IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 35, NO. 8, AUGUST 2017 1867 Spectral Efficiency Improvement With 5G Technologies: Results From Field Tests Jian Wang, Member, IEEE, Aixiang Jin, Dai Shi, Lei Wang, Hui Shen, Member, IEEE, Dan Wu, Liang Hu, Liang Gu, Lei Lu, Yan Chen, Jun Wang, Yuya Saito, Anass Benjebbour, Senior Member, IEEE, and Yoshihisa Kishiyama Abstract Spectral efficiency is always a key factor to be improved and optimized along mobile communication networks evolving generation by generation. 5G enabling technologies must take spectral efficiency into consideration. In this paper, we show the performance of three key 5G technologies in sense of spectral efficiency improvement. Sparse code multiple access, polar codes, and filtered orthogonal frequency-division multiplexing are novel multiple access technology, channel coding scheme, and waveform, respectively. The combination of them is implemented in a 5G field trial testbed by NTT DOCOMO and Huawei for the first time. According to the field test results, we achieve over 100% spectral efficiency improvement comparison with baseline, where orthogonal frequency-division multiple access and turbo coding as LTE are used. Index Terms 5G, spectral efficiency, SCMA, polar code, f-ofdm, field test. I. INTRODUCTION THE fifth generation (5G) cellular communications have drawn lots of attentions from both academic and industrial fields. Among all the critical requirements such as high peak throughput, ultra-low latency and large device density, etc., a superior spectral efficiency (SE) is also emphasized [1], [2]. Obviously, the radio spectrum resource is of fundamental importance for wireless communications, especially when considering the fact that most of the available spectrum has been allocated to existing wireless communication systems. Inheriting the requirements from the whole 5G system, the enabling technologies for 5G should take SE improvement into consideration. Breakthroughs in baseband and RF architecture, advanced RF domain processing and network architecture are necessary for the development of 5G [3]. Besides, new physical layer technologies are also called for, where multiple access scheme, channel coding scheme and waveform are three key aspects. Multiple access schemes are milestones for different generation wireless systems from 1G to 4G. Frequency- Manuscript received December 18, 2016; revised April 1, 2017; accepted May 5, 2017. Date of publication June 8, 2017; date of current version July 15, 2017. (Corresponding author: Jian Wang.) J. Wang, A. Jin, D. Shi, L. Wang, H. Shen, D. Wu, L. Hu, L. Gu, L. Lu, Y. Chen, and J. Wang are with Huawei Technologies Co., Ltd., Shenzhen 310051, China (e-mail: wangjian23@huawei.com; jinaixiang@huawei. com; sd11102001@huawei.com; wanglei888@huawei.com; henry.shenhui@ huawei.com; tony.wudan@huawei.com; huliang.hu@huawei.com; albert. guliang@huawei.com; kevin.lu@huawei.com; bigbird.chenyan@huawei.com; justin.wangjung@huawei.com). Y. Saito, A. Benjebbour, and Y. Kishiyama are with the 5G Laboratory, NTT DOCOMO Inc., Tokyo 239-8536, Japan (e-mail: yuuya.saitou.fa@ nttdocomo.com; benjebbour@nttdocomo.com; kishiyamag@nttdocomo.com). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JSAC.2017.2713498 division multiple access (FDMA) for 1G, time-division multiple access (TDMA) for 2G, code-division multiple access (CDMA) for 3G and orthogonal frequency-division multiple access (OFDMA) for 4G are all orthogonal multiple access (OMA) schemes. Wireless radio resources are divided orthogonally and allocated to different users. Data traffic, device density and service diversity are expected to have a huge increasing for 5G. To satisfy these requirements, several types of non-orthogonal multiple access (NOMA) schemes are proposed, while [4] [6] make good overviews on NOMA. Instead of not allowing interference between users as in OMA, NOMA allows controllable interferences by non-orthogonal resource allocation, while the interference impacts are alleviated with an affordable increasing in receiver complexity. Most of the NOMA schemes can be grouped into two categories: power domain multiplexing with successive interference cancellation (SIC) receiver and code domain multiplexing with message passing algorithm (MPA) receiver. As one of the NOMA schemes, a novel sparse code multiple access (SCMA) is firstly introduced in [7] and [8]. The term sparse code comes from the fact that by using SCMA, multiple user equipments (UEs) share the same time-frequency resources through mapping their data onto different sparse codewords. All codewords in the same codebook (or say SCMA layer) contain zeros in the same position, and layers are distinguished with each other through different positions of zeros. The key benefit of SCMA is that, besides overloading as other NOMA schemes, a multi-dimensional constellation is designed to bring the coding gain and shaping gain. With employing MPA at the receiver side, data from different UEs are separated and reconstructed. Overloading together with extra gains helps SCMA to provide high throughput with given spectrum bandwidth. After Shannon s information theory being established [9], channel coding schemes began to play an important role in communication systems. Linear codes were first developed to deal with errors made by the channel based on the binary symmetric channel (BSC) assumption. At the same time, convolutional codes with trellis-based decoding methods