1 / 39

Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)

Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [ mmWave Multi-Resolution Beamforming ] Date Submitted: [ January 15, 2008 ] Source: [ Ismail Lakkis 1 , Shuzo Kato 2 , Su- Khiong Yong 3 , Pengfei Xia 3 ]

ganya
Télécharger la présentation

Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) • Submission Title: [mmWave Multi-Resolution Beamforming] • Date Submitted: [ January 15, 2008] • Source:[Ismail Lakkis1, Shuzo Kato2 , Su-Khiong Yong3, Pengfei Xia3] • Company [ (1) Tensorcom, (2) NICT, (3) Samsung Electronics] • Address [] • Voice:[], FAX: [], E-Mail:[ilakkis@tensorcom.com, Shu.kato@nict.go.jp, sk.yong@samsung.com, pengfei.xia@samsung.com] • Re: [] • Abstract: [Multi-Resolution OFDM & SC Beamforming] • Purpose: [Multi-Resolution OFDM & SC Beamforming] • Notice: This document has been prepared to assist the IEEE P802.15. It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein. • Release: The contributor acknowledges and accepts that this contribution becomes the property of IEEE and may be made publicly available by P802.15. I. Lakkis & al.

  2. Content • Beamforming System Model • Potential Antenna Elements and Arrays • Multi-Resolution Beams • Multi-Level Beamforming Mechanism I. Lakkis & al.

  3. Beamforming System Model I. Lakkis & al.

  4. Beamforming System Model • A device using the same antennas for transmission and reception, i.e. same physical antennas, is referred to as a Antenna Symmetric System (ASS), otherwise it is referred to as Antenna Asymmetric System (AAS); • The ASS is a special case of AAS; • Consider a link between two AAS devices, DEV1 and DEV2, as shown in Fig. 1: • DEV1 has MT Transmit antennas and MR Receive antennas; • DEV2 has NT transmit antennas and NR receive antennas; • The developed model makes no assumptions regarding the antenna arrangements whether these are 1-D or 2-D antenna arrays, or any other arrangement; • A cyclic prefixed OFDM system with an FFT length of N subcarriers and a cyclic prefixed Single Carrier (SC) system with a burst length of N have the same system model. • We assume that the cyclic prefix is longer than any multipath delay spread between any pair of antenna elements; I. Lakkis & al.

  5. Beamforming System Model DEV1 DEV2 DEV1 Transmitter DEV2 Receiver DEV2 Transmitter DEV1 Receiver I. Lakkis & al.

  6. Beamforming System Model • Consider an OFDM symbol (SC burst) transmitted from DEV1 to DEV2: where Tc is the sample (chip) duration and skare the complex data ; • At DEV1’s transmitter, the bit stream is modulated by the beamformer vector: and then transmitted into a MIMO channel with a frequency domain Channel State Information (CSI) at frequency bin number n: where hi,j(n) denotes the cascade of the transmit/receive filtering, and the channel response between DEV1 j-th transmit antenna and DEV2 i-th receive antennas; I. Lakkis & al.

  7. Beamforming System Model • At DEV2’s receiver, the received signals are processed through the combiner vector: • The equivalent channel between DEV1’s transmitter and DEV2’s receiver is a Single Input Single Output (SISO) channel, with frequency response at frequency bins 0 to N-1 given by: • The discrete-frequency received signal model becomes: where is the OFDM data symbol (FFT of the SC data burst), and is the additive white Gaussian noise vector; • The equivalent model from DEV2’s transmitter to DEV1’s receiver have the same formulation but with a different SISO channel (see previous figure): I. Lakkis & al.

