Đang tải... (xem toàn văn)
In this paper, we investigate and compare the systemlevel throughput of nonorthogonal multiple access (NOMA) with a successive interference canceller (SIC) and orthogonal multiple access (OMA) in the downlink using a realistic file transfer protocol (FTP) traffic model. In the FTP traffic model, a user is assigned a finite payload to transmit when it arrives, and it leaves the system after the payload transmission is completed. This model is more realistic than the fullbuffer traffic model, which is assumed in previous NOMA investigations. Furthermore, we evaluate the systemlevel throughput with various user scheduling criteria. Under the realistic finitepayload traffic model, a very high throughput gain in the vicinity of the cell using NOMA can be translated to improve the celledge user experience. Through extensive computer simulations, we clarify the behavior of NOMA with a SIC in the realistic traffic model in conjunction with various user scheduling criteria. The simulation results suggest that NOMA with a SIC is a promising multiple access scheme for systems beyond 4G.
IEEE ICC 2015 - Workshop on 5G & Beyond - Enabling Technologies and Applications Performance and Design of SIC Receiver for Downlink NOMA With Open-Loop SU-MIMO Keisuke Saito, Anass Benjebbour, Yoshihisa Kishiyama, Yukihiko Okumura, and Takehiro Nakamura 5G Laboratory, NTT DOCOMO, INC 3-6 Hikari-no-oka, Yokosuka-shi, Kanagawa-ken 239-8536 Japan keisuke.saitou.xt@nttdocomo.com Abstract— This paper investigates the performance and design of successive interference cancellation (SIC) receiver for downlink non-orthogonal multiple access (NOMA) combined with 2-by-2 open-loop single-user (SU)-MIMO Compared to NOMA with 1by-2 SIMO (Single Input Multiple Output), for NOMA with 2by-2 SU-MIMO we have to deal with both inter-user and interstream interference at the receiver of each user Therefore, we introduce different weight generation schemes for both before SIC and after SIC according to the transmission rank combination between the user equipment (UEs), while taking into account inter-user interference In addition, we compare the performance of different SIC receivers; specifically, codeword level SIC (CWIC), symbol level SIC (SLIC), and ideal SIC Assuming LTE (Long Term Evolution) Transmission Mode (TM3), link-level simulations are conducted under different allocated transmit powers, rank combinations, and modulation and coding schemes (MCS) The link-level simulation results showed that the CWIC receiver achieves higher performance compared to SLIC and achieves almost the same performance compared to ideal SIC when the power ratio of cell-center UE is below 0.35, and the MCS of cell-center UE and cell-edge UE are 16QAM (R = 0.49) and QPSK (R = 0.49), respectively Keywords—5G; NOMA; Successive interference cancellation; SIC; SU-MIMO I INTRODUCTION The commercial service of mobile communications system based on Long Term Evolution (LTE) Release [1, 2] standardized by the 3rd Generation Partnership Project (3GPP) was recently launched worldwide by many operators LTE adopts the full IP packet-based radio access and provides higher-speed, larger-capacity, and lower-latency radio access than 3rd or 3.5th generation systems LTE also adopts an intracell orthogonal multiple access (OMA) such as orthogonal frequency division multiple access (OFDMA) and singlecarrier frequency division multiple access (SC-FDMA) in downlink and uplink, respectively Nevertheless, following the anticipated exponential increase of mobile traffic, significant gains in capacity and quality of user experience (QoE) are required In order to achieve much higher levels of system performance than those for LTE Release 8, LTE-Advanced specified