Ali Sadri, PhD Sr. Director Intel Corporation Past and Future - PowerPoint PPT Presentation

Ali Sadri, PhD Sr. Director Intel Corporation

Past and Future Capacity Improvement Air Interference Mitigation, Full Duplex Air New Waveform, MU-MIMO Interface Beamforming, etc Interface Available Available Licensed, Unlicensed, Shared Spectrum Spectrum mmWave Small Cell 3G-4G Densification Relay Edge Cloud Small Cell Mesh Backhaul Fronthaul 4G-5G 2 iCDG - Intel Communication and Devices Group

Search for Alternate Spectrum Current IMT 24 GHz Band LMDS Band 40 GHz Band 50 GHz Band 60 GHz Band 70-80 GHz Bands bands Licensed Licensed Licensed Licensed Unlicensed Minimal Licensed 5+5 GHz <1 GHz <4 GHz <4 GHz <3 GHz 7 GHz Global MS Global MS Global MS Global MS Global MS No Mobile Allocation In Region 1 & 2 30 1 3 10 40 20 50 70-80 24.25 25.25 38.6 42.5 47.2 50.2 27 31 57 64 3 iCDG - Intel Communication and Devices Group

Reuse mmWave Knowledge Bands > 6 New Bands Legacy Bands GHz < 3.8 GHz < 6 GHz (+mmWAVE) Licensed Unlicensed Licensed Unlicensed Licensed Unlicensed 28 39 45 60 6-24 70-90 WiFi WiFi LAA GHz GHz GHz GHz GHz GHz * Categorized based on channel models and path loss ** Potentially the same technology elements could be used across a range of frequencies 4 iCDG - Intel Communication and Devices Group

mmWave Path-Loss Comparisons Oxygen Absorbance 2.3 GHz 28 GHz 38 GHz 73 GHz 5 iCDG - Intel Communication and Devices Group

HetNet with mmWave Capable Small Cells (MCSC) eNB Aggregation MCSC eNB eNB MCSC MCSC MCSC MCSC MCSC 28 ,39 or 60 GHz 6 iCDG - Intel Communication and Devices Group

Network Densification Topology Fiber Node Distribution Node Access Node 7 iCDG - Intel Communication and Devices Group

High Frequency Beam Forming 8 iCDG - Intel Communication and Devices Group

Challenges in mmWave Systems Design • Higher Path Loss • To compensate with the high path loss higher gain antenna and/or higher transmit power is required • EIRP, TX power and RF exposure limit are regulated • Massive MIMO is required for high gain antennas • Transmission becomes highly directional • With Narrow beams, tracking of the UE becomes challenging • Feed line loss • Diminishing return occurs as size of array increases • Transmission loss increases as function of frequency 9 iCDG - Intel Communication and Devices Group

Challenges in RF & Antenna 28 or 39 GHz • Feed line loss: (8-by-8) elements 60 GHz λ 𝑑/𝑔 7.69𝑛𝑛 λ 𝑑/𝑔 5𝑛𝑛 2 = 2 = Antenna spacing: 2 = Antenna spacing: 2 = = 2.5mm 2 2 @ 28 GHz is 5.36mm and @ 39 GHz is 3.85mm From 60 GHz to 28 GHz (or 38 GHz), • The required area getting bigger then feed line getting longer (roughly double). • Feed loss is also a function of frequency (higher loss at 60 GHz) 10 iCDG - Intel Communication and Devices Group

Modular RFEM Configurations 60GHz Operation 16 Elements 25.2 mm x 9.8 mm Antenna Side Shield Side 16 elements 32 elements 64 elements 128 elements 128 elements iCDG - Intel Communication and Devices Group

MAA POC Evolution • Stack up PCB MAA 128 (2x4) • • Maple-M & R EIRP ~ 43 dBm • Reduce Side lobe • Stack up PCB • MAA 128 (1x8) • Maple-M & R • Partial PCB • Discrete • G3M EIRP ~ 43 dBm • MAA 128 (2x4) MAA 128 (1x8) • • • 160 x 140 x 110 Indoor • Maple-M & R • Maple-M & R EIRP ~ 43 dBm EIRP ~ 43 dBm • • • Stack up PCB 300 x 220 x 150 160 x 140 x 110 • • MAA 128 (2x4) • Maple-M • • MAA-RFEM EIRP ~ 48 dBm • 160 x 140 x 110 • G3S Indoor G3 Indoor 190 x 170 x 140 • G1 G2 Indoor Indoor G4 Outdoor 12 iCDG - Intel Communication and Devices Group

GEN3+ Evaluation Kit Hardware Overview Indoor Design • Easy access to ports • Easy signal breakout for chamber tests • Easy tabletop, tripod, post, ceiling installation • Antenna Array • 128 elements - 8x16 array - balanced feeds • Tiled 8x RFEMs based on Intel WiGig product • 1x Intel WiGig Baseband Modem Module • 13 Intel Confidential iCDG - Intel Communication and Devices Group

Link Budget Calculation Calculate SNR values and find supportable MCS in AWGN channels ITU Region N (1 Gbps threshold) LOS Backhaul Access No rain 650 m 380 m 99.00% availability 600 m 360 m 99.90% availability 470 m 290 m 99.99% availability 350 m 230 m Assumptions • Noise figure + implementation loss: (10.5 + 3) dB • PER < 1% • AWGN channel (phase impairment considered) iCDG - Intel Communication and Devices Group

