MIMO - A Key to Broadband Wireless. Volker Jungnickel. Fraunhofer. Institut. Nachrichtentechnik Heinrich-Hertz-Institut

Transcription

1 MIMO - A Key to Broadband Wireless Volker Jungnickel

2 Outline Introduction Channel properties Algorithms Real-time implementation Conclusions 2

3 Introduction People really want wireless internet access anywhere, anytime and at low cost per bit in future: bi-directional multi-media communications Challenges develop a full-coverage symmetric broadband wireless access system (4G) for large numbers of users to communicate simultaneously in multiple cells realize high bandwidth for each user in limited spectrum Promising approach: Exploit the spatial domain! enormous bandwidth is hidden in wireless channel bandwidth can be reused by the so-called spatial multiplexing 3

4 Principle of MIMO MIMO = multiple-input multiple-output rich scattering wireless channel R1 Multiple TX antennas or users, each transmitting an independent data stream. Signal is received with multiple antennas as well. TX data DEMUX a 1 a 2 a 3 a 4 T1 T2 T3 T4 R2 R3 R4 R5 R6 Space-Time Signal Processing RX data Training phase: Based on known pilot signals, the RX learns the channel between each TX and each RX antenna. Data transmission phase: Based on the channel side information, data streams can be separated at the RX. 4

5 Channel properties: Indoors capacity [bps/hz] Tx-Rx distance ca. 1 m NLOS with LOS Rayleigh σ= 20 db number of antennas (n = n t ) r Due to rich scattering, indoor capacity increases linearly with number of antennas, with no line-of-sight (NLOS). With line-of-sight (LOS), there is a minor saturation with more than 10 RX antennas. Up to 16 times the bandwith in the same spectrum. Gbit/s WLANs, n, wireless networks in factory halls, 5

6 Channel properties: Outdoors Multi-cell multi-user scenario HHI Two applications 1 HFT single-base to mobile MIMO at link level ILR multi-base to mobile: use multiple base station antennas MATH Cross-cell MIMO at system level 6

7 Antennas Mobile terminal Base station 10 antenna elements (5 x dual polarized patches) at mobile terminal 16 antenna elements (8 x dual polarized) at base station 7

8 Short range (with free LOS) power at short distance in outdoor scenario, the probability of free line-of-sight is significant since terminal antenna elements look into different directions, just one element (h/v pol.) receives significant power singular values MIMO capacity only two useful spatial channels, due to two polarization modes reduced capacity compared to rich scattering scenario but received power is higher at short distance use adaptive modulation (per-antenna rate control) for short range 8

9 Wide area (no LOS) power at longer distance in outdoor scenario (> m), the probability of LOS is negligible all spatial channels can be loaded with data, since rich scattering is dominant singular Values MIMO capacity capacity is similar to indoors, but the received power is reduced even non-adaptive MIMO techniques may work properly important for highly mobile users 9

10 Multiple cells: Interference total coverage (power of best base in 5m square) SIR due to other bases red = minor interference, blue = interference limited path gain [db] Use antennas from multiple base stations to jointly detect users in adjacent cells! 10

11 Enhanced capacity per cell (SNR = 10 db) HHI -8 x10 2 HFT MATH ILR 1 pa th gain [lin.u.] cdf Pos capacity per cell [bps/hz] with orthogonal scheduling (GSM), resources must be shared reduced cell capacity mean capacity of isolated cells = expected with cross-cell interference cancellation joint channel capacity is even larger, due to enhanced channel rank Cross-cell MIMO processing may be a key to full-coverage broadband wireless! 11

12 Algorithms ZF: multiplies with (pseudo-) inverse channel matrix ML: looks for closest signal constellation x = [ x 1, x2] H n + y 1 H Zero Forcing decided xˆ here TX space RX space signal space ML looks here for min. dist.! 12

13 Performance 8 Tx 8 Rx, BPSK uncoded P: perfect channel information at RX CE: with realistic channel estimation (8 orthogonal pilot sequences) ML realizes full spatial diversity due to channel est. errors Better algorithms perform better are more robust against channel estimation errors but do have higher complexity 13

14 Combination with OFDM broadband MIMO OFDM: Parallel transmission on multiple narrow-band sub-carriers Waveforms remain orthogonal after passing through a multi-path channel Le istung figure from SIEMENS Zeit No intra-cell interference, unlike WCDMA Complexity is significantly reduced 14

15 MIMO-OFDM xperimental System Principle multipath channel 5 x FFT channel joint 5 detection x estimation FFT joint detection OFDM-BLAST Pilots Data compute weight matrices per OFDM Tone

16 Real-time implementation 3-antenna Tx 5-antenna Rx 3 Tx, 5 Rx antennas, 100 MHz jointly developed by HHI, IAF and SIEMENS first-in-the-world real-time data transmission at 1 Gbit/s over the air with omni-directional antennas and NLOS (no line-of-sight) efficient implementation TX : 2 FPGAs (V2/6000) RX : 2 FPGAs (V2Pro-100) 1 DSP (TI 6713/225) 100 MHz signal monitoring 16

17 1 Gbit/s transmission experiments 3 omni-directional Tx antennas are moved along a 4 m track in an office low mobility is supported 17

18 The 1 Gbit/s team Photo is taken on Nov. 30, 2004, after first trials with 3x4 configuration.

19 Thanks for cooperation Andreas Forck Thomas Haustein Holger Gäbler Dr. Udo Krüger Kirsten Krüger Volker Pohl Malte Schellmann Dr. Clemens von Helmolt Stefan Schiffermüller Armin Brylka Georghe Istoc Haifeng Chen Stefan Jaeckel Christoph Juchems Frank Luhn Marian Pollock Dr. Egon Schulz Wolfgang Zirwas Joseph Eichinger Fraunhofer HHI für angewandte Funksystemtechnik (IAF), Braunschweig Siemens München, Future Radio Concepts 19

20 Conclusions In general, MIMO is an efficient technique to exploit the spatial dimension in mobile communications. MIMO channel properties are excellent in indoor and wide-area outdoor scenarios. In short-range outdoor scenario, fading may be correlated. Adaptive transmission may be helpful in such cases. The reduced number of streams is compensated by higher order modulations, which is feasible since SNR is higher. Cross-cell MIMO-OFDM processing has the potential to remove both intra- and inter-cell interference in cellular networks with low computational complexity. We have demonstrated that MIMO-OFDM can be implemented in real-time at data rates up to 1 Gbit/s. The technology is ripe for WLAN, cellular applications need further research. MIMO-OFDM has the potential to realize full-coverage broadband wireless access with reuse of resources in each cell. 20

21 Thank you very much for listening. I am looking forward to your questions. 21