as LTE Release 10 [3, 4] and its enhancements are being studied in future releases (LTE Release 11 and beyond) As a candidate technology for further performance enhancements of LTE and LTE-Advanced, a downlink nonorthogonal multiple access (NOMA) with successive interference cancellation (SIC) receiver, which utilizes an additional new domain, i.e., the power domain for multiplexing 978-1-4673-6305-1/15/$31.00 ©2015 IEEE the signals of multiple users at the transmitter side and utilizes SIC receiver to separate their signals at the receiver side, is proposed With respect to previous works related to downlink NOMA, the concept and system-level performance are discussed and investigated under 1-by-2 SIMO (Single Input Multiple Output) systems in [5 - 8] In addition, the performance of codeword level SIC (CWIC) receiver considering the influence of EVM (Error Vector Magnitude) is investigated for the case of 1-by-2 SIMO using link-level simulations in [9] Furthermore, NOMA extensions to 2-by-2 single-user (SU) and multi-user (MU)-MIMO are proposed and evaluated using system-level simulations in [10 - 12] The goal of this work is two-fold: The first is to investigate the design of the SIC receiver for downlink NOMA combined with 2-by-2 open-loop SU-MIMO based on LTE TM3 (Transmission mode 3) [13] The second is to clarify the linklevel performance of different types of SIC receiver under different allocated transmit powers, rank combinations, and modulation and coding schemes (MCS) The rest of this paper is organized as follows In section II, we describe the concept of downlink NOMA SIC with SUMIMO Section III describes the SIC receiver design for NOMA with SU-MIMO In section IV, after describing the link-level simulation assumptions, we provide and discuss the results of the link-level performance of downlink NOMA SIC for 1-by-2 SIMO and 2-by-2 open-loop SU-MIMO for several types of SIC receivers Finally, we conclude the paper in Section V II DOWNLINK NOMA SIC WITH SU-MIMO Figure illustrates the concept of downlink NOMA SIC with SU-MIMO for the case of one base station (BS) and two user equipment (UEs); Figure 2(a) shows the transceiver configuration of BS transmitter In this paper, we assume that the number of UE is two, with one UE located at cell-center and the other UE at cell-edge At the BS transmitter, the transmit signal vector X is generated as follows: X P1 WTx,1X1 P2 WTx, X2 , (1) where Xi represents the transmit signal vector of UE #i (i = 1, 2), where E[|Xi|2] = 1, and Pi (P1 < P2, P1 + P2 = 1) represents the allocated transmit power to UE #i This means that the transmit signals for UE #1 and UE #2 are multiplexed in the power domain based on the allocated transmit power Pi WTx, i is the precoding weight matrix of UE #i At the UE receiver, Y, which represents the received signal vector of UE #i (i = 1, 2), 1161 IEEE ICC 2015 - Workshop on 5G & Beyond - Enabling Technologies and Applications Precoding Weight P2 P1 UE #1 BS Data buffer Cancellation Turbo encoder CRS Non-orthogonal MUX Power P2 P1 Frequency Power setting Data modulator Power setting UE #2 Stream #1 - SIC (UE #2) Decoding (UE #1) UE #1 - Decoding (UE #2) Ant #1 - (a) BS Transmitter Fig Concept of downlink NOMA SIC with SU-MIMO Received signal is represented by Yi HX Ni , MUX Ant #1 - CRS (2) Channel estimation MMSE where H represents the complex channel matrix of UE #i, and Ni is the additive white Gaussian noise (AWGN) vector of UE #i, where E[NiNiH] = 2I (I is the identity matrix) In order to combine 2-by-2 SU-MIMO transmission with NOMA, SU-MIMO is applied to each UE independently with up to layer (rank 2) transmissions per UE As a result, up to layer transmission is enabled by applying 2-by-2 SU-MIMO on the top of