Decoding of uplink control information encoded with placeholders in long term evolution

Transcription

1 Journal of Chongqing University (English Edition) [ISSN ] Vol. 10 No. 3 September 2011 Article ID: (2011) To cite this article: WANG Dan, LIAO Yong, FAN Tong-liang. encoded with placeholders in long term evolution [J]. J Chongqing Univ: Eng Ed [ISSN ], 2011, 10(3): encoded with placeholders in long term evolution WANG Dan 1,, LIAO Yong 2, FAN Tong-liang 1 1 College of Communication Engineering, Chongqing University, Chongqing , P.R. China 2 Center of Communication and Tracking Telemetering Command, Chongqing University, Chongqing , P.R. China Received 22 February 2011; received in revised form 30 May 2011 Abstract: In 3GPP (the 3rd Generation Partnership Project) LTE (Long Term Evolution) systems, the physical uplink shared channel (PUSCH) conveys uplink control information (UCI) back to enodeb with or without UL-SCH (uplink shared channel) data. Placeholders are inserted into the UCI to scramble in a way that maximizes the Euclidean distance of modulation symbols. Considering the attribution of encoding with placeholders, a simple and efficient decoding scheme is proposed in this paper. As shown in our simulation results, improved performance is achieved. Keywords: decoding; placeholder; constellation; acknowledge/non-acknowledge (ACK/NACK); rank indication (RI); long term evolution (LTE); physical uplink shared channel (PUSCH) CLC number: TN911 Document code: A 1 Introduction a In LTE (long term evolution) uplink, a physical uplink shared channel (PUSCH) carries uplink control information (UCI) and uplink shared channel (UL-SCH) data. UCI at the coding unit is in form of channel quality information (CQI and/or PMI), rank indication (RI) and acknowledge/non-acknowledge (ACK/NACK), assisting enodeb to perform channel scheduling and resource allocation. When UCI is transmitted in the PUSCH, the channel coding for ACK/NACK, RI and CQI is done independently. In Ref. [1], ACK/NACK and RI are encoded with placeholders and scrambled in a way that maximizes the Euclidean distance of modulation symbols. Based on the encoding scheme, we proposed a simple and efficient scheme to decode the ACK/NACK and RI information in this work. Corresponding author, WANG Dan ( 王丹 ): swnucquwd@163.com. Funded by the Fundamental Research Funds for the Central Universities (XDJXS ). Conventionally, a general bit log-likelihood ratio (LLR) expression is provided for Gray coded rectangular quadrature amplitude modulation (R-QAM) signals. But the log MAP and max-log-map algorithms for LLR calculation are complex. To reduce the complexity of the bit metric calculation, several methods [2-10] have been proposed for Gray coded signals. Especially in Refs. [11-12], a simple arithmetic function is proposed to replace the mathematical max or min function of conventional LLR expression. All the above algorithms have contributed to improving performance or simplifying the demodulation for higher order QAM. Nevertheless the constellation size of ACK/NACK and RI information is limited to binary phase shift keying (BPSK) or quadrature phase shift keying (QPSK) and therefore they fail to perform well when decoding with conventional schemes. In this work, we built up a simple and efficient receiver scheme making use of the attribution of encoding. The scheme provides obvious performance gain in most fading channel situations over conventional method. Furthermore, it demodulates 1- bit or 2-bit information respectively as BPSK or QPSK 133