  8. Beamforming Cost Function • For both OFDM and SC, the Signal to Noise Ratio (SNR) on the nth subcarrier is given by: • Define the Effective SNR (ESNR) as a mapping from the instantaneous subcarriers SNRs to an equivalent SNR that takes into account the FEC (Forward Error Correction). There are many methods that can be used to compute the ESNR: • Mean of SNRs over the different subcarriers • QSM (Quasi-Static Method: 3GPP2 1xEV-DV/DO method 1) • CESM (Capacity effective SINR mapping: 3GGP2 1xEV-DV/DO method 2) • CECM (CESM based on Convex Metric: 3GGP2 1xEV-DV/DO method 3) • EESM (Exponential Effective SIR Mapping: 3GGP) • Etc,… • The ultimate objective of a beamforming algorithm is to select the optimal beamformer vectors w1(DEV1) and w2(DEV2) and optimal combiner vectors c2(DEV2) and c1(DEV1) that maximize the effective SNR (or any other optimality criterion); I. Lakkis & al.

  9. Beamforming Cost Function • Different ESNR methods can be used for SC and OFDM. A MMSE SC equalizer for example tends to have an ESNR which can be approximated by the average of the SNRs over the different subcarriers, whereas OFDM tends to have an ESNR which can be approximated by the geometric mean of the SNRs over the different subcarriers. More accurate methods that we listed before take into account the FEC, imperfections in the receiver, BER, … • For an ASS we have the following: • And consequently, it is enough to consider one direction of the link, i.e. DEV1-Tx to DEV2-Rx or DEV2-Tx to DEV1-Rx but not both; • For an AAS, both directions of the link should be considered to estimate the four vectors w1(DEV1), w2(DEV2), c1(DEV1), and c2(DEV2); • So for the general case of AAS, the following two steps are required: • 1. DEV1-Tx → DEV2-Rx, in order to estimate w1(DEV1) and c2(DEV2), DEV2 has to acquire the CSI matrices ; • 2. DEV2-Tx → DEV1-Rx, in order to estimate w2(DEV2) and c1(DEV1), DEV1 has to acquire the CSI matrices I. Lakkis & al.

  10. Potential Antenna Elements & Arrays I. Lakkis & al.

  11. Standing-Wave Dipole Antenna I. Lakkis & al.

  12. Broadside main lobe l / 2 Iin l Broadside main lobe Iin 3l / 2 Main lobes Iin I. Lakkis & al.

  13. Uniform Rectangular Aperture I. Lakkis & al.

  14. y x l/2 l/2 y x l l y x 2l 2l I. Lakkis & al.

  15. z d y x 1-D Antenna Array • The power gain of an antenna array = power gain of a single antenna × abs(array factor)2 • Array factor of a uniformly-spaced antenna array along the z-axis: • Antenna array directivity: • Maximum possible directivity: Dmax = N I. Lakkis & al.

  16. 2-D Antenna Array z • Array factor of a 2-D antenna array : dz dx • Separable 2-D antenna array with Nx elements along the x-direction and Ny elements along the y-direction: y x • Array factor of a separable 2-D antenna array : I. Lakkis & al.

  17. Antenna Array Codebook • In the following, we assume separable 2-D antenna arrays  to form the 2-D codebook we need only to specify codebooks for 1-D antenna arrays along the x-axis and y-axis ; • The simplest phased antenna array is an antenna array implementing the following phases 0o or 180o for each antenna element. We refer to this as scalar transmitter and the beamformer (combiner) weights are selected from {+1 , -1} • This means each antenna element will transmit (receive) I+Q (phase 0o), or -(I+Q) (phase 180o). We refer to such a system as a scalar-Tx (Rx) with weights +1 or -1; • A more flexible phased antenna array is an antenna array implementing the following phases 0o, 90o, 180o or 270o for each antenna element. • Using a quadrature transmitter with I & Q (In-phase and Quadrature), this means each antenna element will transmit (receive) I (phase 0o), -I (phase 180o), Q (phase 270o), –Q (phase 90o). An equivalent set of signals would be: I+Q, I-Q, -I+Q, -I-Q. We refer to such a system as vector-Tx (Rx) with weights selected from {+1, -1, +j, –j}; I. Lakkis & al.

  18. Beam Codebooks I. Lakkis & al.

  19. Beam & Sector codebooks • Beam codebooks for N antennas and M beams (M ≥ N) can be generated as follows: • Sector (multi-beam) codebooks for N antennas and M = N/2 beams can be generated as follows: • Fix can be substituted with round(). I. Lakkis & al.

