Performance Evaluation of MIMO midamble design for IEEE

Performance Evaluation of MIMO midamble design for IEEE 802. 16 m IEEE 802. 16 Presentation Submission Template (Rev. 9) Document Number: IEEE C 802. 16 m-09/1237 r 2 Date Submitted: 2009 -01 -07 Source: Jerry Pi, Jiann-An Tsai, Bruno Clerkx, David Mazzarese Samsung Electronics zpi@sta. samsung. com Venue: Re : 802. 16 m amendment working document Base Contribution: IEEE C 802. 16 m-09/1237 r 2 Purpose: To discuss and adopt the proposed text in the revision of the 802. 16 m SDD Notice: This document does not represent the agreed views of the IEEE 802. 16 Working Group or any of its subgroups. It represents only the views of the participants listed in the “Source(s)” field above. It is offered as a basis for discussion. It is not binding on the contributor(s), who reserve(s) the right to add, amend or withdraw material contained herein. Release: The contributor grants a free, irrevocable license to the IEEE to incorporate material contained in this contribution, and any modifications thereof, in the creation of an IEEE Standards publication; to copyright in the IEEE’s name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEE’s sole discretion to permit others to reproduce in whole or in part the resulting IEEE Standards publication. The contributor also acknowledges and accepts that this contribution may be made public by IEEE 802. 16. Patent Policy: The contributor is familiar with the IEEE-SA Patent Policy and Procedures: <http: //standards. ieee. org/guides/bylaws/sect 6 -7. html#6> and <http: //standards. ieee. org/guides/opman/sect 6. html#6. 3>. Further information is located at <http: //standards. ieee. org/board/pat-material. html> and <http: //standards. ieee. org/board/pat >. 1

Background • AWD, section 15. 3. 5. 4. 2 – “MIMO midamble is used for PMI selection in closed loop MIMO. For OL MIMO, midamble can be used to calculate CQI. The midamble signal occupies the one OFDMA symbol in a DL sub-frame. For the type-1 subframe case, the remaining 5 consecutive symbols form a type-3 subframe. For the type-2 subframe case, the remaining 6 consecutive symbols form a type-1 subframe. ” • AWD text proposal for detailed MIMO midamble design is provided in C 80216 m-09_1236 • This contribution provides performance evaluation of the MIMO midamble design as in C 80216 m-09_1236 2

Time domain interlacing • Time domain interlacing enables collision avoidance of midamble among neighboring cells – Allow power boosting of midamble to improve channel estimation performance • The subframe index for MIMO midamble is explicitly signaled • The OFDM symbol index within the subframe is derived from the Cell_ID – MIMO midamble is transmitted on the first OFDM symbol of a subframe if the Cell_ID is even – MIMO midamble is transmitted on the last OFDM symbol of a subframe if the Cell_ID is odd 3

Frequency domain placement • Cell-specific frequency domain shift applied – To further minimize collision of midamble among neighboring cells • Terminology – – i : antenna index, i = 1, …, NT j: subcarrier index, j = 0, 1, …, 17 k: Cell_ID n: Frame index • In frame n, midamble for antenna i is transmitted on subcarrier j in cell k – if and only if the following two conditions are met • mod(j + k/2 , 9) 0 • mod(j + k/2 , 9) + 4 mod(n, 2) – 1, 8) = BRO(i – 1 , 3) – BRO(x, 3) is the 3 -bit bit-reversal value of x • Subcarriers not used for midamble transmission are reserved as null subcarriers – The unused power can be allocated to midamble for power boosting – Null subcarriers can be used for interference estimation 4

Properties of the midamble • Up to 16 time-domain interlaces to reduce midamble collision • Midamble is transmitted once every frame in the same OFDM symbol • Midamble for each antenna is regularly spaced in frequency (1 in every 9 subcarriers) A single midamble channel estimator for all antennas in 2, 4, or 8 Tx cases • Midamble in adjacent frames are staggered – Increase the frequency sampling rate to 16 frequency samples per subband per antenna – Improve channel estimation performance by 2 -D MMSE channel estimation across frames • Midamble among neighboring cells are shifted in frequency – Further minimize midamble collision – Particularly beneficial in TDD system with limited DL subframes for midamble transmission • Enable channel estimation to benefit from power boosting – 2 Tx : 6. 5 d. B, 4 Tx : 3. 5 d. B, 8 Tx : 0. 5 d. B – Further midamble boosting possible if low PAPR sequences are designed 5

