An Advanced Wireless System with MIMO Spatial Scheduling

Transcription

1 An Advanced Wireless System with MIMO Spatial Scheduling Jan., 00 What is the key actor or G mobile? ) Coverage High requency band has small diraction & large propagation loss ) s transmit power Higher data rate requires larger Tx power, but battery capacity is limited ) Reduction o control signaling Current scheduling or MIMO system requires large amount o signaling ) High spectral eiciency Spectrum is always scarce. Yoshitaka Hara Mitsubishi Electric Corporation Challenges towards a new system solution : Multiband System + OFDMA/TDD with MIMO Spatial Scheduling Multiband System : high requency (small diraction) 0 : low requency (large diraction) Solution to coverage problem Cell (Hierarchical Sectors) Low req. Multiband System Coniguration Paging channel in low requency Fast band allocation in MAC layer depending on band quality Gradual migration to G system, keeping low investment risk 0 0 Frequency (.GHz) 0 (900M, GHz) G system Cooperation G system Selected requency High requency covers most o area, whereas low requency supports small area not covered by high requency Multiband system supports both high data rate & wide coverage Supported by Low Freq. Supported by High Freq. High req..ghz Stand-by terminal GHz, or 900MHz High Data Rate Channel Beacon, Paging, Basic control ino. Data Transmission Band quality reporting ( --> ) Allocation o high requency band New call arrival Low Data Rate Channel

2 Harmonization o Duplexing Methods Harmonization o FDD & TDD would be G issue Duplexing FDD TDD Feature Guarantee coverage and minimum data rate at cell edge Support high data rate Frequency Band -Advantage in large cell size Low requency band (e.g. 900MHz, GHz) -Eicient MIMO system using channel reciprocity -Antenna calibration is important High requency band (e.g..ghz) FDD (Low Freq)TDD (High Freq.) would be a good harmonization. Downlink Uplink OFDMA/TDD with MIMO spatial scheduling A New System Solution to reduce terminal s transmit power, reduce amount o control signaling, & improve spectral eiciency subband group group subband subband subband subband requency 's position s Feature o proposed OFDMA/TDD with MIMO spatial scheduling Key concept Exploit channel reciprocity more in wireless systems New technologies ) Pilot-based CSI reporting (OFDMA, MIMO mode) ) Remote control o transmit beamorming in MIMO ) Uplink and downlink spatial scheduling ) Partial CSI reporting Pilot-based CSI reporting (OFDMA, MIMO mode) ---Technology to reduce amount o control signaling and coordinate intererence between cells ---

3 's CSI Reporting in OFDMA CSI (or CQI) reporting on subband basis is essential or requency and spatial scheduling CSI (or CQI) Almost lat ading Data Packet CQI reporting scheduled packet CQI CQI CQI(e.g. bits) () () () (F) A large amount o CQI inormation bits needed on uplink New Challenge to Use Channel Reciprocity in CSI Reporting I terminal transmits UL pilot signal with constant power in subbands, Pilot signal h P s ReceivedPower Transmit Power P s h can get channel gain h, but cannot get h (Constant) Proposed Pilot-Based CSI Reporting (OFDMA mode) controls transmit power to be /I in the -th subband (I : terminal s intererence+noise power in the -th subband) Intererence Pilot signal h I : Intererence-plus-noise power in the -th subband P s = I P s ReceivedPower Transmit Power I I h h can get terminal s using channel reciprocity CSI Reporting in MIMO Channel Transmit Power: P s Channel State Ino. intererence near-by [Problem] It was diicult to report spatial intererence inormation. [Proposed method] can report spatial intererence inormation easily