  20. Multi-Resolution Beams • Ex: Linear array with 8 elements: • Level-1 (Coarse): Four sectors (each sector = 2 beams): • r = 1, Dthe_deg= 53.54, TheMax_deg = 13.78, GdB = 6.22 • r = 2, Dthe_deg= 20.35, TheMax_deg = 35.69, GdB = 6.36 • r = 3, Dthe_deg= 14.75, TheMax_deg = 58.07, GdB = 6.22 • r = 4, Dthe_deg= 13.50, TheMax_deg = 74.76, GdB = 5.37 • +1 +1 +1 +1 • -j -1 -1 -1 • +1 +1 -1 -1 • -j -1 +1 -1 • +1 -1 +1 -1 • -j +1 -1 +1 • +1 -1 -1 +1 • -j +1 +1 +1 Sector # 0 Sector # 3 I. Lakkis & al.

  21. Multi-Resolution Beams • Level 2 (fine): 8 beams • r = 1, Dthe_deg= 54.63, TheMax_deg = 0.00, GdB = 9.03 • r = 2, Dthe_deg= 20.13, TheMax_deg = 40.32, GdB = 8.38 • r = 3, Dthe_deg= 14.85, TheMax_deg = 60.01, GdB = 9.03 • r = 4, Dthe_deg= 13.19, TheMax_deg = 74.79, GdB = 8.38 • r = 5, Dthe_deg= 12.81, TheMax_deg = 89.98, GdB = 9.03 • r = 6, Dthe_deg= 13.13, TheMax_deg = 103.74, GdB = 8.38 • r = 7, Dthe_deg= 14.85, TheMax_deg = 119.99, GdB = 9.03 • r = 8, Dthe_deg= 19.32, TheMax_deg = 137.52, GdB = 8.38 • +1 +1 +1 +1 +1 +1 +1 +1 • -1 -1 -j -j +1 +1 +j +j • +1 +j -1 -j +1 +j -1 -j • -1 -j +j -1 +1 +j -j +1 • +1 -1 +1 -1 +1 -1 +1 -1 • -1 +1 -j +j +1 -1 +j -j • +1 -j -1 +j +1 -j -1 +j • -1 +j +j +1 +1 -j -j -1 • condition number = 1.000, GSUM = 69.63710 I. Lakkis & al.