lead to a different way, which are adoptable to any channels. Then the error correction coding becomes a more general concept channel coding. Although Shannon proved that the error-free transmission can be achieved for every noisy channel through channel coding with any code rate up to the channel capacity, he did not elaborate how to construct capacity-achieving codes. Turbo codes introduced by Berrou et al. [10] and LDPC codes introduced 0733-8716 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1868 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 35, NO. 8, AUGUST 2017 by R. G. Gallager in his doctoral dissertation [11] and rediscovered by MacKay and Neal [12] are achievements for finding codes approaching channel capacity. Readers are suggested to refer to [13] for more information about channel coding. Besides all the above mentioned codes, Polar codes are the first proven capacity-achieving channel coding scheme, which has good coding gain without any error floor [14] [16]. Compared to LTE Turbo, UEs who use Polar can obtain higher successful decoding rate under the same modulation coding scheme (MCS) level, or even use a higher MCS level under the same signal to noise ratio (SNR). As a result, throughput and spectral efficiency performance can be improved. It is well known that, in LTE, 10% of the allocated bandwidth is reserved as guard band to meet the spectrum mask and adjacent channel leakage ratio (ALCR) requirement. To alleviate this waste of frequency resource, and meet the spectrum mask and ALCR requirement simultaneously, out-of-band emission (OOBE) suppressing method is needed. Meanwhile, the existing OFDM waveform requires synchronization which calls for much signaling coordination. Aiming at reducing OOBE and/or relaxing the requirement on synchronization, quite a lot of novel waveforms are proposed for 5G, including generalized frequency division multiplexing (GFDM) [17], filter bank multi-carrier (FBMC) [18], universal filtered multicarrier (UFMC) [19] and filtered-ofdm (f-ofdm) [20] [22]. Comparisons between these waveforms have been done in [22], [23], where it is found that f-ofdm appears as the most promising waveform contender for 5G. It maintains the advantages of OFDM on high multiplexing flexibility and low equalization complexity. Moreover, through careful and smart filter design, low OOBE, relaxed synchronization, affordable complexity and backward and forward compatibility can be expected [22], [23]. With low OOBE, desirable frequency localization can be achieved to take the former unused guard spectrum band into use, without influencing the central spectrum band and the neighbour band. This additional spectrum band can provide extra throughput, thus results into a spectral efficiency improvement. In this paper, we implement the above mentioned technologies into our 5G field test platform and conduct several trials to evaluate the performance of them in sense of spectral efficiency improvement. The LTE settings, i.e. OFDMA, Turbo codes and no filter, are taken as the baseline. Same definitions as LTE are used in this paper for the sake of easy understanding. After the baseline runs stably, we turn on SCMA, Polar codes and f-ofdm one by one to see how much each technology impacts on the overall system spectral efficiency. The results show that, with the help of these technologies, the system spectral efficiency can be improved more than 100% compared to the baseline, where SCMA provides more than 90% gain compared to OFDMA for the case investigated, where large SNR difference exists between the UEs sharing the same time-frequency resources. The gain is obtained through optimization on SCMA codebook design, layer allocation and power allocation. Fig. 1. Fig. 2. Factor graph and signature matrix of a 6 4 codebook. SCMA multiplexing. Polar coding provides extra gain, which is around 10%. The gain mainly comes from larger coding gain compared to other coding scheme. Through using f-ofdm, another 7% extra gain is obtained stably, since 10 more resource blocks can be used. Meanwhile, by using f-ofdm, we found that interference from neighbour spectrum band results in limited impacts. This paper is organized as follows. In section II, we give a brief description on the involved technologies. In section III, the field test results are shown to verify their performance. And finally, we make some conclusion in Section IV. II. THE BACKGROUND OF ENABLING TECHNOLOGIES In this section, the basic concepts of the three enabling technologies are introduced briefly. More details can be found in the references mentioned below. A. SCMA The signature of SCMA codebooks can be represented by a factor graph G(V, N) with V variable nodes (VNs) and N function nodes (FNs) [7], [8]. The VNs represent data layers, and the FNs represent time-frequency resources share by data layers. Fig. 1 gives an example of a factor graph with 6 VNs and 4 FNs, which is a 6 4 codebook or 4-point codebook, and its signature matrix. For each data layer, every P = log 2 M bits are mapped to a N-dimensional SCMA complex codebook of size M. Each codeword of a codebook is a sparse vector with K < N non-zero entries. Each UE can use one or multiple data layers. Denote the bits mapped on the v-th data layer as a v = (a v,1, a v,2,...,a v,p ), then (a 1, a 2,...,a V ) is the coded bits of all the V data layers. a v is mapped to a SCMA codeword x v = (x v,1, x v,2,...,x v,n ), wherein only K entries of x v are non-zero.