4 Downlink Principle o proposed CSI reporting Joint Channel J MIMO Whitening Channel Filter -/ H(M N) R IN (M M) Intererence Intererence+noise components are white with power ( White Noise) Intererence is similar to white noise ater whitening process Uplink Joint Channel J MIMO Whitening Channel Filter H T (N M) -/ (R IN ) T (M M) r(p) I can get knowledge o channel J, can optimize DL transmission considering terminal s intererence. Pilot-based CSI reporting (MIMO mode) -/ ) transmits pilot signals using precoding matrix (R IN ) T H T Linear Precoding -/ r (p) r (p) (R IN ) T r M (p) Pilot Signals (R IN : Intererence-plus-noise covariance matrix) ) can compute optimal transmit beam & predict estimates Channel Response: J=(R Tx beam weight : Received : (*: Conjugate, T : Transpose, e n : n-th eigenvector, n : eigenvalue ) -/ IN H) T w =e n (J * J T n )=e =P s n (J * J T n(h R - IN H) )=P s n (H R - n IN H) Throughput Perormance ) can consider spatial directions o DL intererence in transmission control prediction n (pre) w Intererence w Adaptive Modulation & Coding(AMC) (pre) n Modulation db No use.0.db QPSK..dB QAM.dB QAM MMSE Rx beamorming Throughput (Packets/Frame) Pilot-based CSI reporting Pilot eedback w/o precoding No intererence intererence signal intererence signals P DL, S /P DL, N [db] X MIMO Intererence i.i.d Gaussian channel H Higher throughput is obtained using pilot-based CSI reporting st nd T

5 Pilot-Based CSI Reporting in Cellular Systems can control MIMO transmission, considering DL co-channel intererence. This pilot signal is also used as sounding signal or UL scheduling. orthogonal pilot transmission based on CDM co-channel intererence Remote control o transmit beamorming in MIMO ---Technology to reduce amount o control signaling--- co-channel intererence Precoding Precoding Precoding (long reporting period) Transmit power is controlled by open-loop power control. For Eicient Multiuser MIMO Uplink ) decides the k-th terminal s transmit weight based on system optimization ) inorms the k-th terminal o transmit weight v k Remote control o terminal s Tx beam is necessary. Transmit weight v k Instruction o transmitbeamorming optimizesterminal's transmitbeamorming instructs terminal on his transmit beamorming Eicient Remote Control Scheme in TDD controls terminal s Tx beam using channel reciprocity. ) sends orthogonal pilots using N adaptive beams (N: Number o s antennas) Pilot signals w w w N K [ w,...,w N ] =N / (B * B T ) - B* tr{(b * B T ) - } / T T [ B= H v / H v T T,..., H K v K / H K v K, b K+,...,b N ] T T H v,..., H K v K b K+... b N v k : Target Tx weight or k-th terminal ) computes Tx weight v k rom response o pilots v v k v K v k = (A * T k A +P 0 I) - k a* k k (A k* A kt +P 0 I) - a k k * P 0 : Pseudo-noise Power A =[a k k,..., a N k ] a j k: Response vector o j-th pilot at k-th terminal

6 In TDD systems, can control many terminals Tx beams simultaneously with high accuracy, using channel reciprocity. Amount o DL signaling is small ( symbols in case o antennas at ) Pilot signals w Pilot signal power : P s w N Noise power P z K Error [db] error=e[ n (ideal) - n ] / (ideal) n : Ideal received at n : Actual received at s s P s [db] /P z Multiuser MIMO with Spatial Scheduling - Uplink Scheduling - Downlink Scheduling ---Technology to reduce terminal s transmit power and improve spectral eiciency --- Uplink Spatial Scheduling Control Procedure o UL Spatial Scheduling v Appropriate combination o terminals is selected. Selected terminals sends packets based on SDMA v v k Further perormance improvement K Multiuser(MU)-MIMO w/o spatial scheduling Optimization s Tx & Rx beam Channel measurement K K Scheduling instruction Uplink Downlink Uplink Packet Transmission obtains channel knowledge decides terminal k(n), Tx weight v n, (n=,..,n) sends UL scheduling instructions to terminals (terminals, Tx weights, )