Antenna Field Regions D iCDG - Intel Communication and Devices Group

Anant Gupta, UCSB Under the direction of: Professor Madhow of UCSB and Professor Amin of Standard Oct 31, 2016

Sparse Array of Sub-Arrays Goal: Sparse array of subarrays for directive & steerable beams with: Sparse placement of subarrays (4x4 element arrays). • Optimal phase shifts for beamsteering. • conventional array Attribute: Sparse array Larger aperture  Directivity ↑ and BW ↓ • Sparse arrays with same/fewer elements • Challenge: Sub-Nyquist generates aliasing and grating lobes • Problem different from traditional 2D placement (subarray • elements are fixed) Approach: Non uniform configurations perform better in all metrics Intel MAA-RFEM Optimal placement of sub-arrays and phase processing • 4x4 Module Algorithmic/application-level resiliency to aliasing (e.g. for • imaging) iCDG - Intel Communication and Devices Group

Early Insights Trade-offs in different architectures: Metrics: G final Horizontal ? =0 ° 0 SEP2 MRA BW, Grating/side lobes, Directivity -5 Uniform Benchmark Normalized Gain -10 -15 -20 -25 -30 -60 -40 -20 0 20 40 60 3 ° Directivity v/s Subarray Seperation 25 plus Square Directivity saturates beyond certain 24 23 aperture size Sparse Non-uniform G D (dB) 22 21 Benchmark 20 19 18 0 2 4 6 8 10 Subarray seperation ( 6 ) iCDG - Intel Communication and Devices Group

Major Metrics & Approach Cost functions MSLL: Maximum Side lobe level(relative to main lobe) • Directivity Gain- • 2D Beamwidth: (3 dB beam) Max * (3 dB beam) Min • ASLL (Average Side Lobe Level) • Sub-Array Placement: Greedy search Sequentially search for subarray positions on all possible • locations of grid (dx=0.5 λ , dy=0.6 λ ). Sequential Steering weight Optimization Steering weight optimization: Sequential Optimization -8.4 Round 1 Scan for best steering weight across all elements to • Round 2 Round 3 -8.6 reduce MSLL. Round 4 MSLL -8.8 -9 -9.2 0 20 40 60 80 100 120 140 Elements iCDG - Intel Communication and Devices Group

Tradeoffs in Performance Plus Circle 10 10 5 5 y - 6 units y - 6 units Observations and tradeoffs 0 0 Tradeoff between beamwidth and sidelobe level as -5 -5 aperture size increases. -10 -10 -10 0 10 -10 0 10 x - 6 units x - 6 units Pseudo Linear Benchmark Beamwidth ∝ ( Aperture area) -1 10 10 5 5 y - 6 units y - 6 units 0 0 -5 -5 -10 -10 -10 0 10 -10 0 10 Directivity Gain G D MSLL(rel. to mainlobe) x - 6 units x - 6 units -5 28 -10 26 dB dBi -15 24 -20 22 crc + lin B crc + lin B Configs Configs Beamwidth(deg 2 ) ASLL(rel. to mainlobe) 10 3 -15 deg 2 dB 10 2 -20 -25 10 1 crc + lin B crc + lin B Configs Configs Naive Seq-phase-Optimized Ideal iCDG - Intel Communication and Devices Group

Early Results; trade-offs in beam steering Plus Circle 10 10 Phase optimization to ↓ MSLL causes ↓ Directivity. 5 5 y - 6 units y - 6 units 0 0 Directivity v/s steering -5 -5 Plus Circle 28 28 -10 -10 -10 0 10 -10 0 10 x - 6 units x - 6 units 26 26 Pseudo Linear Benchmark G D (dBi) G D (dBi) 10 10 24 24 5 5 y - 6 units y - 6 units 0 0 22 22 -5 -5 0 20 40 60 0 20 40 60 -10 -10 Steering Angle(El.) Steering Angle(El.) Theory -10 0 10 -10 0 10 Ideal x - 6 units x - 6 units Rounding Linear Benchmark 28 27 Phase-opt Observations and tradeoffs 27 26 G D (dBi) G D (dBi) 26 Tradeoff between Directivity gain & sidelobe level 25 with phase optimization 25 24 24 23 23 0 20 40 60 0 20 40 60 Steering Angle(El.) Steering Angle(El.) Beamwidth ∝ ( Aperture area) -1 iCDG - Intel Communication and Devices Group

Beamwidth and Aperture Beam width is roughly inverse of physical array aperture width iCDG - Intel Communication and Devices Group

Conclusion Substantial effort has been focused in the industry on the 5G access • technology to improve capacity, latency, throughput, scalability and quality of service; Access technology alone cannot significantly improve network capacity; • An end-to-end 5G system need be augmented by flexible and high • throughput backhaul and fronthaul; mmWave technology is a great candidate for both access and backhauling to • increase network throughput and capacity, and lower interference; Sparse array architecture provides additional feature to optimize array • performance 23 iCDG - Intel Communication and Devices Group