WANG et al.: SPECTRAL EFFICIENCY IMPROVEMENT WITH 5G TECHNOLOGIES 1869 Fig. 3. Diagram of channel polarization. Time-frequency resources are shared among SCMA data layers of multiple active users. For examples, with the 4-point codebook, six data layers share four resource elements as shown in Fig. 2, which means a 150% overloading. Due to the sparsity feature of SCMA codewords, MPA can be employed to provide a moderate complexity multi-user detection algorithm [7]. Multiple dimensions of optimization can be considered in SCMA systems. Codebook can be designed to get higher coding gain. Design of rotation degrees of constellations in each codebook can help with fast converging of MPA decoding. Power allocation also helps the MPA decoder to distinguish codewords from each other, which in turn helps to speed up the decoding and improve the performance. Overloading and extra gains (e.g., coding gain) help SCMA to improve the throughput of the system. B. Polar Coding Polar codes are a set of error-correcting codes that utilize the channel polarization phenomenon [14], which is shown in Fig. 3 and can be described as follows. The same independent channel W can be transformed into two types of synthesized channels W + and W with slightly different capacities (I (W+) and I (W )) or say. The one with higher reliabilities are regarded as good channels and other ones are bad channels. Applying this polarization transformation recursively on the resulting channels will make the good channels better and better. According to this operation, we can transform multiple independent uses of a given binary memoryless symmetric (BMS) channel into a set of successive uses of synthesized binary input channels with different reliabilities. For example, for a binary erasure channel (BEC) W with the erasure probability ɛ = 0.5. Repeating the polarization operation for 10 times, we got the 1024 channels. Their symmetric capacities can be calculated through the recursive relations I (W (2i 1) N ) = I (W (I ) N/2 )2, I (W (2i) (I ) (I ) N ) = 2I (W N/2 ) I (W N/2 )2, (1) with I (W (1) ( j) 1 ) = 1 ɛ = 0.5, where I (W N ) denotes the symmetric capacity of the j-th channel in all the N synthesized channels. Fig. 4 shows the symmetric capacities of these 1024 channels. Then, transmitting free bits (information bits) over the good channels and fixed bits (frozen bits) on the bad ones is the basic idea of Polar codes [15]. Fig. 4. Plot of symmetric capacities for a binary erasure channel (BEC) with ɛ = 0.5. By mathematical representation, polar codes are based on [ ] n choosing a set of 2 n 1 0 R rows of the matrix G n = 1 1 to form a 2 n R 2 n matrix which is used as the generator matrix in the encoding procedure. The mother code length is limited to be 2 n, while puncturing or shortening method can help to generate codes with required length. Successive cancellation (SC) is the basic decoding algorithm for Polar codes, based on which many kinds of decoders are invented, e.g., SC list (SCL), SC stack (SCS) and CRCaided SCL (CA-SCL) [16]. To find a good tradeoff between performance and complexity, we can choose different decoders with different list/stack sizes. Simulations show that Polar coding has better BLER performance than LTE Turbo. With using the same MCS level, Polar helps users to have higher successful decoding rate than those using LTE Turbo under the same channel condition [16]. Moreover, with adaptive modulation and coding (AMC) scheme on, users with Polar coding may even use a higher MCS level. Throughput of the system can be improved through replacing LTE Turbo by Polar. Meanwhile, As Polar coding is superior to other existing coding schemes, it brings extra benefit, especially for edge users. Polar has a lower required SINR when calculating the maximum coupling loss (MCL), which means besides throughput, it also provides better coverage. C. f-ofdm f-ofdm is a novel waveform framework that enables spectrum slicing and coexistence of multiple subbands [21], [22]. As shown in Fig. 5, K band-limited filters are employed on the baseband OFDM signals to reduce the interference between adjacent subbands. So that the OOBE is suppressed and the interference from adjacent subbands is restrained to a low enough level. The numerology for each of such subbands
1870 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 35, NO. 8, AUGUST 2017 Fig. 5. Diagram of f-ofdm systems [24]. Fig. 6. Field test environment. can be configured to meet the requirement according to the application scenario. In LTE specifications, 10% of system bandwidth is reserved as guard band to meet the adjacent channel leakage ratio (ACLR) and spectrum mask requirement. By applying f-ofdm, the baseband OFDM signal can be shaped with ultra-narrow transition region and meets the OOBE requirement of LTE spectrum mask, thus the guard band can be re-utilized to transmit useful signals. The filter used in f-ofdm should be carefully designed to achieve a tradeoff between time and frequency localization. Usually, a windowed-sinc method is used, where the infinite impulse response of a Sinc function is multiplied by a finite time domain window, leading to a FIR filter. The time domain window can be chosen from the existing welldefined windows, such as rooted raised cosine (RRC) window and Hanning window. Wu et al. [24] have evaluated the performance of RRC windowed FIR filter, Hanning windowed FIR filter and Remez FIR filter. The results show that RRC window provides a better trade-off between time and frequency localization due to its freedom in the roll-off factor α selection. The length of the window is also a parameter to design, since a long window has better localization performance may cause higher computational complexity and longer processing latency. By implementing f-ofdm, 10 more resource blocks (RBs) can be exploited in a 20 MHz system. Meanwhile, by designing the filter carefully, interference from the neighbour spectrum band can be omitted. III. FIELD TEST RESULTS AND ANALYSIS As mentioned above, with the available spectrum bandwidth fixed, improvement of SE is equivalent to increment of throughput. We set up a field test environment to investigate how SCMA, Polar codes and f-ofdm can help to increase the total throughput. The number of UEs in the test is fixed and the total throughput of all the UEs is