MIMO midamble for 2 Tx systems 6

MIMO midamble for 4 Tx systems 7

MIMO midamble for 8 Tx systems 8

Half-symbol-delay Time-domain Correlation 1, 2, 4 -Tx midamble has lower correlation than random signal 8 -Tx midamble has the same correlation as random signal 9

Link level simulation • Simulation configuration – Geometry simulated: [-6, -3, 0, 3, 6, 9, 12, 15] d. B – MMSE receiver with estimated interference and channel matrix from dedicated pilot – 4 -PRU, rank-1 transmission – PMI/CQI selection/feedback modeled • • 4 -bit quantized CQI (16 m AWD MCS Table) 16 m AWD codebook Midamble placed in the 2 nd subframe. CQI/PMI available in the next frame 10% CQI/PMI erasure (BS uses the previous CQI/PMI in case of erasure) – Closed-loop rate adaptation to control the BLER to 10% – Ped B, 3 kmph – 100 seconds (160000 subframes) • Compared cases: – Reuse 1, INR = 0 d. B • Frequency reuse 1, time-domain interlacing, no dominant interferer – Reuse 1, INR = 10 d. B • Frequency reuse 1, time-domain interlacing, with dominant interferer – Reuse 3 : All midamble in the same OFDM symbol, frequency reuse of 3 • Frequency reuse 3, time-domain collision 10

Interference model • Only used for simulation of frequency reuse 1 cases • Two equal strength dominant interferer • Interference midamble placement – Interferer 1 transmit midamble at the 1 st OFDM symbol of subframe 3 – Interferer 2 transmit midamble at the last OFDM symbol of subframe 3 – Making subframe 3 the “worst” subframe • • Midamble structure as of page 7 simulated (4 Tx midamble) Interference-to-Noise-Ratio simulated: [0, 10] d. B Midamble staggers across frames (for frequency reuse of 1) Midamble boosting – 3 d. B beyond data (for frequency reuse of 1) – 5 d. B improvement on midamble SNR (for frequency reuse of 3) 11

LLS results 12

LLS results • Observation – No performance difference observed between INR = 0 d. B and INR = 10 d. B – The suspected impact of degraded channel estimation due to collision between dedicated pilot and other cell midamble is not seen (or is dominated by the improvement of performance due to more accurate PMI/CQI feedback) – Frequency reuse 1 outperforms Frequency reuse 3 by 0. 5 d. B 13

System level simulation • Comparison – Fixed subframe transmission vs. non-fixed subframe transmission – Symbol interlacing vs. Non Symbol interlacing • Simulation configuration – – – – System topology: 19 cells / 57 sectors wrap-around MIMO configuration: (2, 4, or 8) Tx, 2 Rx System bandwidth: 10 MHz Subband size: 4 PRUs CQI feedback: once per 5 ms (all sub-bands) Rank adaptation: ON Receiver: MMSE Scheduling: Proportional Fair • Metric: – System throughput – 5 -percentile user throughput 14

SLS Results 8 TX Midamble Relative Cell Edge Throughput (%) Relative Sector Throughput (%) Full loaded (Non Fixed subframe TX Only) 100 50% loaded (Non Fixed subframe TX Only) 112 107 25% loaded (Non Fixed subframe TX Only) 115 110 50% load (Symbol Interlacing Only) 107 103 50% loaded (Non Fixed subframe TX+ Symbol Interlacing) 118 108 15

SLS Results 4 TX Midamble Relative Cell Edge Throughput (%) Relative Sector Throughput (%) Full loaded (Non Fixed subframe TX Only) 100 50% loaded (Non Fixed subframe TX Only) 111 106 50% loaded (Symbol Interlacing Only) 107 103 16

SLS Results 2 TX Midamble Relative Cell Edge Throughput (%) Relative Sector Throughput (%) Full loaded (Non Fixed subframe TX Only) 100 50% loaded (Non Fixed subframe TX Only) 107 104 50% loaded (Symbol Interlacing Only) 103 102 17

Summary • Both schemes of non-fixed subframe transmission and symbol interlacing provide significant system gain over fixed subframe/fixed symbol transmission • As shown, gain is particularly significant for lightloaded system • As observed, the improvement on cell edge throughput is substantial as compared to sector throughput • Recommendation – TGm adopt the text proposal in C 80216 m-09_1236 into AWD 18