NOMA with UE multiplexing Regarding the transmit power, for the case of OMA with SU-MIMO, the transmit power of each user is split equally among transmission layers For the case of NOMA with SU-MIMO, the transmit power ratio of each UE is set based on the power allocation scheme such as full search power allocation (FSPA), fractional transmit power allocation (FTPA), or pre-defined user grouping and per-group fixed power allocation (FPA) [8]; then the transmit power of each layer within each UE is split equally among transmission layers in the same manner as OMA case III SIC RECEIVER DESIGN FOR NOMA WITH SU-MIMO When applying NOMA transmission, inter-user interference is caused and the amount of inter-user interference depends on the allocated power ratio between the UEs The larger is the difference in allocated power ratio between the UEs, smaller is the inter-user interference On the other hand, when NOMA is combined with SU-MIMO transmission, each UE will be subject to inter-stream interference in addition to inter-user interference Therefore, we investigate two different types of SIC receiver, symbol level SIC (SLIC) receiver and CWIC receiver, in order to suppress the inter-user and inter-stream interference effectively The SLIC receiver detects the interfering modulation symbols and cancels them without decoding, whereas the CWIC receiver detects and decodes the interfering data then cancels them Therefore, the CWIC receiver has better performance than the SLIC receiver However, the CWIC receiver may involve higher implementation complexity and larger latency as compared with the SLIC receiver On the other hand, in order to mitigate inter-user interference, we apply different received weight generation schemes between before SIC and after SIC according to transmission rank combination between the UEs Figure 2(b) shows the transceiver configuration of cellcenter UE (UE #1) receiver for the case of CWIC receiver First of all, UE #1 receiver calculates the received weight Interference user (UE #2) Demodulation & Decoding Interference replica generation Stream #1 - Desired user (UE #1) Interference cancellation MMSE Demodulation & Decoding Recovered data Stream #1 - (b) CWIC receiver for cell-center UE Fig Transceiver configuration matrix WRx, using the channel coefficients estimated using CRS (Cell-specific Reference Signal), which is specified by LTE Release [14], and generates the symbol vector of interfering UE (UE #2) Srep, as follows: S rep, WRx , Y1 (3) Then, when applying the SLIC, the replica signal vector of the interfering UE (UE #2), Xrep, is generated by detecting the modulated symbol Srep, On the other hand, for the case of CWIC receiver, Xrep, is generated by detecting and decoding Srep, After generating Xrep, and applying SIC, the received signal vector, Yc, is calculated as ˆ PW X Yc,1 Y1 H Tx , rep, (4) , ˆ is the estimated complex channel matrix Then, the where H demodulation and decoding are applied to obtain the transmitted binary sequence On the other hand, the cell-edge UE (UE #2) does not apply SIC as the interference signal from UE #1 is treated as noise; thus, only demodulation and decoding are applied to the received signal Y2 When applying SU-MIMO transmission on NOMA with SIC, we apply different received weights before and after SIC in such a way that inter-user interference is taking into account When decoding the signal of UE #2, we calculate the received weight matrix considering the inter-user interference based on the MMSE (Minimum mean squared error) criterion as: ˆ HH ˆ H ˆ HH ˆ 2I 1 H ˆH WRx , H 2 1 (5) ˆ is defined as follows: Where H i ˆ H ˆ PW H i i Tx , i (6) On the other hand, when decoding the signal of UE #1 after applying SIC, the received weight matrix