recorded. In this section, we show the field test settings, procedures and results, respectively. Then we give some discussions on the reasons of the gain. A. Field Test Environment and Settings We consider a one-cell scenario in this field test, where three UEs are scheduled by one base station (BS). As shown in Fig. 6, the antennas of the BS are located on the 108 meters high tower on top of the Yokomaha Docomo building. Each UE is installed into one car, while the UE antennas are on the top of the car with a height of about three meters. One of the three UEs is about 300 meters away from the BS, hence has a relatively high SNR (around 21dB). The distance between other two UEs and the BS are about 850 meters, and the measured SNRs of them are about 1-3dB. The positions of the BS and UEs are also given by the map in Fig. 6. Generally speaking, all the three UEs belong to LOS scenario. The difference in channel quality, or say SNR, comes from their separation distances from the BS. For the sake of convenience, we call the UE with higher SNR as central UE and the other two UEs with lower SNR as edge UEs. The downlink traffic is measured in the field test, while uplink is only used to feedback and only occupies the central 18MHz. The parameters are listed in Table I, which are similar to those in LTE with 20MHz system bandwidth [25]. Note that there are two CP lengths, 5.2us for 0th, 1st, 2nd and 6th symbol in one slot, and 4.17us for the 3rd, 4th and 5th symbol in one slot. A RRC windowed-sinc filter with parametersshownintableiisusedforbothtxfilterand Rx filter. B. Field Test Cases Design In this field test, the baseline system uses OFDMA and LTE Turbo as its multiple access and channel coding scheme, respectively. To avoid interference to the neighbour band, only the central 18MHz, which means 100 RBs are allocated among the 3UEs. The central UE uses 34 RBs while each of the edge UE uses 33 RBs, as shown in Fig. 7. HARQ is not enabled in this trial. After the baseline is running stably, the above mentioned three technologies are turned on one by one to see the throughput gain, which is averaged over time:
WANG et al.: SPECTRAL EFFICIENCY IMPROVEMENT WITH 5G TECHNOLOGIES 1871 TABLE I FIELD TEST PARAMETERS Fig. 7. Resource allocation diagram. 1) Turn on SCMA, and still use Turbo and 100 RBs. All the UEs use the full band, the multiplexing of UEs are done through SCMA data layers allocation among UEs, as shown in Fig. 7. A 6 4 codebook is used in the trial, where there are six data layers in total. The central UE uses four of them and the edge UEs use one layer each. 2) Turn on Polar, and still use SCMA and 100 RBs. All the resource allocation settings are kept unchanged. Then comparing the resulting throughput with the one obtained in Step 1, we get the gain from using Polar coding. 3) Turn on filter, use the guard band while SCMA and Polar are still on. Now, 110 RBs can be used to transmit data without interference on the neighbour band, as shown in Fig. 7. The extra 10 RBs bring additional throughput. 4) Add neighbour band interference, to see if the system is stable while neighbour band interference 1 exists. With the guard band being used, the frequency spacing between the working band and its neighbour band is only 15kHz (one subcarrier), we should make sure the neighbour band interference does not degrade the total throughput much. To support AMC, new MCS levels and corresponding transmission block (TB) sizes are designed and shown in Table II. The design principle is similar to that of LTE. MCS levels above 16 for the one SCMA layer case do not exist since the code rate in those levels are larger than 1. Since, the bit-tosymbol mapping is done by SCMA encoding in SCMA cases, no modulation order is provided for those cases. TB is divided into code blocks (CBs) according to the method used in LTE, which is designed for LTE Turbo codes. Test cases for both Turbo and Polar follow this same MCS table in the trial. A reference power allocation scheme is searched carefully through simulations first, based on which a manual adjustment is performed during the field trials. In the simulations for searching the reference power allocation scheme, we fix the power of two SCMA layers of the central UE first, and then search good values for the other four SCMA layers (two for the central UE and two for the two edge UEs). To limit the number of parameters to be optimized, we set the powers of the two unfixed SCMA layers for the central UE to be equal to each other. Moreover, the two edge UEs experience a similar SNR in the trial, hence the powers of the two SCMA layers used by these two edge UEs are also set to be the same. Denoting the powers of the four SCMA layers for the central UE as [p 1, p 2, p 3, p 4 ] and those for the two SCMA layers for the two edge UEs as [p 5, p 6 ], the searching is performed as shown by the pseudo-code given in Algorithm 1. After finding a proper power allocation scheme, which means the throughput of edge UE does not degrade compared to the baseline and the throughput of the central UE is maximized, we keep on using it throughout the whole trials. The throughputs of all the three UEs are measured and recorded in each trial step, together with the corresponding MCS indices, code rates, block error ratios (BLERs) and SNRs. Each trial step lasts for 5 to 10 minutes. Finally, the average values of the above mentioned parameters are calculated for each durations. 2 Since the advantage of Polar codes over Turbo codes is widely validated in the references [16], and the extra 10% resource can be exploited through using f-ofdm is quite obvious, we only simulated the SCMA system and OFDMA system to see how much gain we can expect in the trial. Same to the settings of field trials, one central UE with high SNR and two edge UEs with low SNR are considered in the simulation. According to the pre-test in the field, around 20dB 1 An OFDM signal occupied the whole 20MHz neighbour band are used as the interference. The subcarrier spacing is 30kHz, which is different from that used in the working band. For sake of simplicity, only one interferer is added in the low-frequency side neighbour band. 2 Due to AMC, MCS index is changing during the trial according to the channel condition. We put the one chosen for the most times into the resulting table.
