December 2 nd, 2016 Ivan Seskar WINLAB

Size: px

Start display at page:

Download "December 2 nd, 2016 Ivan Seskar WINLAB"

Sydney Perkins
6 years ago
Views:

1 FALL 2016 RESEARCH REVIEW Software-defined Infrastructure for Advanced Wireless Testbeds December 2 nd, 2016 Ivan Seskar Department of ECE Rutgers, The State University of New Jersey seskar (at) winlab (dot) Rutgers (dot) edu

Technical Challenges Faster Cellular Radios Access ~1-10 Gbps ~1000x capacity

Dynamic Spectrum Access Next-Gen Mobile Network Wideband PHY Cloud RAN arch

) Custom PHY for IoT New MAC protocols RAN redesign Light-weight control

60 Ghz & other new bands New unlicensed/shared spectrum Dynamic spectrum

2 Technical Challenges Faster Cellular Radios Access ~1-10 Gbps ~1000x capacity Low- Latency/ Low-Power Access Network For Real- Time IoT New Spectrum & Dynamic Spectrum Access Next-Gen Mobile Network Wideband PHY Cloud RAN arch Massive MIMO mmwave (60 Ghz) Multi-Radio access HetNet (+WiFi, etc.) Custom PHY for IoT New MAC protocols RAN redesign Light-weight control Control/data separation Network protocol redesign. 60 Ghz & other new bands New unlicensed/shared spectrum Dynamic spectrum access Spectrum sharing techniques Non-contiguous spectrum Network/DB coordination methods. Mobile network redesign Convergence with Internet Clean-slate Mobile Internet Software Defined Networks Open wireless network APIs Cloud services & computing Edge cloud/fog computing Virtualization, NFV.

3 Wireless Network Softwarization Use software programming to design, implement, deploy, manage and maintain wireless network equipment, components and services Goals: Increase re-usability Rapid re-design of network and service architectures Optimize processes in networks Reduce costs Bring added value to infrastructures

4 Softwarization in Wireless Networks In radio access networks - Agility in spatial, temporal and frequency dimensions enabling: Fine-grained physical layer/network programmability Flexibility in spectrum management Dynamic provisioning Heterogeneous deployments. In mobile edge networks - Extend softwarization from the conventional data center to the edge of wireless networks: Enable on demand service deployment at the most effective locations based on application requirements Automate service establishment/maintenance mechanisms (in a timely fashion)

5 Capacity, Capacity, Capacity 1000 = bits/sec km 2 = Research Viewpoint bits/sec/hz cell Hz cell km 2 4G ~1000x ~20x Industry Viewpoint 5G ~1000x ~3-5x ~5-10x Air Interface Spectrum Spectral efficiency ~25x ~1-2x Networking Bandwidth Cell density Giuseppe Caire: Massive MIMO: implementation issues and impact on network optimization 2016 Tyrrhenian International Workshop on Digital Communications (TIW16) ~1-2x Qian (Clara) Li,Huaning Niu, Apostolos (Tolis) Papathanassiou, and Geng Wu: 5G Network Capacity IEEE vehicular technology magazine, March 2014

6 Capacity, Capacity, Capacity (cont d) Spectral Efficiency/Air Interface Cell Density/Networking Bandwidth/Spectrum Academia Massive MIMO: Serve users per sector with antennas per BS Small Cells & Heterogeneous SoNs: From 300m to 90m cell radius on average mmwaves: From 2-6 GHz to GHz Industry Coordinated Multipoint Tx/Rx 3-D/Full-Dimensional MIMO New Modulation and/or Coding Schemes Cell Densification WLAN Offloading Integrated MultiRAT Operation Device-to-Device Joint Scheduling, Nonorthogonal Multiple Access Information and Communication Technology Coupling More Licensed and Unlicensed Spectrum, mmwaves Licensed Shared Access Unlicensed Spectrum Sharing

7 5G Spectrum Coverage Mérouane Debbah, 5G: Can we make it by 2020? 2016 Tyrrhenian International Workshop on Digital Communications (TIW16)

Basestation Architecture Evolution Traditional Design Current Design Cloud Radio Access Network (CRAN)

Core Network Core Network Remote Radio Head (RRH) Baseband Unit (BBU) FRONTHAUL Common Public Radio