7 SDM-based Scheduling Instructions UL Spatial Scheduling Algorithm v w w N ) transmits instructions (pilot, terminal ID, ) based on SDM towards terminals o interest. Pilot 's Pilot signal 's ID instruction w Pilot 's signal signal ID ID Receive Beamormig ) Each terminal perorms Rx beam and checks all instructions at individual Rx beam outputs. v K selects terminal k(n) and his Tx beam v n, which achieve largest, successively. (n=,,n) v ) I an instruction includes his ID, the terminal computes Tx weight v n using responses o DL v k pilot signals and transmits UL packet according to the instructed Tx beam & v Largest K v Largest K New SDM-based instruction signaling with beamorming gain v v N v System Throughput [b/s/hz] Perormance o UL Spatial Scheduling : antennas/ : antennas 0 0 MU-MIMO with scheduling MU-MIMO w/o scheduling SU-MIMO db P z [db] P s SU-MIMO terminal is supported. #l Subcarrier - MU-MIMO with scheduling terminals are selected among terminals. - MU-MIMO w/o scheduling terminals are always supported. Spatial scheduling has db transmit power gain, compared to SU-MIMO Downlink Spatial Scheduling Appropriate combination o terminals is selected. transmits packets to selected terminals based on SDM w w w k K Optimization s Tx & Rx beam Tx beam nulliies intererence to other terminals Rx beams

8 Control procedure o DL spatial scheduling SDM-based DL Packet Transmission Channel measurement w K w k w Packet Transmission Uplink Downlink Packet Transmission obtains channel knowledge decides terminal k(n), s Tx weight w n, (n=,..,n). sends DL packets based on SDM w w Rx beamorming w N Pilot 's signal ID Pilot 's signal ID Preamble ) multiplexes DL packets spatially. Tx weight w Data Data Payload Tx weight w N ) perorms Rx beamorming, checks terminal ID, and receives packet o interest based on inormation K v v Max DL Spatial Scheduling Algorithm v N v DL algorithm is similar to UL algorithm. selects terminal and his virtual beam v n Max on virtual UL, successively. v Finally, decides s Zero-Forcing beams. v w w N w v N v Zero-Forcing Beamorming System Throughput [b/s/hz] 0 0 Perormance o DL Spatial Scheduling : antennas/ : antennas Spatialscheduling ( terminals among terminals) MU-MIMO ( terminals) /P z [db] P s. SU-MIMO SU-MIMO terminal is supported. Subcarrier - MU-MIMO with scheduling terminals are selected among terminals. - MU-MIMO w/o scheduling terminals are always supported. UL & DL spatial scheduling has similar perormance. MIMO spatial scheduling enhances throughput signiicantly. #l

9 Partial CSI reporting in MIMO Partial CSI Reporting in TDD/MIMO [Full CSI reporting] transmits M pilot signals on UL (M: number o terminal s antennas) [Partial CSI reporting] transmits m 0 (<M) pilot signals rom m 0 eigenbeams. ---Technology to reduce amount o control signaling--- M orthogonal pilot signals m 0 orthogonal pilot signals M antennas m 0 eigenbeams Transmission control under partial CSI perorms transmission control, supposing Nxm 0 virtual MIMO H k which includes terminal s linear transorm. N antennas Nxm 0 Virtual MIMO channel m 0 H k Linear transorm outputs k H k' m 0 k' Linear transorm outputs virtually supposes that k-th terminal has m 0 antennas System Throughput [b/s/hz] Throughput o spatial scheduling using partial CSI reporting (UL, Ps/Pz=0dB, : antennas, : antennas) 0 Full CSI reporting (m 0 =) Partial CSI reporting (m =) 0 Partial CSI reporting (m 0 =) Full CSI reporting (m 0 =) 9 0 Number o s K Partial CSI reporting (m 0 =) keeps high throughput, reducing UL signaling 0%.