1872 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 35, NO. 8, AUGUST 2017 TABLE II MCS TABLE FOR SCMA SYSTEMS Algorithm 1 Power Allocation Scheme 1: Initial power allocation: p 1 = 1;p 2 = 1; 2: Initial maximum throughput: T max = 0; 3: Initial adjust step: 1 = 2 = 0.1; 4: for i = 0; i <= 15; i = i + 1 do 5: x = 10 i/10 ; 6: p 3 = x; 7: p 4 = x; 8: for j = 0; j <= 15; j = j + 2 do 9: y = 10 j/10 ; 10: p 5 = y; 11: p 6 = y; 12: Calculate the total throughput T tmp ; 13: if T tmp > T max then 14: T max = T tmp ; 15: x opt = x; 16: y opt = y; 17: end if 18: end for 19: end for 20: p 3 = x opt ;p 4 = x opt ;p 5 = y opt ;p 6 = y opt ; 21: Normalize [p 1, p 2, p 3, p 4, p 5, p 6 ] Fig. 8. Resource allocation diagram. get, the AWGN channel is used for the central UE and its SNR is increased to 30dB. As the results shown in the second line of Table III, a gain of 107% is obtained. gap between the high and low SNRs can be expected, hence SNR gap is set to be 20dB in our simulation. The EPA channel model is used, and the results are shown in the first line of Table III. The gain of SCMA over OFDMA is about 63% in this case. Furthermore, to probe the maximum gain we can C. Field Test Results and Discussions The test results are shown in Table IV. For the sake of clearness, the throughputs of all UEs and the total throughput are drawn in Fig. 8. With all the three UEs fixed in their positions, the SNRs are nearly the same during all the tests as shown in Table IV.
WANG et al.: SPECTRAL EFFICIENCY IMPROVEMENT WITH 5G TECHNOLOGIES 1873 TABLE III SIMULATION RESULTS TABLE IV FIELD TEST RESULTS From the results, we can observe that: SCMA related gain: After turning on SCMA, the total throughput of the system achieves more than 90% gain compared to the baseline. All the three UEs have larger throughputs with using SCMA, where the central UE contributes the most for the total throughput gain. Compared to the baseline, where RBs and power are allocated equally among all the three UEs, better resource (data layers) allocation and power allocation scheme are adopted in the SCMA case. Meanwhile, coding gain and shaping gain of SCMA also help to enlarge the total throughput. 91.11% gain is obtained in this trial, which falls in the range of expected results shown in Table III. Polar related gain: Replacing Turbo by Polar, the total throughput becomes 59.14Mbps, which is more than 110% of that of the baseline. From the results, we know that in this trial, Polar codes provide extra (59.14 53.32)/53.32 = 10.92% gain. Compared to the SCMA and Turbo case, after turning on Polar codes while keeping other settings unchanged, the MCS levels of all the three UEs increase one or two levels, i.e., from (19,9,9) to (21,10,10). Hence, the gain of total throughput mainly comes from the larger coding gain provided by Polar codes compared to Turbo codes. Moreover, the gain of each of the three UEs is almost the same to each other, which means Polar codes provide gain for both central UE and edge UE. Filter related impacts: In Step 3, the filter is used and extra 10 RBs can be taken into use. The total throughput gain over the baseline increases to over 125%. Compared to Step 2 (SCMA and Polar without filter), the gain is (63.56 59.14)/59.14 = 7.47%. Although the number of RBs is increased by 10%, throughput gain increment cannot be expected as that high due to the MCS table design, where only about 9% gain in TB size exists in many MCS levels. Besides making use of the guard band, filter also helps to suppress interferences to the neighbour bands. Impacts of neighbour band interference: At last, the neighbour band interference is turned on. We can see from Table IV that the total throughput almost has no degradation. Although the frequency spacing between two bands are only one subcarrier (15kHz), with the help of the carefully designed filter, the impacts of the interference from the neighbour band can be limited to an acceptable degree. Also the OOBE of the working band can be suppressed.