(ETSI-ORI) BACKHAUL S11,R4,R6 Power Amplifi er Power Amplifi er Baseband Power Amplifi er Baseband

8 Basestation Architecture Evolution Traditional Design Current Design Cloud Radio Access Network (CRAN) Power Amplifier Baseband Transport Control & Mgmt. Core Network Power Amplifier Baseband Transport Control & Mgmt. Core Network Remote Radio Head (RRH) Baseband Unit (BBU) FRONTHAUL Common Public Radio Interface (CPRI) Open Base Station Architecture Initiative (OBSAI) Open Radio Equipment Interface (ETSI-ORI) BACKHAUL S11,R4,R6 Power Amplifi er Power Amplifi er Baseband Power Amplifi er Baseband Transport Baseband Control Transport & Mgmt. Control Transport & Mgmt. Control & Mgmt. Core Network Core Network

Paging in 5G RAN Protocol Stack Considerations Functionality on a Faster Time Scale: Agile Traffic

9 METIS-II Key 5G Architecture Paradigms 5G RAN a Harmonized and Integrated Landscape of AIVs A Logical CN/RAN Split with evolved Interfaces Moving Functionality from Core Network to RAN: Mobility and Paging in 5G RAN Protocol Stack Considerations Functionality on a Faster Time Scale: Agile Traffic Steering and Resource Mgmt RAN Enablers for Network Slicing Physical Architecture and Possible Function Splits

10 Service Service Service Service Service Service Service Service METIS-II: RAN Support for Network Slicing It is foreseen that network slices will be used to form logical E2E networks for particular business constellations The 5G RAN should be slice-aware Offer means for slice isolation and protection Provide means for efficient resource reuse Key questions are yet the assignment of devices to slices and multi-slice connectivity. Example network slice (E2E logical network) Completely independent realization of network slices in the core network Likely individual logical protocol instances for different services, highly tailored to these. Possibly slice-specific processing of services Likely multiple slices and the services therein multiplexed into common instances for lower MAC, PHY, and sharing the same radio. Note that MAC or PHY functions may still be highly slice- or service tailored MUX RRC PDCP RLC MUX MUX RRC PDCP RLC Possibly slice-specific MAC scheduler MUX MUX RRC PDCP RLC MUX MUX RRC PDCP RLC Possibly slice-specific MAC scheduler Possibly common lower MAC (but with slice-specific and/or service-specific behaviour) Possibly common PHY (but with slice-specific and/or service-specific behaviour) Possibly common radio / spectrum CN Domain RAN Domain

11 PDCP RLC (asynch.) RLC (synch.) MAC (e.g. RRM) METIS-II: Function Splits MAC (e.g. HARQ) FEC Scrambling Modulation, Layer mapping, Precoding Resource element mapping & IFFT D/A Conversion Antenna S1* Resource element mapping & IFFT D/A Conversion Antenna Resource element mapping & IFFT D/A Conversion Antenna M6-M4 Uncoded user data (without H- ARQ retransmissions) M3 Coded user data M2 I/Q samples in frequency domain M1 I/Q samples in time domain (e.g. CPRI) M0 Coaxial cable Scaling with user data rates Scaling with bandwidth and # of antennas Relaxed latency requirements Requiring low-latency fronthaul

12 Typical Fronthaul BW Requirements

13 METIS-II: Deployment Scenarios Scenario 1 Standalone access nodes Each node with one or more (colocated) air interfaces Non-ideal backhaul* Site A: BB- + Central Cloud / Aggregation point Non-ideal backhaul* (optional) Non-ideal backhaul* Site B: BB- + Scenario 2 Central baseband processing unit for high number of access nodes Ideal fronthaul Site A: Optional BB- + Central Cloud / Central BB- Ideal fronthaul Site B: Optional BB- + Scenario 3 Local baseband processing unit for low to medium number of access nodes Local BB- Ideal fronthaul Site A: Optional BB- + Non-ideal backhaul* Local BB- Central Cloud / Aggregation point Non-ideal backhaul* Ideal fronthaul Site B: Optional BB- + Scenario 4 Self-back/fronthauling scenario Non-ideal backhaul (Ideal fronthaul) Site B taken (e.g. novel 5G Wireless selfback/fronthaul Site C: (Optional) BB- + Site B: (Optional) BB- + from Scenario 1-3 (optional e.g. LTE-A Wireless selfback/fronthau l Site D: (Optional) BB- + (e.g. LTE-A (optional e.g. novel 5G (e.g. novel 5G (optional e.g. LTE-A (e.g. LTE-A (optional e.g. novel 5G (e.g. novel 5G (optional e.g. LTE-A (e.g. LTE-A (optional e.g. novel 5G (e.g. novel 5G (optional e.g. LTE-A (e.g. LTE-A (optional e.g. novel 5G (e.g. novel 5G (optional e.g. LTE-A