  22. Multi-Resolution Beams • Level-3 (tracking = super-fine): 32 beams • r = 1, Dthe_deg= 54.63, TheMax_deg = 0.00, GdB = 9.03 • r = 2, Dthe_deg= 54.63, TheMax_deg = 0.00, GdB = 9.03 • r = 3, Dthe_deg= 46.04, TheMax_deg = 25.16, GdB = 8.39 • r = 4, Dthe_deg= 24.85, TheMax_deg = 32.91, GdB = 8.30 • r = 5, Dthe_deg= 20.13, TheMax_deg = 40.32, GdB = 8.38 • r = 6, Dthe_deg= 18.13, TheMax_deg = 45.38, GdB = 8.18 • r = 7, Dthe_deg= 16.60, TheMax_deg = 50.88, GdB = 8.18 • r = 8, Dthe_deg= 16.19, TheMax_deg = 57.42, GdB = 8.25 • r = 9, Dthe_deg= 14.85, TheMax_deg = 60.01, GdB = 9.03 • r = 10, Dthe_deg= 14.85, TheMax_deg = 60.01, GdB = 9.03 • r = 11, Dthe_deg= 13.69, TheMax_deg = 66.11, GdB = 8.39 • r = 12, Dthe_deg= 13.50, TheMax_deg = 70.17, GdB = 8.30 • r = 13, Dthe_deg= 13.19, TheMax_deg = 74.79, GdB = 8.38 • r = 14, Dthe_deg= 13.03, TheMax_deg = 78.33, GdB = 8.18 • r = 15, Dthe_deg= 12.91, TheMax_deg = 82.48, GdB = 8.18 • r = 16, Dthe_deg= 13.63, TheMax_deg = 87.80, GdB = 8.25 • r = 17, Dthe_deg= 12.81, TheMax_deg = 89.98, GdB = 9.03 • r = 18, Dthe_deg= 12.81, TheMax_deg = 89.98, GdB = 9.03 • r = 19, Dthe_deg= 12.56, TheMax_deg = 95.45, GdB = 8.39 • r = 20, Dthe_deg= 12.85, TheMax_deg = 99.24, GdB = 8.30 • r = 21, Dthe_deg= 13.13, TheMax_deg = 103.74, GdB = 8.38 • r = 22, Dthe_deg= 13.38, TheMax_deg = 107.30, GdB = 8.18 • r = 23, Dthe_deg= 13.81, TheMax_deg = 111.64, GdB = 8.18 • r = 24, Dthe_deg= 15.47, TheMax_deg = 117.49, GdB = 8.25 • r = 25, Dthe_deg= 14.85, TheMax_deg = 119.99, GdB = 9.03 • r = 26, Dthe_deg= 14.85, TheMax_deg = 119.99, GdB = 9.03 • r = 27, Dthe_deg= 15.69, TheMax_deg = 126.52, GdB = 8.39 • r = 28, Dthe_deg= 17.10, TheMax_deg = 131.34, GdB = 8.30 • r = 29, Dthe_deg= 19.32, TheMax_deg = 137.52, GdB = 8.38 • r = 30, Dthe_deg= 22.10, TheMax_deg = 142.90, GdB = 8.18 • r = 31, Dthe_deg= 29.60, TheMax_deg = 150.34, GdB = 8.18 • r = 32, Dthe_deg= 55.85, TheMax_deg = 164.09, GdB = 8.25 • +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 • -1 -1 -1 -1 -1 -1 -1 -1 -j -j -j -j -j -j -j -j +1 +1 +1 +1 +1 +1 +1 +1 +j +j +j +j +j +j +j +j • +1 +1 +1 +1 +j +j +j +j -1 -1 -1 -1 -j -j -j -j +1 +1 +1 +1 +j +j +j +j -1 -1 -1 -1 -j -j -j -j • -1 -1 -1 -j -j -j +1 +1 +j +j +j -1 -1 -1 -j -j +1 +1 +1 +j +j +j -1 -1 -j -j -j +1 +1 +1 +j +j • +1 +1 +j +j -1 -1 -j -j +1 +1 +j +j -1 -1 -j -j +1 +1 +j +j -1 -1 -j -j +1 +1 +j +j -1 -1 -j -j • -1 -1 -j -j +1 +j +j -1 -j -j +1 +1 +j -1 -1 -j +1 +1 +j +j -1 -j -j +1 +j +j -1 -1 -j +1 +1 +j • +1 +1 +j -1 -j -j +1 +j -1 -1 -j +1 +j +j -1 -j +1 +1 +j -1 -j -j +1 +j -1 -1 -j +1 +j +j -1 -j • -1 -1 -j +1 +j -1 -j +1 +j +j -1 -j +1 +j -1 -j +1 +1 +j -1 -j +1 +j -1 -j -j +1 +j -1 -j +1 +j I. Lakkis & al.

  23. Planar Array • Planar arrays can be obtained from 1-D linear arrays by specifying the codebooks along the z-axis and x-axis . • Ex: Planar array 8x8 showing all 64 beams I. Lakkis & al.

  24. Planar Array • Contour of planar array 8x8 showing all 64 beams theta I. Lakkis & al.

  25. Multi-Level Beamforming Mechanism I. Lakkis & al.

  26. Multi-Level Beamforming Mechanism • Association • Exchange beamforming capabilities • Antenna training: Two level training • Coarse Level/Sector based • Fine Level/High resolution beams based • Tracking Level: • Super-high resolution beams based • Coupled-clustering approach for tracking of channel changes I. Lakkis & al.