1874 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 35, NO. 8, AUGUST 2017 To make a summary, SCMA provides more than 90% gain compared to the baseline in this field test. Polar coding gives extra 10% gain due to its larger coding gain. With the help of filter, extra 10 RBs can be used with robustness to the neighbour band interference, which leads to extra 7% gain. In total, by employing SCMA, Polar and f-ofdm, more than 120% gain over the baseline is obtained in the field test. IV. CONCLUSIONS In this paper, we provide results from 5G joint trial from DOCOMO and HUAWEI. In the trial, three SE improving technologies are investigated. Firstly, SCMA, as a nonorthogonal multiple access technology, enables overlapping of data from different UEs and provides both coding gain and shaping gain. The throughput and SE gain obtained by using SCMA is 91.11% compared to the baseline when the SNR difference between central and edge UEs was about 19dB. Secondly, Polar coding, a proven capacity achieving coding scheme, provides extra 10.92% gain on throughput and SE due to its larger coding gain than Turbo coding. Finally, with the help of f-ofdm, the guard band 10 RBs are used to transmit data, which improves the total throughput and SE by 7.47%. The overall gain of these three technologies is 127.81%, which means we can double the system total throughput and SE through using these three technologies. REFERENCES [1] J. G. Andrews et al., What will 5G be? IEEE J. Sel. Areas Commun., vol. 32, no. 6, pp. 1065 1082, Jun. 2014. [2] P. Banelli, S. Buzzi, G. Colavolpe, A. Modenini, F. Rusek, and A. Ugolini, Modulation formats and waveforms for 5G networks: Who will be the heir of OFDM?: An overview of alternative modulation schemes for improved spectral efficiency, IEEE Signal Process. Mag., vol. 31, no. 6, pp. 80 93, Nov. 2014. [3] Huawei. (2013). 5G: A Technology Vision (White Paper). [Online]. Available: http://www.huawei.com/5gwhitepaper/ [4] A. Benjebbour, K. Saito, A. Li, Y. Kishiyama, and T. Nakamura, Nonorthogonal multiple access (NOMA): Concept and design, in Signal Processing for 5G: Algorithms and Implementations. Chichester, U.K.: Wiley, 2016, pp. 143 168. [5] L. Dai, B. Wang, Y. Yuan, S. Han, C.-L. I, and Z. Wang, Nonorthogonal multiple access for 5G: Solutions, challenges, opportunities, and future research trends, IEEE Commun. Mag., vol. 53, no. 9, pp. 74 81, Sep. 2015. [6] Z. Wei, J. Yuan, D. W. K. Ng, M. Elkashlan, and Z. Ding, A survey of downlink non-orthogonal multiple access for 5G wireless communication networks, ZTE Commun., vol. 14, no. 4, pp. 17 25, Oct. 2016. [7] H. Nikopour and H. Baligh, Sparse code multiple access, in Proc. IEEE 24th Int. Symp. Pers. Indoor Mobile Radio Commun. (PIMRC), Sep. 2013, pp. 332 336. [8] M. Taherzadeh, H. Nikopour, A. Bayesteh, and H. Baligh, SCMA codebook design, in Proc. IEEE 80th Veh. Technol. Conf. (VTC Fall), Sep. 2014, pp. 1 5. [9] C. E. Shannon, A mathematical theory of communication, Bell Sys. Tech. J., vol. 27, no. 3, pp. 379 423, 1948. [10] C. Berrou, A. Glavieux, and P. Thitimajshirna, Near Shannon limit error-correcting coding and decoding: Turbo-codes. 