is calculated as: 1162 IEEE ICC 2015 - Workshop on 5G & Beyond - Enabling Technologies and Applications ˆ HH ˆ 2I WRx ,1 H 1 1 subframe (1 ms) ˆH H (7) Fig Radio frame structure TABLE I xi (k 1) xi (k )* , , Subframe length 1.0 ms Symbol duration Effective data: 66.67 s + CP: 4.69 s Antenna configuration 1-by-2, 2-by-2 (uncorrelated) Channel coding / decoding Turbo coding (Constraint length: bits) / Max-Log-MAP decoding (6 iterations) SLIC, CWIC, Ideal SIC R1:R2 = 2:2 According to (5) [or (7)] R1:R2 = 1:1, R1:R2 = 2:1 According to (5) [or (7)] after extension to (10) Channel model Exponentially decaying 6-path Rayleigh fading Channel estimation CRS-based channel estimation Based on the aforementioned types of SIC receiver and the received weight generation schemes, we evaluate the performance of the SIC receiver IV SIMULATION EVALUATIONS (10) where hij(k) represents the complex channel coefficient of k-th subcarrier of i-th received antenna and j-th transmitted antenna Then, we calculate the received weight matrix by applying (10) to (5) and (7) 2) R1:R2 = 2:2 The complex channel matrix Hi is represented as follows: h (k ) h12 (k ) WTx,i H i Pi 11 , h ( k ) h ( k ) 21 22 15 kHz of after SIC, we define the channel matrix as in (11) to detect the cell-edge UE (UE #2, rank = 2) (9) and we define the precoding weight matrix WTx, i as the ~ identity matrix I The extended complex channel matrix H is defined as follows: ~ ~~ ~ Yi = HX Ni , x1 (k ) x (k ) ~ P2 *2 X P1 * x (k 1) , x ( k ) h11(k ) h12 (k ) h22 (k ) ~ h21(k ) H * * h12 (k 1) h11(k 1) h* (k 1) h* (k 1) 21 22 1200 Subcarrier separation MMSE receiver before SIC [or after SIC] where xi(k) represents the transmitted signal of k-th subcarrier of UE #i Moreover, we extend the received signal vector as follows: 20 MHz Number of subcarriers SIC receiver (8) T ~ Yi y1 (k ), y2 (k ),y1* (k 1),y2* (k 1) SIMULATION ASSUMPTIONS System bandwidth 1) R1:R2 = 1:1 For this case, SFBC is applied to both UE #1 and UE #2 The transmitted signal matrix ((2 transmit antennas) × (k-th and k + 1-th subcarriers)) is represented as follows: xi (k ) * xi (k 1) CRS (Antenna #1) CRS (Antenna #2) Control signal Data signal RB bandwidth (12 subcarriers, 180 kHz) Moreover, when applying 2-by-2 open-loop SU-MIMO transmission based on the LTE TM3, SFBC (Space-Frequency Block Coding), which encodes the same data differently and increases the SNR of the recombined data streams to obtain the transmit diversity, and large delay CDD (Cyclic Delay Diversity), which applies spatial multiplexing by precoding vector in order to increase throughput, are applied for rank and rank transmission, respectively Therefore, we generate the different received weights according to the transmission rank combination (UE #1:UE #2 = R1:R2) (11) where, the precoding weight matrix WTx, is applied based on large delay CDD as specified in LTE Release [13], and WTx, is set equal to WTx, over all subcarriers In this case, the received matrix is calculated by applying (11) to (5) and (7) 3) R1:R2 = 2:1 In order to detect cell-center UE first (UE #1, rank =1), we define the transmit signal vector and the channel matrix as in (10) for the case of before SIC On the other hand, for the case A Simulation Assumptions Link-level simulations are conducted to evaluate the performance of the SIC receiver on downlink NOMA with 2by-2 open-loop SU-MIMO The radio frame structure is given in Fig 3, and the simulation assumptions are given in Table I The radio frame structure is based on LTE Release specifications [14] The control signal is multiplexed on the 1st symbol, the data signal is multiplexed on after the 2nd symbol, and CRS, which is used for channel estimation, is multiplexed with the 1st, 4th, 7th, 11th symbol