2 regardless of real modulation schemes, and thus reduces the complexity and is suitable for practical implementations. 2 Encoding of ACK/NACK and RI in physical uplink control channel (PUSCH) with placeholders RI and CQI/PMI are always reported in the same subframe in PUSCH aperiodically or periodically. The CQI/PMI is calculated given the simultaneously reported rank. The bit widths for RI feedback for PDSCH transmissions are 1 or 2, which are determined by the maximum number of layers according to the corresponding enodeb antenna configuration and user equipment (UE) category. For TDD, two HARQ (hybrid automatic repeat request)-ack feedback modes, i.e. HARQ-ACK bundling and HARQ-ACK multiplexing, are supported. For TDD HARQ-ACK bundling, HARQ-ACK consists of 1- or 2-bit information. For TDD HARQ-ACK multiplexing, HARQ-ACK consists of 1- to 4-bit information. For the case that HARQ-ACK consists of 1- or 2-bit information, placeholders are embedded into the encoded blocks. Therefore we only focus on the instance of 1- or 2-bit ACK/NACK in this paper. The 1- or 2-bit ACK/NACK or RI information, denoted by d 0 or d 1 respectively, is first encoded according to Table 1, [1,13] where d 2 =(d 0 +d 1 ) mod 2. In the table, x and y are placeholders. Then the encoded block is repeated multiple times and the resulting sequence is concatenated together to obtain Q ACK/RI bits, which is determined by the resource allocation for PUSCH. UCI and data may or may not be multiplexed on PUSCH. ACK/NACK and RI are mapped consecutively to reference symbols and CQI time-first mapping. ACK/NACK and RI resources are punctured into data starting from the bottom of the resource grid and predetermined locations next to the demodulation reference signal (DMRS) as shown in Fig. 1, whereas CQI resources are placed at the beginning of the data resources. ACK/NACK and RI occupy less than 4 single-carrier frequency-division multiplexing access (SC-FDMA) symbols respectively. [14] Multiplexing and interleaving of UCI and data in conjunction with the resource element mapping for PUSCH can implement the mapping scheme above. After that, the block of bits b(0),,b(m bit -1), where M bit is the number of bits, shall be scrambled with a UE-specific scrambling sequence prior to modulation, resulting in a block of scrambled bits ~ b ~ (0),..., b ( M bit 1). [15] Meanwhile, the position of x is replaced by 1 and y by the previous bit. Considering all of the above processes, the sequence carrying ACK/NACK or RI information may turn to be multiple repeated blocks as Table 2. Then the sequence of scrambled bits shall be modulated with QPSK, 16QAM or 64QAM configured by higher layers. Obviously, encoding with placeholders maximizes the Euclidean distance of the modulation symbols carrying ACK/NACK or RI information. According to the modulation mapper defined in Ref. [15], Figs. 2 to 4 illustrate the possible constellation points of ACK/NACK or RI information in comparison with the conventional points. As shown in Figs. 2 to 4, the possible constellation points scatter at the edge of normal points. As a result, the modulation of 1- or 2-bit ACK/NACK or RI information enables limiting the constellation size to BPSK or QPSK, respectively. 3 Receiver process of ACK/NACK and RI with placeholders Ref. [1] does not specify the decoding of ACK/NACK and RI in receivers. As is shown in Figs. 2 to 4, if demodulation is done in a usual way which neglects the attribution of encoding with placeholders, some constellation points may be judged to be the nomapper points in transmitters. Consequently, a higher bit-error ratio (BER) may be produced. According to Section 2, the demodulation of information with placeholders is only related to the phase of a received signal but not to both the phase and amplitude in normal 16QAM or 64QAM. Therefore, the receiver of ACK/NACK and RI with placeholders can concern with only the number of input bits but not the modulation scheme. In the case of 1-bit, demodulation can be processed as BPSK, whereas the 2-bit case corresponds to QPSK. Obviously, this simplifies the demodulation of 16QAM or 64QAM to that of BPSK or QPSK. What s important is that a lower BER can be obtained. The process of decoding is as follows. Step 1. Demodulate the received sequence which can be denoted by r 0, r 1,,r Msym, where M sym is the number of symbols given by M sym =M bit /Q m, where Q m is 2, 4 or 6 for QPSK, 16QAM or 64QAM, respectively. 134 J. Chongqing Univ. Eng. Ed. [ISSN ], 2011, 10(3):

3 Modulation scheme Table 1 Encoding of 1-bit and 2-bit ACK/NACK or RI 1-bit ACK/NACK or RI Encoded HARQ-ACK/RI 2-bit ACK/NACK or RI QPSK [ d0 y ] [ d0 d1 d2 d0 d1 d 2] 16QAM [ d0 yxx ] [ d0 d1 xx d2 d0 xxd1 d2 xx ] 64QAM [ d0 yxxxx ] [ d0 d1 xxxx d2 d0 xxxx d1 d2 xxxx ] Notes: ACK, acknowledge; NACK, non-acknowledge; RI, rank indication; HARQ, Hybrid automatic repeat request; QPSK, quadrature phase shift keying; QAM, quadrature amplitude modulation; d 0, 1-bit ACK/NACK or RI information; d 1, 2-bit ACK/NACK or RI information; d 2, (d 0+d 1)mod 2; x and y, placeholders. Table 2 Scrambling of 1-bit and 2-bit ACK/NACK or RI Encoded HARQ-ACK/RI Modulation scheme 1-bit ACK/NACK or RI 2-bit ACK/NACK or RI QPSK [ d 0 d 0] [ d d d d d d ] QAM [ d 0 d 0 1 1] [ d d 1 1 d d 1 1 d d ] 64QAM [ d 0 d ] [ d d d d d d ] Notes: ACK, acknowledge; NACK, non-acknowledge; RI, rank indication; HARQ, Hybrid automatic repeat request ; QPSK, quadrature phase shift keying; QAM, quadrature amplitude modulation; d 0, 1-bit ACK/NACK or RI information; d 1, 2-bit ACK/NACK or RI information; d 2, (d 0+d 1)mod 2; x and y, placeholders. Fig. 1 Resource mapping for acknowledge/non-acknowledge (ACK/NACK) and rank indication (RI) Fig. 2 Quadrature Phase Shift Keying constellation points J. Chongqing Univ. Eng. Ed. [ISSN ], 2011, 10(3):