1, in Proc. Int. Conf. Commun., 1993, pp. 1064 1070. [11] R. G. Gallager, Low density parity check codes, Ph.D. dissertation, Massachusetts Inst. Technol., Cambridge, MA, USA, 1963. [12] D. J. MacKay and R. M. Neal, Good codes based on very sparse matrices, in Proc. IMA Int. Conf. Cryptograph. Coding, 1995, pp. 100 111. [13] S. Lin and D. J. Costello, Error Control Coding. Engle Wood Cliffs, NJ, USA: Prentice-Hall, Pearson Education, 2004. [14] E. Arıkan, Channel polarization: A method for constructing capacityachieving codes for symmetric binary-input memoryless channels, IEEE Trans. Inf. Theory, vol. 55, no. 7, pp. 3051 3073, Jul. 2009. [15] R. Pedarsani, S. H. Hassani, I. Tal, and E. Telatar, On the construction of polar codes, in Proc. IEEE Int. Symp. Inf. Theory (ISIT), Jul. 2011, pp. 11 15. [16] K. Niu, K. Chen, J. Lin, and Q. T. Zhang, Polar codes: Primary concepts and practical decoding algorithms, IEEE Commun. Mag., vol. 52, no. 7, pp. 192 203, Jul. 2014. [17] G. Fettweis, M. Krondorf, and S. Bittner, GFDM-generalized frequency division multiplexing, in Proc. IEEE 69th Veh. Technol. Conf. (VTC Spring), Apr. 2009, pp. 1 4. [18] F. Schaich, Filterbank based multi carrier transmission (FBMC) Evolving OFDM: FBMC in the context of WiMAX, in Proc. Eur. Wireless Conf. (EW), 2010, pp. 1051 1058. [19] F. Schaich and T. Wild, Waveform contenders for 5G OFDM vs. FBMC vs. UFMC, in Proc. 6th Int. Symp. Commun., Control Signal Process. (ISCCSP), May 2014, pp. 457 460. [20] J. Li, K. Kearney, E. Bala, and R. Yang, A resource block based filtered OFDM scheme and performance comparison, in Proc. ICT, May 2013, pp. 1 5. [21] J. Abdoli, M. Jia, and J. Ma, Filtered OFDM: A new waveform for future wireless systems, in Proc. IEEE 16th Int. Workshop Signal Process. Adv. Wireless Commun. (SPAWC), Jun. 2015, pp. 66 70. [22] X. Zhang, M. Jia, L. Chen, J. Ma, and J. Qiu, Filtered-OFDM Enabler for flexible waveform in the 5th generation cellular networks, in Proc. IEEE Global Commun. Conf. (GLOBECOM), Dec. 2015, pp. 1 6. [23] A. Roessler, 5G waveform candidates application note, Rohde&Schwarz, Munich, Germany, Tech. Rep. 1MA271, 2016. [24] D. Wu et al., A field trial of f-ofdm toward 5G, in Proc. IEEE Global Commun. Conf. (GLOBECOM), Dec. 2016, pp. 1 6. [25] Evolved Universal Terrestrial Radio Access (E-UTRA); Physical Channel and Modulation, Ver 12.3.0, 3GPP, Valbonne, France, document TS 36.211, 2014. Jian Wang (S 10 M 14) received the B.Sc. degree in information engineering and the Ph.D. degree in information and communication systems from Zhejiang University, Hangzhou, China, in 2008 and 2013, respectively. He is currently a Senior Engineer with Huawei Technologies Co., Ltd., Hangzhou. His research interests mainly focus on wireless network and communications, including transmission control protocol performance, cross-layer optimization, and quality-of-service provisioning in wireless networks, channel equalizer, and channel coding in wireless communications. His current interests focus on channel coding standardization and field trial verification for NR. Aixiang Jin received the B.Sc. degree in electronic information of science and technology from the Nanjing University of Aeronautics and Astronautics. He was with Nanjing YongXin Company from 2010 to 2012. He is currently a Researcher with Huawei Technologies Co., Ltd. He participated in the project of asynchronous chip developing. He is currently involved in researching technologies of communications of 5G. Dai Shi received the B.Sc. and master s degrees from the Electronic Information College, Tongji University, in 2005 and 2008, respectively. She joined Huawei Technologies Co., Ltd., Shanghai, in 2008, where she was a Researcher in a research project on next generation of communication technologies from 2008 to 2017. Her research interests mainly focus on 5G key technologies PoC verification, such as user centric no cell, non-orthogonal multiple access schemes, new waveform and grantfree transmission procedures, and field trial test to verify the performance of 5G new air interface performance.
