data rate (even for high mobility)
Massive device connectivity: 100 x connected devices (even in crowded areas)
5GTARGETS
We are very far from the 5G targets
Meeting them requires a new approach to network deployment
11
Problem: typical Radio Access Network (RAN) deployment
• High cost
• High power (EMF exposure)
• Very long deployment time (12+ months)
• Low spectral efficiency [bit/s/Hz/m2] (capacity per m2)
• Significant coverage holes
• Poor indoor signal penetration
12
Problem: zoom-in
scheduler
3GPP stack
➢ Core centralized and separated from RAN
eNB/gNB
chipsetCore NetworkInternet / OTT
➢ RAN functionality locked to a particular chipset
➢ Scheduler per eNB/gNB causing fierce competition for resources
➢ Single eNB/gNB covering large area –low frequency reuse
➢ Low spectral efficiency [bit/s/Hz/m2]
➢ Interference between eNB/gNB
➢ High power eNB/gNB
13
Problem: zoom-in
scheduler
3GPP stack
➢ Core centralized and separated from RAN
eNB/gNB
chipsetCore NetworkInternet / OTT
➢ RAN functionality locked to a particular chipset
➢ Scheduler per eNB/gNB causing fierce competition for resources
➢ Single eNB/gNBcovering large area –low frequency reuse
➢ Low spectral efficiency [bit/s/Hz/m2]
➢ Interference between eNB/gNB
➢ High power eNB/gNB
14
Solution: Open Software-Defined RAN
• Low cost
• Low power (negigible EMF expsure)• Very short deployment time (days)
• High spectral efficiency [bit/s/Hz/m2] (capacity per m2)
• No coverage holes
• Easy indoor installation and great signal penetration
15
VNF Composition Framework
RAN controller
3GPP stack
Solution: zoom-in
Radio Heads
Edge ServerInternet / OTT
✓ RAN / Core / MANO running on off-the-shelf Edge server
✓ Hardware provider independent from functionality
✓ High level of portability
✓ Aggressive frequency reuse - very high spectra efficiency
✓ Interference solved by RAN controller
✓ High capacity, high number of users served
✓ Low power Radio Heads
16
VNF Composition Framework
RAN controller
3GPP stack
Solution: zoom-in
Radio Heads
Edge ServerInternet / OTT
✓ RAN / Core / MANO running on off-the-shelf Edge server
✓ Hardware provider independent from functionality
✓ High level of portability
✓ Aggressive frequency reuse - very high spectra efficiency
✓ Interference solved by RAN controller
✓ High capacity, high number of users served
✓ Low power Radio Heads
18
Open RAN (O-RAN)
● Founded as xRAN in 2016 by AT&T, Deutsche Telekom, SK Telecom and Stanford University
○ 2018 xRAN joined C-RAN Alliance + extending the scope to form O-RAN Alliance
● Scope:○ promote a software-based, extensible,
open Radio Access Network (xRAN)○ standardizing critical elements of the xRAN
architecture (interfaces).● Aims to
○ bring cloud scale economics to the RAN○ bring agility to the RAN
● Themes:○ Openness (new interfaces vs 3GPP, IMS-like
approach*)○ Intelligence (AI/ML, policies, 3rd party apps)
● Focus ○ on specifications and PoC
19
Telecom Infra Project (TIP)
Key members
● Founded 2016 by Facebook○ Deutsche Telekom, Facebook, Intel, Telefonica,
Nokia, Vodafone, BT in the Board of Directors○ 500+ members - many only observers
● Scope:○ building new technologies, examining new
business approaches, and spurring investment in the telecom space - from access to core
● Aims to○ VPs are often targeting rural deployments○ So far 2G and 4G based, kickstarting group for
5G NR○ Develop RAN technologies based on GPP and
disaggregated software (OpenRAN, OpenCellular)
● Focus ○ implementation and how to build SW and HW○ MVP
21
3GPP O-RAN TIP
Goal Allow large vendors to
interoperate and lower the
costs of network
development
Allow small vendors to interoperate and lower the costs even
further
Members Vendors mainly Operator driven: China Mobile,
AT&T + vendors (large and small)
Facebook + operators, main
vendors
3GPP compliance Yes Yes, complies and extends Yes
