Multiple Antenna OFDM solutions for enhanced PHY

1. Multiple Antenna OFDM solutions for enhanced PHY Presented by: Alexandre Ribeiro Dias Contributors: A. Ribeiro Dias, Stéphanie Rouquette-Léveil, Sébastien Simoens Motorola Labs Paris +33 (1) 69 35 48 32 alexandre.ribeirodias@motorola.com Alexandre Ribeiro Dias - Motorola Labs

2. Introduction (1/2) • PAR requirements: • Define modifications to both 802.11 PHY and MAC so that a maximum throughput of at least 100Mbps at the MAC SAP is enabled • Increase PHY performance (1/2): • If same bandwidth (proposed functional requirement) multiple antenna techniques are required to increase the peak data rate with good coverage (add advanced coding schemes?) • Which multiple antennas techniques should be used? • How many antennas can be considered? Alexandre Ribeiro Dias - Motorola Labs

3. Introduction (2/2) • Increase PHY data rate (2/2): • How does « 100Mbps at the MAC SAP » translates in terms of PHY data rate requirements ( depends on MAC efficiency  depends on MAC amendment)? • Increase MAC SAP goodput: • How high can the throughput be with an enhanced PHY and 802.11 or 802.11e MACs? • How can this efficiency be increased with backward compatibility constraints ? Alexandre Ribeiro Dias - Motorola Labs

4. UL Maximum Ratio Combining (MRC) Multiple antenna techniques (1/5) • How can multiple antennas at the AP be used to improve performance of legacy STAs? Partial channel state information, Closed-Loop (CL) technique Transmit Selection (TS, per subcarrier, per antenna) DL Alexandre Ribeiro Dias - Motorola Labs

5. Multiple antenna techniques (2/5) • Space Time Block Codes (STBC) to benefit from spatial diversity (with MRC) • Increase communication reliability/coverage • Not optimal for high data rates Alexandre Ribeiro Dias - Motorola Labs

6. ML, ZF, MMSE, SIC… based receivers S/P … … Multiple antenna techniques (3/5) • Spatial Division Multiplexing (SDM): • Data rate multiplied by number of transmit antennas • Transmit diversity not exploited • In general (depending on decoder) Nr  Nt Alexandre Ribeiro Dias - Motorola Labs

7. SDM combined with STBCs (Open Loop OL) • SDM combined with TS (Closed-Loop CL) N<Nt data streams Transmit Selection (TS, per subcarrier, per antenna) S/P S/P Multiple antenna techniques (4/5) • Hybrid schemes: increase data rate and exploit transmit diversity for higher robustness/good range Alexandre Ribeiro Dias - Motorola Labs

8. Multiple antenna techniques (5/5) • Open-Loop vs Closed-Loop? • CL techniques known to provide substantial gain but how CSI can be obtained at the TX? • Reciprocity assumption in TDD systems? • Are RF front-ends identical on UL and DL? Additional overhead for calibration? • Delay between UL and DL needs to be controlled • Feedback link? • Needs to be implemented and additional overhead… Alexandre Ribeiro Dias - Motorola Labs

9. Multiple antenna PHY example • How many antennas should be used? • Antenna configuration should depend on STA size (laptop, mobile handset) • Ex: fixed number of antennas at the AP, several number of antennas (hence multiple antenna techniques) at the STA depending on dimensions/legacy Alexandre Ribeiro Dias - Motorola Labs

10. Alternative OFDM modulators • Cyclic Prefix OFDM (CP-OFDM) is an interesting option with simple equalisation scheme: • BUT: • Zero Padded OFDM (ZP-OFDM, CP replaced by zeros) • solves the sensibility to channel zeros locations • Pseudo-Random Postfix OFDM (PRP-OFDM, CP replaced by pseudo-randomly weighted known postfix): • Keeps all advantages of ZP-OFDM • Low complexity channel/synchronisation estimation and tracking possible, support for increased mobility (longer packets?) • Alternative OFDM modulators worth being considered Alexandre Ribeiro Dias - Motorola Labs

11. Multiple Antenna System Processing OFDM modulation OFDM modulation scrambler coder interleaver mapping IEEE802.11a block New block Legend: PHY simulation results (1/2) • Simulation parameters: • 256QAM investigated to increase spectral efficiency • NLOS TGn D channels (no spatial correlation, ≠ path loss model) • Packet size: 512Bytes, 23dBm transmit power • ZF receiver when SDM techniques (not optimal but least complex) • Perfect MIMO channel estimation • All IEEE802.11a functional blocks are kept: Alexandre Ribeiro Dias - Motorola Labs

12. PHY simulation results (2/2) • Hybrid schemes for reliability and high data rate? • Hybrid schemes: range increase not very important, however reduces constraints on SNR requirements Alexandre Ribeiro Dias - Motorola Labs

13. MAC simulation results • Simulation Assumptions: • .11e MAC with Group Acknowledgement • 1500 byte payload • 16 MPDU per Group • Simulation Results: • Only 100 Mbps effective vs 162Mbps theoretical with 3 streams and 64QAM • MAC Efficiency is of only 54% for the 216Mbps mode • .11e with Group Ack is too inefficient ! Alexandre Ribeiro Dias - Motorola Labs

14. Possible MAC improvements • Need to increase the PHY burst size by aggregating numerous packets (without SIFS between each packet) • Keep individual FCS per packet for SR-ARQ • Possibility to rely on the Contention Free Period of .11e to introduce more efficient TDD/TDMA MAC, to reduce constraints on PHY Alexandre Ribeiro Dias - Motorola Labs

15. Conclusions • .11e MAC not efficient with high data rate PHY • Increasing the MAC efficiency would relax the constraint on PHY peak data rate • Number of antennas/techniques used should be chosen to reduce constraints on SNR requirements and keep reasonable range • Antenna configurations should depend on STA size (set of different cost/performance trade-offs) Alexandre Ribeiro Dias - Motorola Labs