4 Fig. 3 16QAM constellation points, where UCI is uplink control information, and QAM is quadrature amplitude modulation Fig. 4 64QAM constellation points, where UCI is uplink control information, and QAM is quadrature amplitude modulation j /4 Firstly, the sequence is multiplied by e π. j /4 ri () = ri () e π, i= 0,1, Msym Secondly, the decoded bit stream is determined with the phase angle of soft symbols as follows. For 1-bit information, the decoded bit di () is defined by 1, di ang( ri ( )) ( π/ 4,3 π/ 4) () = 0, others, i= Msym 0,1,. Then, the demodulated sequence can be obtained as [ di ( ), di ( )] for QPSK, [ b ( j),..., b ( j+ Qm 1)] = [ di ( ), di ( ),1,1] for 16QAM, [ di ( ), di ( ),1,1,1,1] for 64QAM, i j= 1,2, Q,..., M / Q. = 0,1, Msym, m bit m For 2-bit information, the decoded bits di () is defined by [0,0], ang( ri ( )) ( π/ 4, π/ 4), [1, 0], di ang( ri ( )) π ( / 4,3 π/ 4), () = i= 0,1, Msym. [1,1], ang( ri ( )) (3 π/ 4, π 3 / 4), [0,1], ang( ri ( )) ( 3 π/ 4, π/ 4), Note that di () stands for 2 bits here. Then the demodulated sequence can be expressed as [ di ( )] for QPSK, [ b ( j),... b ( j+ Qm 1)] = [ d ( i),1,1] for 16 QAM, [ di ( ),1,1,1,1] for 64 QAM, i j= 1,2, Q,..., M / Q. = 0,1, Msym, m bit m Obviously, the sequence b ( j) are the demodulated bits of possible constellation points in Figs. 2 to 4. Step 2. Descramble the demodulated sequence. The scrambling sequence c(i) and the positions of placeholders in the decoded sequence are known in a receiver unit. Therefore, the descrambled sequence b ˆ(0),..., b ˆ( M bit 1) can be obtained according to the following pseudo code. Set i = 0. while i< Mbit, if i is the position of x bi ˆ( ) = bi () else if i is the position of y bi ˆ( ) = bi () + ci ( 1) mod2 ( ) else b ˆ( i) = ( b ~ ( i) + c( i) ) mod2. end if end if i = i J. Chongqing Univ. Eng. Ed. [ISSN ], 2011, 10(3):

5 end while 4 Simulation and results In this section, link-level simulation results of the proposed algorithm are presented in comparison with the results of demodulating with max-log-map algorithm which disregards for the attribution of encoding with placeholders. The simulation model consisted of a single cell with one UE and an enodeb. Bit error ratios (BER) of ACK/NACK and RI were measured. The total bandwidth considered was B=10 MHz at 2 GHz, subdivided into 50 resource blocks of 12 subcarriers each. Extended typical urban channel model (ETU) [16] with a maximum Doppler frequency of 5 Hz scenario was studied. 4.1 Performance results of 1-bit ACK/NACK Simulation results are shown in Fig. 5 for 1-bit ACK/NACK decoded by the proposed algorithm and conventional algorithm. The BER was studied with the input sequence of scrambling and the output sequence of descrambling. placeholders, the performance of QPSK, 16QAM and 64QAM is similar as BPSK with proposed algorithm and it provides a gain of around 2.2 db over QPSK with conventional algorithm. Although a larger constellation distance is gained by encoding processing, the conventional algorithm demodulates the constellation size of BPSK as it does for QPSK, 16QAM or 64QAM. It may bring in some error detection to the no-mapper points in Figs. 2 to 4 and increase much decoding complexity. As a result, it performs worse than the proposed algorithm. Comparing Figs. 5 and 6 indicates that the proposed algorithm provides a similar improvement over conventional algorithm. Meanwhile, the performance of QPSK, 16QAM and 64QAM is almost identical in the proposed algorithm, and hence it s also the same with QPSK in conventional algorithm. Additionally, the proposed algorithm for RI information provides an improvement over conventional algorithm for ACK/NACK. However, both algorithms perform worse for RI than for ACK/NACK, because the resource mapping of RI is farther from reference signal than of ACK/NACK BER BER SNR/dB Fig. 5 BER (bit error ratio) versus SNR (signal noise ration) for 1-bit ACK/NACK (acknowledge/non-acknowledge) 4.2 Performance results of 2-bit ACK/NACK Simulation results of 2-bit ACK/NACK decoded with proposed algorithm and conventional algorithm are shown in Fig. 6. For the 1-bit case, due to the limitation of the constellation size to BPSK by encoding with SNR/dB Fig. 6 BER (bit error ratio) versus SNR (signal noise ration) for 2-bit ACK/NACK (acknowledge/non-acknowledge) 5 Conclusions In this paper, the receiver of 1- or 2-bit ACK/NACK and RI information on PUSCH is discussed. An efficient and simple receiver scheme is also proposed. The essence of this scheme is to utilize encoding with placeholders and scrambling in a special way which J. Chongqing Univ. Eng. Ed. [ISSN ], 2011, 10(3):