Outcomes Specifications Specifications and PoC MVPs (demonstrators)
Summary Good approach for 2G-3G Must have approach for 5G (and 4G)
Comparison: O-RAN, TIP and 3GPP
22
Network deployment for the coming networking revolution
Operator’s perspective
23
Cost of SD RAN vs. MACRO, DAS, SmallCell – simulation assumptions
Main assumptions of simulation:
To compare costs and implementation impact on similar coverage area for alternative radio solutions
● Deployment scenario: Indoor● Area: 4 km2, Dense Urban● Population: 10.000, Market share 50%, 500 MB/user@BH● One frequency band is used: 2.6 GHz● Bandwidth: 20 MHz, FDD● MIMO: 2x2● Sectorization: 3-sectors for Macro sites, Omni for remaining solutions● Building penetration loss: 20 dB for Macro sites, 3dB for remaining solutions
24
Improvement of capacity with lower investment
0,0
1,0
2,0
3,0
4,0
5,0
Network CAPACITY Gbps@BH
MACRO
DAS
SmallCells
SD-RAN
0
50
100
150
200
250
Number of sites/cells
MACRO
DAS
SmallCells
SD-RAN
Summary:
● Coverage: the same (4 km2)
● Capacity: 1,86x more capacity in SD-RAN comparing to MACRO
● TCO costs: alternatives 1,3-1,7x more expensive vs. SD-RAN
First live implementation shows 50%-70% lower costs vs. Macro (Ref: Latin America)
● Mbps cost efficiency: 2,7x lower costs of Mbps in SD-RAN vs. MACRO (TCO based)
Tota
l co
sts
[EU
R]
TCO
MACRO
DAS
SmallCells
SD-RAN
x1,4 x1,7 x1,3
29
Early adopters for the new paradigm
Rakuten and NEC will jointly develop a 3.7 GHz massive MIMO 5G antenna radio unit (RU), which will be manufactured by NEC based on architecture. It is planned to launch 5G services in June 2020.
This innovative network is fully virtualized from radio access network to core and adopts OPEN-RAN 5G systems architecture from launch.
Japanese operator NTT DoCoMo said that it is adopting open interface specifications from the O-RAN Alliance to allow it to mix and match its 4G and 5G networks.
The operator is set to launch its commercial 5G network in “pre-service” operation on 20 September.
30
Summary – a new approach
• To be able to purchase separately hardware and functionality (software)
• To foster innovation by building far broader ecosystem of players
• To significantly lower the cost of [GB] transfer
• To meet the 5G capacity targets
• To ease the network rollout (nobody thinks about Ethernet cable installation – it’s already there)
• To engage local players (municipalities, property owners, etc)
• To safeguard a future-proof computing infrastructure able to run telco functionalities as well as services
• To ultimately enable low latency applications
• To engage local players (municipalities, property owners, etc)
Open RAN architecture
Massive deployment of Radio Heads in existing infrastructure
Deployment of Edge computing resources
32
„A good hockey player plays where the puck is. A great hockey player plays where the puck is going to be.” Wayne Gretzky
„należy jechać tam, gdzie krążek zmierza, a nie tam, gdzie znajduje się w danym momencie”
Final word
33
Contact us
IS-WirelessPuławska 45b05-500 Piaseczno/near WarsawPOLAND
34
VNF
Composition
Framework
RAN
controller
3GPP stack
Products
Radio Heads
Edge ServerInternet / OTT
- Executable on any GPU / Server- Realizes slicing- Lego-brick flexible functionality
- 3GPP Standard compliant RATs(4G LTE, 5G NR,NB IoT)
- Support for WiFi
- Efficient use of radio resources- KPI optimization- Support for various QoS /traffic
- Core framework able to connect3rd party solutions
- MANO orchestrating underlyingVNFs
- Commercial off-the-shelf computing device
- Low power- Rapidly
deployable- Hidden- No permit
35
Strategy: universal software platform for numerous markets
By customizing our solution we can address other vertical markets
vRAN Hardware RAN
We start by addressing enhance mobile broadband
Massive MTC (IoT) Low latency Ultra-reliability