14 Example: OpenAirInterface enodeb and UE Challenge : Efficient LTE implementation that uses general-purpose x86 processors (GPP) for base-band processing front-end, channel decoding, phy procedures, L2 protocols Key elements: Real-time extensions to Linux OS x86-64 multicore arch Real-time data acquisition to PC SIMD optimized integer DSP 64-bit MMX 128-bit SSE2/3/4 256-bit AVX2 ifft/fft, Channel Estimation, Turbo Decoding SMP Parallelism Master-worker model Courtesy: Navid Nikaein, Eurecom/Open Air Interface

OAI Roadmap: Toward Software-defined 5G Network Cloud-native 5G networks Phase 1:

Architecture and NFV Supported projects: FP7 MCN, FUI ELASTIC Network Orchestration

service-oriented deployment (https://jujucharms.

Network Programmability network slicing Agent-controller protocol and southband API

programmability Network controller: network abstraction (network state graphs),

15 OAI Roadmap: Toward Software-defined 5G Network Cloud-native 5G networks Phase 1: Stateless through distributed shared memory, multitenancy Phase 2: Mircoservice Architecture and NFV Supported projects: FP7 MCN, FUI ELASTIC Network Orchestration Approach 1) Openstack and heatstack orchestrator Approach 2) Juju modeling for service-oriented deployment ( Supported project: FP7 MCN, FP7 FLEX, Canonical partnership program Network Programmability network slicing Agent-controller protocol and southband API in support of SDN+MEC agents: in charge of network function monitoring and programmability Network controller: network abstraction (network state graphs), network application realtime, standalone mode or as a plugin Supported projects: H2020 Coherent, H2020 Q4Health, ETSI MEC PoC Courtesy: Raymond Knopp, Euricomm

Fronthaul At Scale: ORBIT Testbed Massive-MIMO 40 USRP X310s Available FPGA resources: Resource Type DSP48 Blocks Block Rams (18 kb) Number 58K 14K Logic Cells 7.2M Slices (LUTs) 1.

16 Fronthaul At Scale: ORBIT Testbed Massive-MIMO 40 USRP X310s Available FPGA resources: Resource Type DSP48 Blocks Block Rams (18 kb) Number 58K 14K Logic Cells 7.2M Slices (LUTs) 1.5M 2 x UBX-160 (10 MHz - 6 GHz, 160 MHz BB BW) 2 x 10G Ethernet for fronthaul/interconnect Four corner movable mini-racks (4 x 20 x 20 -> 1 x 80 x 80) > 500+ GPP Cores/CloudLab Rack Number of GPU platforms 32x40G SDN aggregation switch

17 Fronthaul (M1) Requirements for Massive MIMO Single-Ethernet/ Single-host Limit Current BW Requirement???

18 City Scale: CRAN Expanded Massive CPU,GPU,FPGA Cloud

19 Program Experimental License (cont d) Rules have officially become effective as of January 14, But: new forms and reporting WEB site are not yet up (will take a few more weeks according to reliable sources ) Perfect fit: intended to foster innovation Has significant implication on wide area experimentation especially with SDRs and nontraditional front-ends. (the rule also includes two other experimental licenses: the Medical Testing License and the Compliance Testing License)

20 Missing Link : Outdoor Deployable SDR Wireless Units? Modest power amplifier Wideband Antenna Firewall (GENI) Rack SDR front-end Local processing (Open Programmable) COTS BS/AP

21 More wiser.orbit-lab.org wimax.orbit-lab.org metis-ii.5g-ppp.eu

ORBIT Pilot September 24 th, 2017 Ivan Seskar WINLAB

IEEE 5G and Beyond Testbed Workshop ORBIT Pilot September 24 th, 2017 Ivan Seskar Department of ECE Rutgers, The State University of New Jersey seskar (at) winlab (dot) Rutgers (dot) edu IEEE 5G Testbeds