Fast Millimeter Wave Beam Alignment Omid Abari Haitham Hassanieh, - PowerPoint PPT Presentation

Fast Millimeter Wave Beam Alignment Omid Abari Haitham Hassanieh, Michael Rodriguez , Mohammed Abdelghany, Dina Katabi, and Piotr Indyk

Emerging Applications VR stream 7 Gbps of data from PC to headset One autonomous car will generate 4TB of data per day 5G promises 1000 times faster than 4G Today’s wireless networks can not support future applications

mmWave Technology

mmWave Technology Currently we operate here Millimeter Wave Bands 60GHz 70GHz 80GHZ 90GHz 100GHz 0 10GHz 20GHz 30GHz 40GHz 50GHz Frequency Spectrum Enable high-throughput wireless links

mmWave changes how wireless systems operate Today: Broadcast

mmWave changes how wireless systems operate mmWave: Pencil-beam Antennas Problem Statement: how can we align the beams to establish a link?

mmWave radios use phased arrays to create a beam

Naïve Approach: Exhaustive Scan 𝑂 : number of possible directions 𝑂 directions 𝑂 directions AP Client O 𝑂 - measurements à Too slow

802.11ad Scan Stage 1: Client uses omni-directional; AP scans directions AP Client

802.11ad Scan Stage 2: AP uses omni-directional; client scans directions AP Client O 𝑂 measurements à Still Too Slow [MOBICOM’14, SIGMETRICS’15, NSDI’16]

Can we find the best beam alignment without scanning the space?

Agile-Link • A millimeter wave system that can quickly establish a link without scanning the space • Works within the existing 802.11ad standard, and can support both clients and access points • Implemented and evaluated in practical settings

Agile-Link Idea Potential Direction of the Client: 0 o , 60 o , 90 o or 120 o AP 40 o , 60 o , 100 o or 150 o 60 o 60 o is direction of client Client Construct a Multi-Armed Beam: Simultaneously collects signals from multiple directions. 90 90 90 90 120 120 60 60 120 120 60 60 150 30 150 30 150 150 30 30 180 180 0 0 0 0 180 180

1. How do we create multi-armed beams? 2. What is the best choice of multi-armed beams to minimize the number of measurements?

Creating Multi-Armed Beams Phase Shifters For an Antenna Array: 𝑏 / 𝒊 = 𝑮 7 𝒚 , 𝑮′ is Inverse Fourier Matrix ℎ / 𝒛 𝒋 = 𝒃 𝒋 𝒊 = 𝒃 𝒋 𝑮 7 𝒚 ℎ - 𝑧 𝑗 𝒃 𝒋 90 𝑮 𝑏 0 ℎ 0 180 0 Divide 𝒃 𝒋 into segments 90 𝒃 𝒋 = 𝑮 180 0

1. How do we create multi-armed beams? 2. What is the best choice of multi-armed beams to minimize the number of measurements?

Why do we need to choose the beams carefully? 180 150 120 AP Client 90 60 30 0 Signals traveling along different paths can cancel each other

Best Choice of Multi-Armed Beams 180 150 Spatial Directions 0 o 30 o 120 60 o 90 o 120 o 150 o 180 o AP 90 60 o 60 30 0 Bins: 1 2 3 4 Client Hashing Pick multi-armed beams to create random hash functions • Voting Estimate the true direction using voting •

Theorem (Informal) - Assuming: - N possible directions - K signal paths - Our algorithm finds the optimal beam alignment in 𝑃 𝐿 log 𝑂 measurements.

Experimental Results

Implementation and Evaluation Built a Millimeter Wave Radio with a Phased Array.

Number of Measurements Reduction in the Number 10000 Our system vs. Exhaustive Agile-Link vs Exhaustive Search Reduction in Search Time Our system vs. 802.11ad of Measurements Agile-Link vs 802.11ad 1000 100 10 1 0 50 100 150 200 250 Number of Potential Directions ( N ) Phased Array Size Our system requires orders of magnitude fewer measurements

Beam Alignment Latency Number of 802.11ad 802.11ad Directions with our algorithm 16 1ms 0.5ms 64 4ms 0.8ms 128 106ms 0.9ms 256 310ms 1.0ms Achieved fast beam alignment (less than 1ms)

Beam Alignment Accuracy 1 0.8 0.6 CDF 0.4 Agile-Link 802.11ad 0.2 Exhaustive Search 0 0 1 2 3 4 5 6 7 8 SNR Loss (dB) Agile-Link performs much better than Exhaustive search and 802.11ad

Related Work • Past Work on Beam Alignment: [D. Araujo EUSIPCO’14, J. Kim Jour. Communications’14, B. Li Trans. Wireless’13, Y. M. Tsang Globecom’11, J. Wang Communications’09, W. Yuan PIMR’15, L. Zhou PIMRC’12, D. Ramasamy Allerton’12, A. Alkhateeb IEEESP’14, B. Gao. IET’14, B.Li Trans. Wireless’13, T. Nitsche ENET’15, T. Nitsche Infocom’15,etc] • Past Work on mmWave Channel Studies: [C. R. Anderson Trans. Wireless’04, S. Collonge Trans. Wireless’04, S. Rangan IEEE’14, M. Smulders Trans on Ant.’09, J. Violette NASA report, J. Comm. ’02, S. Sur SIGMETRIC’15, A. Saeed VTC’16, X. Tie PAM’11, etc] • Past Work on using Sparsity: [E. Eltayeb GLOBESIP’15, A. Alkhateeb J. IEEE’14, B. Gao’ IET14, D. Ramasamy, IEEE’14, etc]

Conclusion • Establishing communication links in millimeter wave networks is challenging due to directionality. • Agile-Link: millimeter wave system that can quickly establish a link without having to scan the space. • Exciting time for millimeter wave networks!

Fast Millimeter Wave Beam Alignment Omid Abari Haitham Hassanieh, Michael Rodriguez , Mohammed Abdelghany, Dina Katabi, and Piotr Indyk