WANG et al.: SPECTRAL EFFICIENCY IMPROVEMENT WITH 5G TECHNOLOGIES 1875 Lei Wang received the B.Sc. degree in communication engineering and the Ph.D. degree in communication and information systems from the Nanjing University of Aeronautics and Astronautics in 2005 and 2012, respectively. He is currently a Senior Researcher with Huawei Technologies Co., Ltd., Shanghai. His research interests include wireless communication and signal processing. Hui Shen (M 09) was born in 1975. He received the Ph.D. degree in electronics and communication engineering from the Huazhong University of Science and Technology, China, in 2004. From 2004 to 2007, he was with the Technical Center, Research Department of ZTE Corporation, Shenzhen, China, as a Researcher and a Standard Senior Engineer. He is currently with Huawei Technologies Co., Ltd., Shenzhen. His research interests lie in the areas of wireless communications, including multipleantenna systems, multi-user MIMO precoding, interference alignment, and polar code. Dan Wu received the bachelor s degree in chemical physics from the University of Science and Technology of China in 2004, and the Ph.D. degree in physical chemistry from the Chinese Academy of Sciences, China, in 2009. From 2009 to 2011, he was an Assistant Professor with the Chinese Academy of Sciences. Since 2012, he has been with Huawei Technologies, China. His research interests are communications theory, including waveform and channel coding. Liang Hu received the bachelor s and master s degrees in energy engineering and automation from the Harbin Institute of Technology in 2000 and 2004, respectively. In 2012, he joined Huawei Technologies Co., Ltd., Hangzhou, where he has been the Project Leader of 5G UE since 2012. He is currently focusing on researching, field trial, and prototype verification of 5G key technologies. Liang Gu has over 16 years of experience on wireless product, standardization, and advanced research. In 2005, he joined Huawei Technologies Co., Ltd., where he is currently a Principle Engineer. He was actively involved in WiMAX and LTE standard development. Many of his ideas have been adopted in standards and products. He is also leading 5G system research and 5G trials. He and his team have released the Huawei first 5G trials. He hold over 60 granted patents. Lei Lu received the B.Sc. and master s degrees from the Electronic Information College, Tongji University, in 2004 and 2007, respectively. He was a Visiting Researcher with the École Supérieure d Informatique, Électronique, Automatique, France, in 2005. In the same year of graduation, he joined Huawei Technologies Co., Ltd., Shanghai, where he was a Researcher in the Huawei research project Next Generation of Communication Technologies from 2007 to 2011. Since 2012, he has been the Project Leader on Huawei 5G Air Interface PoC and Field Trial focusing mainly on 5G key technologies PoC verification, such as user centric no cell, non-orthogonal multiple access schemes, new waveform and grant-free transmission procedures, and field trial test to verify the performance of 5G new air interface performance. Yan Chen received the B.Sc. degree from the Chu Kochen Honored College, Zhejiang University, in 2004, and the Ph.D. degree from the Institute of Information and Communication Engineering, Zhejiang University, in 2009. She was a Visiting Researcher with the University of Science and Technology from 2008 to 2009. In the same year of graduation, she joined Huawei Technologies Co., Ltd., Shanghai. She was the Team Leader of the Huawei research project Green Radio Excellence in Architecture and Technology from 2010 to 2013, during which, she was also the Project Leader of the Green Transmission Technologies in GreenTouch Consortium. Since 2013, she has been the Technical Leader on Huawei 5G air interface design focusing mainly on nonorthogonal multiple access, advanced receiver and interference management, flexible duplex, grant-free transmissions, system design towards ultra-low latency, and ultra-high reliability performance. Jun Wang, photograph and biography not available at the time of publication. Yuya Saito received the B.S. degree in electrical engineering from Hosei University, Tokyo, Japan, in 2002, the M.S. degree from the Tokyo Institute of Technology, Tokyo, in 2004, and the Ph.D. degree from the University of Colorado Boulder, Boulder, Co, USA, in 2009. In 2009, he joined NTT DOCOMO, Inc. His research interests include the fifth generation of mobile communications system, including massive MIMO and non-orthogonal multiple access. Anass Benjebbour (S 99 M 04 SM 09) received the B.Sc. Diploma degree in electrical engineering, and the M.Sc. and Ph.D. degrees in telecommunications from Kyoto University, Japan, in 1999, 2001, and 2004, respectively. In 2004, he joined NTT DOCOMO, Inc. Since 2010, he has been a leading member of its 5G team. He has authored or co-authored over 100 technical publications, four book chapters, and an Inventor of over 50 patent applications. His research interests include novel system design concepts and radio access techniques for next-generation mobile communication systems (5G), such as massive MIMO, NOMA, and waveform design. He is a Senior Member of the IEICE. He served as a 3GPP and ITU-R Standardization Delegate and a Secretary of the IEICE RCS Conference from 2012 to 2014. He was an Associate Editor of the IEICE Communications Magazine from 2010 to 2014 and an Associate Editor of the IEICE Transactions on Communications from 2014 to 2018. Yoshihisa Kishiyama received the B.E., M.E., and Dr.Eng. degrees from Hokkaido University in 1998, 2000, and 2010, respectively. In 2000, he joined NTT DOCOMO, where he has been involved in the Research and Development of 4G/5G radio access technologies and physical layer standardization in 3GPP. He is currently a Senior Research Engineer with the 5G Laboratory, NTT DOCOMO Inc. In 2012, he received the ITU-AJ Award from the International Telecommunication Union Association of Japan for his contributions to LTE.
本文献由 学霸图书馆 - 文献云下载 收集自网络, 仅供学习交流使用 学霸图书馆 (www.xuebalib.com) 是一个 整合众多图书馆数据库资源, 提供一站式文献检索和下载服务 的 24 小时在线不限 IP 图书馆 图书馆致力于便利 促进学习与科研, 提供最强文献下载服务 图书馆导航 : 图书馆首页文献云下载图书馆入口外文数据库大全疑难文献辅助工具