  27. Association Information Exchange • During association, DEV1 & DEV2 exchange the following information: • # z-antennas, # x-antennas • # sectors • Fine codebook ID (or # fine beams) • Super fine codebook ID (or # super-fine beams) • Cluster size (# super fine beams per cluster) I. Lakkis & al.

  28. Coarse Level Beamforming • DEV1 (source) uses N sectors & DEV2 (dest.) uses M sectors • Coarse training consists of M cycles; • Cycle n: DEV1 sends M repetitions of a sector training sequence (ST) in the same direction (i.e. sector #n) ; • Cycle n: DEV2 attempts to receive each ST sequence using different direction. DEV2 uses sector#0 to listen to TS#0, sector#1 to listen to TS#1, and so on; • After the full N cycles, DEV2 will have an opportunity to receive each combination of DEV1 transmit sectors (0:N-1) and DEV2 receive sectors (0:M-1); • DEV2 selects the best pair of sectors, i.e. DEV1’s optimal Tx sector and DEV2 optimal Rx sector; • DEV2 feedback using its best sector and repeats it N times • DEV1 scans the feedback using its 0:N-1 sectors and decode the feedback; • DEV1 inform DEV2 about the beams it is planning to use next; • Repeat from DEV2 to DEV1 for asymmetric channel; I. Lakkis & al.

  29. Coarse Level I. Lakkis & al.

  30. Coarse Level I. Lakkis & al.

  31. Fine Level • During coarse level, DEV1 has sliced its best sector(s) into K beams and DEV2 has sliced its best sector(s) into J beams & the 2 DEVs have exchange this information • Fine training consists of K cycles; • Cycle j: DEV1 sends J repetitions of a beam training sequence (BT) in the same direction (i.e. beam #j) ; • Cycle j: DEV2 attempts to receive each ST sequence using different beam direction. DEV2 uses beam#0 to listen to BT#0, beam#1 to listen to BT#1, and so on; • After the full K cycles, DEV2 informs DEV1 about the optimal beamformer pattern (or best beam) using its best sector identified in Level1. DEV1 listens using its best sector as well. • DEV1 & DEV2 switch to their best patterns (beams) and test the link (verification); • Repeat from DEV2 to DEV1 for asymmetric channel; I. Lakkis & al.

  32. Fine Level Verification Stage: Is this needed? I. Lakkis & al.

  33. Fine Level I. Lakkis & al.

  34. Tracking • Cluster: a set of adjacent beams • Associate a cluster from DEV1 to a cluster of DEV2 • Tracking will happen at cluster level • Each cluster will have a center beam • Track the center beam(s) in each cluster • Example: • DEV1 Cluster = beams of index 6,7,8 (identified during fine phase) • After some time, beam number 8 became stronger than beam 7; • DEV1 redefine the cluster as 7,8,9 • It is like tracking the center of gravity cluster by cluster • Use a tracking packet concept for tracking (bidirectional for asymmetric channel); I. Lakkis & al.

  35. Tracking Sector Super-fine Beams +p Cluster 1 Cluster 2 -p 0 p I. Lakkis & al.

  36. Tracking Cluster 1 Cluster 2 t = 0 Center of cluster 1 New Cluster 1 Cluster 2 t = T Added to cluster 1 New center of cluster 1 Dropped from cluster 1 I. Lakkis & al.

  37. Pro-Active Beamforming • Coarse training sequences takes place in the preamble and can be used by all DEVs; • Each DEV reserves a CTA for the special purpose of fine beamforming; • Coarse Feedback & Fine training takes place between PNC and a single DEV; • Tracking takes place in the data CTA allocated to PNC & DEV I. Lakkis & al.

  38. Pro-Active Beamforming I. Lakkis & al.

  39. On-Demand Beamforming • It takes place in a CTA between two devices: DEV1 (source) & DEV2(destination); • DEV1 reserves a CTA for the special purpose of beamforming • Two-level training (Coarse & Fine) as described before • Tracking takes place in the data CTA I. Lakkis & al.

More Related