in the time domain and with the every subcarriers per RB (Resource Block) in the frequency domain for each transmit antenna The system bandwidth and the number of subcarriers of the OFDM signal are 20 MHz and 1200, respectively; with a subcarrier separation of 15 kHz At the BS transmitter, the information binary data sequence is turbo encoded with the coding rate of R and modulated using QPSK, 16QAM, or 64QAM The LTE TM3 transmission scheme is applied to resultant signal sequence of each UE, and the signals of both UEs are nonorthogonally multiplexed in the power domain based on the allocated power ratio (P1:P2) After insertion of CRS, the signal sequence is converted into an OFDM symbol with the duration of 66.67 s, followed by the addition of a cyclic prefix (CP) of 4.69 s; then, quadrature modulation is performed The number of multiplexing UEs is two which is based on the system-level simulation results [7], and antenna configuration is 1-by-2 and 2-by-2 The channel model is an instantaneous multipath Rayleigh fading model with a six-path exponentially decayed 1163 IEEE ICC 2015 - Workshop on 5G & Beyond - Enabling Technologies and Applications 10 10 -1 Ideal SIC SLIC CWIC -2 P1:P2 = 3:7 UE #1: 16QAM (R = 0.51) UE #2: QPSK (R = 0.51) -3 for achieving BLER of 10-1 (dB) 10 30 Required SNR of cell center UE Average BLER 10 Ideal SIC CWIC R1:R2 = 1:1 R1:R2 = 2:1 R1:R2 = 2:2 25 20 15 10 -5 10 15 20 25 30 Average received SNR for cell center UE (dB) 10 -1 Ideal SIC SLIC CWIC R1:R2 = 1:1 R1:R2 = 2:1 R1:R2 = 2:2 -2 -3 P1:P2 = 3:7 UE #1: 16QAM (R = 0.49) UE #2: QPSK (R = 0.49) -1 for achieving BLER of 10 (dB) 10 0.5 30 Required SNR of cell center UE Average BLER 10 0.1 0.2 0.3 0.4 Power ratio of cell center UE (P1) (a) MCS of UE #1: 16QAM (a) 1-by-2 SIMO 10 UE #1: 16QAM (R = 0.49) UE #2: QPSK (R = 0.49) 20 15 10 Ideal SIC CWIC R1:R2 = 1:1 R1:R2 = 2:1 R1:R2 = 2:2 -5 10 15 20 25 30 Average received SNR for cell center UE (dB) UE #1: 64QAM (R = 0.51) UE #2: QPSK (R = 0.49) 25 0.1 0.2 0.3 0.4 Power ratio of cell center UE (P1) 0.5 (b) MCS of UE #1: 64QAM (b) 2-by-2 SU-MIMO Fig Required SNR for achieving BLER of 10-1 (UE #1 performance) Fig BLER performance power delay profile model with a root mean squared (r.m.s.) delay spread value of 0.29 s, where the relative path power is decayed by dB and each path is independently Rayleighfaded with a maximum Doppler frequency of 10 Hz At the UE receiver, after CP removal, the received signal is demultiplexed into each subcarrier component using FFT At the cell-center user, UE #1, SLIC or CWIC is applied and compared with ideal SIC Then, MIMO signal detection is applied using the received weights Finally, the sequence of likelihood values is turbo decoded using the Max-Log-MAP algorithm with six iterations to recover the transmitted binary data Note that ideal SIC here refers to the ideal generation of the replica of interfering signal; however, interference cancellation may still remain imperfect due to channel estimation error On the other hand, for the cell-edge user, UE #2, SIC is not applied B Simulation Results First, we clarify the performance of SLIC and CWIC receivers Figure 4(a) shows the block error rate (BLER) performance of NOMA transmission for 1-by-2 SIMO, P1:P2 = 3:7, and SLIC and CWIC are applied The MCS of UE #1 and UE #2 is 16QAM (R = 0.51) and QPSK (R = 0.51), respectively For comparison, we also show the performance of ideal SIC The figure shows that CWIC can achieve almost the same performance as ideal SIC On the other hand, the required SNR of SLIC for achieving the BLER of 10-1 is increasing by 10.7 dB compared with ideal SIC due to the degradation of interference cancellation accuracy Figure 4(b) shows the BLER performance for the case