6 maximizes the Euclidean distance of the modulation symbols. Therefore, we can reduce the complication of the demodulation for 16QAM or 64QAM down to BPSK or QPSK. What s more, simulation results show that the proposed scheme has improved BER performance obviously. References [1] 3GPP Organizational Partners. 3GPP TS V9.2.0 ( ) 3rd Generation Partnership Project; Technical Specification Group Radio Access Network; Evolved Universal Terrestrial Radio Access (E-UTRA); Multiplexing and channel coding (Release 9) [S]. Valbonne, France: 3GPP Organizational Partners Publications Offices, [2] Kwon JK, Park S, Sung DK. Log-likelihood ratio (LLR) conversion schemes in orthogonal code hopping multiplexing [J]. IEEE Communications Letters [ISSN ], 2003, 7(3): [3] Steingrimsson B, Luo ZQ, Wong KM. Soft quasimaximum-likelihood detection for multiple-antenna wireless channels [J]. IEEE Transactions on Signal Processing [ISSN X], 2003, 51(11): [4] Dai XM. Low complexity maximum-likelihood based QRD-M for MIMO systems with S-QAM or APSK [C]. In: ICC, Cape Town, South Africa, May 23-27, [S.l.: s.n.], 2010: 1-6. [5] Yuan JH, Vucetic B, Feng W. Turbo-coded M-QAM for fading channels [J]. IEEE Electronics Letters [ISSN ], 2000, 36(18): [6] Pyndiah RM. Near-optimum decoding of product codes: block turbo codes [J]. IEEE Transactions on Communications [ISSN ], 1998, 46(8): [7] Pyndiah R, Picart A, Glavieux A. Performance of block turbo coded 16-QAM and 64-QAM modulations [C]. In: The Global Telecommunications Conference (GLOBECOM), Singapore, November 13-17, [S.l.]: IEEE, 1995: [8] Steingrimsson B, Luo ZQ, Wong KM. Quasi-ML detectors with soft output and low complexity for PSK modulated MIMO channels [C]. In: 4th IEEE Workshop on Signal Processing Advances in Wireless Communications, Rome, Italy, June 15-18, [S.l.]: IEEE, 2003: [9] Le Goff SL, Glavieux A, Berrou C. Turbo-codes and high spectral efficiency modulation [C]. In: 1994 IEEE International Conference on Communications, Proceedings of the (ICC '94), New Orleans, Louisiana, May 1-5, [S.l.]: IEEE, 1994: [10] Wang CC. A bandwidth-efficient binary turbo coded waveform using QAM signaling [C]. In: 2002 International Conference on Communications, Circuits and Systems and West Sin0 Exposition Proceedings, Chengdu, China, June 29-July 1, [S.l.]: IEEE, 2002: [11] Kim KS, Hyun K, Yu CW, et al. General loglikelihood ratio expression and its implementation algorithm for gray-coded QAM signals [J]. ETRI Journal [ISSN ], 2006, 28(3): [12] Lu HF, Chen W, Huang QY. Simplified soft decision algorithm for OFDM system [J]. Journal of Wuhan University of Technology: Transportation Science & Engineering [ISSN ], 2009, 33(1): (In Chinese). 卢海风, 陈伟, 黄秋元. 一种应用于 OFDM 通信系统的简化软判决算法 [J]. 武汉理工大学学报 : 交通科学与工程版 [ISSN ],2009,33(1): [13] Qualcomm Europe. Update to incorporation decisions from RAN1 #51 bis and RAN1 #52, R [R]. Sorrento, California: Qualcomm Europe, [14] Texas Instruments. Coding of control information on PUSCH, R [R]. Shenzhen: Texas Instruments, [15] 3GPP Organizational Partners. 3GPP TS v Physical channels and modulation (Release 9) [S]. Valbonne, France: 3GPP Organizational Partners, [16] 3GPP Organizational Partners. 3GPP TS v Base station (BS) radio transmission and reception (Release 9) [S]. Valbonne, France: 3GPP Organizational Partners, Edited by LUO Min 138 J. Chongqing Univ. Eng. Ed. [ISSN ], 2011, 10(3):