Software-defined Infrastructure for Advanced Wireless Testbeds - PowerPoint PPT Presentation

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

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

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 WINLAB

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) WINLAB

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

Capacity, Capacity, Capacity (cont’d) Academia Industry Spectral Efficiency/Air Massive MIMO: Coordinated Multipoint Tx/Rx • Interface Serve 10-20 users 3-D/Full-Dimensional MIMO • per sector with New Modulation and/or • 100-200 antennas Coding Schemes • per BS Cell Density/Networking Small Cells & Cell Densification • Heterogeneous WLAN Offloading • SoNs: From 300m Integrated MultiRAT Operation • to 90m cell radius Device-to-Device • on average Joint Scheduling, Nonorthogonal Multiple Access • Information and Communication Technology • Coupling Bandwidth/Spectrum mmWaves: From More Licensed and Unlicensed Spectrum, • 2-6 GHz to 20-60 mmWaves GHz Licensed Shared Access • Unlicensed Spectrum Sharing • WINLAB

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

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

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

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

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

Typical Fronthaul BW Requirements WINLAB

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

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 WINLAB Courtesy: Navid Nikaein, Eurecom/Open Air Interface