of 2by-2 TM3 open-loop SU-MIMO The MCS of UE #1 and UE #2 is 16QAM (R = 0.49) and QPSK (R = 0.49), respectively R1:R2 is set to 1:1, 2:1, and 2:2, which are the major rank combinations selected in system-level evaluations [11] Note that we only show the performance of ideal SIC for the case of R1:R2 = 1:1 and 2:2 since the performance of R1:R2 = 2:1 is the same as that of 2:2 The figure shows that the performance of CWIC is almost the same as that of ideal SIC On the other hand, the required SNR of SLIC for achieving the BLER of 101 is increased by 11.1 dB compared with ideal SIC when R1:R2 = 1:1 This gap is approximately the same as the case of 1-by-2 in Fig 4(a) However, the required SNR of SLIC increases much more compared to ideal SIC for the case of R1:R2 = 2:1 and 2:2 This is due to the inter-stream interference introduced by multi-rank transmission Based on these results, hereafter 1164 IEEE ICC 2015 - Workshop on 5G & Beyond - Enabling Technologies and Applications V CONCLUSION for achieving BLER of 10-1 (dB) Required SNR of cell edge UE 30 UE #1: 16QAM UE #1: 64QAM R1:R2 = 1:1 R1:R2 = 2:1 R1:R2 = 2:2 25 20 15 UE #2: QPSK (R = 0.49) 10 -5 0.1 0.2 0.3 0.4 Power ratio of cell center UE (P1) 0.5 Fig Required SNR for achieving BLER of 10-1 (UE #2 performance) we focus on the CWIC receiver and clarify the influence of allocated power ratio Figure 5(a) shows the required SNR of UE #1 for achieving the BLER of 10-1 The simulation conditions are the same as Fig 4(b) The figure shows that the required SNR of CWIC compared to that of ideal SIC becomes larger from different P1 values according to R1:R2 For the case of R1:R2 = 1:1, which has no inter-stream interference for both UE signals, the required SNR for achieving the BLER of 10-1 is progressively increasing when P1 is greater than 0.35 When R1:R2 = 2:1, which has inter-stream interference for UE #1 signal, the required SNR is progressively increasing when P1 is greater than 0.4, which is slightly lower than the R1:R2 = 1:1 case This is because for the R1:R2 = 2:1 case, the variance of interference signal with rank = is lower than when rank = For the case of R1:R2 = 2:2, which has more severe inter-stream interference for both UE signals, the required SNR is progressively increasing when P1 is greater than 0.3 On the other hand, when applying R1:R2 = 2:1 and 2:2, and P1 = 0.05, the required SNR of CWIC and ideal SIC is increasing because of the residual interference due to the channel estimation error Figure 5(b) shows the required SNR of UE #1 for achieving the BLER of 10-1 when the MCS of UE #1 is 64QAM (R = 0.51) The result shows that the required SNR compared to that of Fig 5(a) is increased due to applying higher order modulation to UE #1 However, the performance gap between CWIC and ideal SIC in terms of required SNR is smaller compared to Fig 5(a) This is because the increase in the required SNR is mainly due to the use of higher order modulation for UE #1 On the other hand, when P1 is below 0.15, the required SNR of CWIC and ideal SIC is increased compared to Fig 5(a) The reason for this increase is that the influence of residual interference due to channel estimation error becomes larger with higher order modulation for UE #1 Figure shows the required SNR of UE #2 for achieving the BLER of 10-1 The MCS of UE #2 is QPSK (R = 0.49) The result shows that when P1 is larger, the required SNR is larger due to the inter-user interference Furthermore, the required SNR is larger when R2 = compared with R2 = due to the inter-stream interference On the other hand, the influence of different modulation scheme of UE #1 is not observed The design and the performance of SIC receiver for downlink NOMA combined with 2-by-2 open-loop SU-MIMO were