10 OFDMA/TDD System with MIMO Spatial Scheduling A New System Solution to reduce terminal s transmit power, reduce control signaling, & improve high spectral eiciency Downlink Uplink subband OFDMA/TDD with Spatial Scheduling Using presented technologies, OFDMA/TDD with MIMO spatial scheduling can be achieved. group group subband subband subband subband requency 's position s s low transmit power, high spectral eiciency, & practical amount o signaling are consistent. Feasibility in Frame Format Overhead and Spectral Eiciency Amount o signaling is acceptable in a realistic rame ormat Subcarriers [Overhead] Including all control signaling (pilot, instructions, guard time, ACK, CSI, etc.), overhead is about 0% (could be urther improved) (00kHz) (00kHz) Orthogonal pilot signals E rom last beamormers ( symbols) Orthogonal pilot signals C rom each antenna ( symbols) Orthogonal pilot signals D rom each beamormer ( symbols) ID (* bits=symbols) (* bits= symbols) Data packet (SDM transmission based on DL spatial scheduling) Data packet (SDM transmission based on UL spatial scheduling) Other controlsignals Other control signals.ms.ms.ms Guard Period DOWNLINK UPLINK GI Data.s.0s Orthogonal pilot signals B ACK rom lastbeamormers (*0bits=symbols).s ACK(*0bits=symbols) ( symbols) ACK or UL packet (SDMTransmission) UL schedulingintructions (SDMTransmission) Feedbackinormation ( symbols / terminal, single-usertransmission) Orthogonal pilot signals A rom terminals or partial CSI reporting Downlink ield Pilot signal C UL scheduling ino. ACK Data Packet Total Symbols 0 Uplink ield Guard time CSI reporting ACK Data Packet Total Symbols [Spectral eiciency] Max. spectral eiciency 9 bps/hz (Under max. antennas at and terminals)

11 Conclusion An new system solution towards G mobile : Multiband system + OFDMA/TDD with MIMO spatial scheduling Multiband system will support wide coverage and high data rate -Gradual migration to G system, keeping low investment risk OFDMA/TDD with MIMO spatial scheduling -Spatial scheduling is promising in TDD system. -High capacity & low Tx power are achievable under practical signaling Use o channel reciprocity will sophisticate uture wireless systems Publications Multiband system [] Y. Hara, K. Oshima,``Multiband mobile communication system or wide coverage and high data rate,'' IEICE Trans. on Commun., pp. -, Sept. 00. [] Y. Hara, A. Taira,``System coniguration or multiband mobile systems,'' WWRF#, Dec. 00. OFDMA/TDD system with MIMO spatial scheduling [] Y. Hara, A.-M. Mourad, K. Oshima,``Pilot-based channel quality reporting or OFDMA/TDD systems with cochannel intererence,'' IEICE Trans. on Commun., pp. -0, Sept. 00. [] Y. Hara, K. Oshima,``Pilot-based CSI eedback in TDD/MIMO systems with cochannel intererence,'' Proc. o VTC00 Fall, Sept. 00. [] Y. Hara, K. Oshima,``Remote control scheme o transmit beamorming in TDD/MIMO systems,'' Proc. o PIMRC0, Sept. 00. [] Y. Hara, L. Brunel, K. Oshima,``Uplink spatial scheduling with adaptive transmit beamorming in multiuser MIMO systems,'' Proc. o PIMRC0, Sept. 00. [] Y. Hara, L. Brunel, K. Oshima,``Downlink spatial scheduling with mutual intererence cancellation in multiuser MIMO systems, Proc. o PIMRC0, Sept. 00. []. Hara, K. Oshima, ``Spatial scheduling using partial CSI reporting in multiuser MIMO systems,' IEICE Trans. on Commun., Feb. 00 (to appear). [9] Y. Hara, A. Taira, K. Oshima,``OFDMA/TDD/MIMO system with spatial scheduling,'' Proc. o VTC00 Spring, April 00. [0] Y. Hara, K. Oshima, ``Transmit power control or pilot-based CQI reporting in OFDMA/TDD systems,'' Proc. o PIMRC00, Sept. 00.