investigated In particular, the performance of CWIC and SLIC are compared with ideal SIC with the received weight generation schemes applied being adapted according to the rank combinations of NOMA UEs and whether the detection is before or after SIC Assuming LTE Release specifications and the number of NOMA multiplexed users being two, we showed that the CWIC receiver achieves higher performance compared with SLIC receiver, and almost the same performance as ideal SIC when P1 is below 0.35, and the MCS of cell-center UE and cell-edge UE is 16QAM (R = 0.49) and QPSK (R = 0.49), respectively This is because the CWIC receiver is able to effectively mitigate both inter-user and interstream interference On the other hand, we showed that the required SNR of cell-edge UE is also impacted by the transmission rank of own signal and that of cell-center UE due to inter-user and inter-stream interference REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] 1165 M Tanno, Y Kishiyama, N Miki, K Higuchi, and M Sawahashi, “Evolved UTRA - physical layer overview,” in Proc IEEE SPAWC 2007, June 2007 S Sesia, I Toufik, and M Baker, “LTE – The UMTS Long Term Evolution from Theory to Practice,” WILEY, 2009 E Dahlman, S Parkvall, and J Sköld, “4G – LTE/LTE-Advanced for mobile broadband,” Academic Press, 2011 NTT DOCOMO, “Requirements, candidate solutions & technology roadmap for LTE Rel-12 onward,” 3GPP RWS-120010, June 2012 K Higuchi and Y Kishiyama, “Non-orthogonal access with successive interference cancellation for future radio access,” in Proc APWCS 2012, Aug 2012 N Otao, Y Kishiyama, and K Higuchi, “Performance of nonorthogonal access with SIC in cellular downlink using proportional fair-based resource allocation,” in Proc ISWCS, pp 476-480, Aug 2012 Y Saito, A Benjebbour, Y Kishiyama, and T Nakamura, “Systemlevel performance evaluation of downlink non-orthogonal multiple access (NOMA),” in Proc IEEE PIMRC 2013, Sept 2013 A Benjebbour, A Li, Y Saito, Y Kishiyama, A Harada, and T Nakamura, “System-level performance of downlink NOMA for future LTE enhancements,” in Proc IEEE GLOBECOM, Dec 2013 K Saito, A Harada, A Benjebbour, Y Kishiyama, and T Nakamura, “Performance evaluation of SIC receiver considering error vector magnitude for downlink NOMA,” IEICE RSC2014-163, vol 114, no 254, pp 43-48, Oct 2014 A Benjebbour, A Li, Y Kishiyama, J Huiling, and T Nakamura, “System-level evaluations of SU-MIMO combined with NOMA,” IEICE RCS2014-141, vol 114, no 180, pp 13-18, Aug 2014 Y Saito, A Benjebbour, Y Kishiyama, and T Nakamura, “A study on performance of downlink non-orthogonal multiple access (NOMA) in various environments,” IEICE RCS2014-162, vol 114, no 254, pp 37-42, Oct 2014 A Benjebbour, A Li, Y Kishiyama, H Jiang, and T Nakamura, “System-level performance of downlink NOMA combined with SUMIMO for future LTE enhancements,” in Proc IEEE GLOBECOM, Dec 2014 3GPP, TS 36.213 (V8.8.0) “Evolved Universal Terrestrial Radio Access (E-UTRA); Physical layer procedures (Release 8),” Sep 2009 3GPP, TS 36.211 (V8.9.0), “Evolved Universal Terrestrial Radio Access (E-UTRA); Physical Channels and Modulation (Release 8),” Dec 2009 ... ideal SIC for the case of R1:R2 = 1:1 and 2:2 since the performance of R1:R2 = 2:1 is the same as that of 2:2 The figure shows that the performance of CWIC is almost the same as that of ideal SIC. .. inter-stream interference On the other hand, the influence of different modulation scheme of UE #1 is not observed The design and the performance of SIC receiver for downlink NOMA combined with 2-by-2 open-loop... and the channel matrix as in (10) for the case of before SIC On the other hand, for the case A Simulation Assumptions Link-level simulations are conducted to evaluate the performance of the SIC