Panel Workshop Starts at 4:30 pm

Transcription

1 Panel Workshop Starts at 4:30 pm

2 Outline of the Panel Discussion Building connections between academic and industry DL NoMA in 3GPP UL NoMA in 3GPP Some thinking towards future Unified framework Some implementation from industry field trials introduction from DoCoMo Open minded discussion towards future study What are your most interested application scenarios for NoMA? What are the most important metrics to evaluate NoMA in your mind? What are your most interested features for NoMA design? Page 2

3 Discussion on NoMA in Standardization Page 3

4 DL NoMA (MUST) in Rel-14 LTE WI in 3GPP Three categories of MUST schemes (power domain reflected in compound modulation) In MUST Category 1, coded bits of two or more co-scheduled UEs are independently mapped to component constellation symbols but the composite constellation does not have Gray mapping. In MUST Category 2, coded bits of two or more co-scheduled UEs are jointly mapped to component constellations and then the composite constellation has Gray mapping. In MUST Category 3, coded bits of two or more co-scheduled UEs are directly mapped onto the symbols of a composite constellation (the power domain superposition is done implicitly during the bit-to-symbol mapping). Figures from Non-Orthogonal Transmission Technology in LTE Evolution, IEEE COMMAG, July 2016 Page 4

5 Candidate Schemes Observed in Rel-14 NR SI in 3GPP MA scheme Company Full Name Main Features Proposed Usage SCMA [R ] Huawei Sparse code multiple access Symbol-level spreading UL/DL PDMA [R ] CATT Pattern division multiple access Symbol-level spreading UL/DL LDS-SVE [R ] Fujitsu Low density spreading-signature vector extension Symbol-level spreading UL/DL MUSA [R ] ZTE Multi-user shared access Symbol-level spreading UL NCMA [R ] LGE Non-orthogonal coded multiple access Symbol-level spreading UL NOCA [R ] Nokia Non-orthogonal coded access Symbol-level spreading UL FDS [R ] Intel Frequency domain spreading Symbol-level spreading UL RSMA [R ] Qualcomm Resource spread multiple access Symbol-level scrambling UL RDMA [R ] MTK Repetition division multiple access Symbol-level scrambling UL GOCA [R ] MTK Group orthogonal coded access Symbol-level scrambling UL IGMA [R ] Samsung Interleave-grid multiple access Symbol-level interleaving UL/DL NOMA [R ] DCM Non-orthogonal multiple access Symbol-level superposition UL/DL LSSA [R ] ETRI Low code rate and signature based shared access Bit-level interleaver UL IDMA [R ] Nokia Interleaver division multiple access Bit-level interleaver UL LCRS [R ] Intel Low code rate spreading Bit-level spreading UL Page 5

6 Agreed Objectives of Rel-15 UL NoMA SI in 3GPP Non-orthogonal multiple transmission scheme Transmitter side signal processing schemes for non-orthogonal multiple access [RAN1]: Modulation and symbol level processing, including spreading, repetition, interleaving, new constellation mapping, etc. Coded bit level processing including interleaving and/or scrambling, etc. Symbol to resource element mapping, sparse or not, etc. Demodulation reference signal. Other signal is not excluded. Receivers for non-orthogonal multiple access: [RAN1, RAN4] MMSE receiver, successive/parallel interference cancellation (SIC/PIC) receiver, joint detection (JD) type receiver, combination of SIC and JD receiver, or other receivers. The study should consider performance, receiver complexity, etc. Procedures related to the non-orthogonal multiple access [RAN1] UL transmission detection HARQ, including transmission scheme, feedback scheme, and combining scheme Link adaptation MA signature allocation/selection Synchronous and asynchronous operation Adaptation between orthogonal and non-orthogonal multiple access Link and system level performance evaluation or analysis for non-orthogonal multiple access continued from performance metrics identified from Rel-14. The benchmark for comparison is OFDM contention based multiple access. Realistic modelling of Tx/Rx impairment including potential PAPR issue, channel estimation error, power control accuracy, collision, etc. should be considered. [RAN1] Traffic model and Deployment scenarios of embb (small packet), URLLC and mmtc Device power consumption Coverage (link budget) Latency and signalling overhead BLER reliability, capacity and system load Physical abstraction (link-to-system mapping model) Transmitter Receiver Procedure Evaluation Page 6

7 Suggestion for Component-wise Study Transmitter How to make good use of the unified framework to converge the study? FEC Bit level operations Symbol level operations OFDM operations Options Bit-level Interleaver /Scrambler UE specific scrambler (LTE) UE specific interleaver Modulated Symbol Sequence Generator Options Symbols to RE Mapping Single tone or multi-tone modulation UE specific symbol spreading User specific symbol interleaver/scrambler Sparse/non-sparse resource mapping Power adjustment IFFT First to identify for each scheme, which component block has UE specific signature design Each company justify the benefit of the component block with UE specific design, comparing with a corresponding baseline Multiple options proved of benefit can be recommended within each component block Page 7

8 Suggestion for Component-wise Study Receiver How to converge the study of advanced receiver? Received symbols 1st Stage Signal Cancel SU Detector such as MF/MMSE SU Detector such as MF/MMSE De- Mapping Signal Rebuild De- Mapping FEC Decoder FEC Decoder Decoded bits Decoded bits Received symbols 1st Stage Signal Cancel MU Detector such as MPA/EPA/ MMSE/ESE MU Detector such as MPA/EPA/ MMSE/ESE De- Mapping Signal Rebuild De- Mapping LLR of undecoded bit streams FEC Decoder FEC Decoder FEC Decoder FEC Decoder Decoded bits Decoded bits 2nd Stage 2nd Stage SU detector + hard SIC MU detector + hybrid soft/hard PIC Page 8

9 Discussion on NoMA in Field Trials Presentation from Yoshihisa Kishiyama, NTT DoCoMo Page 9

10 Discussion on Open Questions Page 10

11 Open Questions Discussion Open minded discussion towards future study Applicability of NoMA to WiFi, especially for cost of 2-5 dollars with unlicensed band Prof. Ding: #UE can be supported can be boosted by NoMA given limited # of channels; so improved connectivity Dr. Ma: many features defined for licensed band can be reused for unlicensed band, NoMA is one example, but some special care needs to pay. Prof. Hanzo: Current WiFi could not work very well with a large crowd, so any improvement in overall connectivity/capacity should be considered. NoMA with advanced detectors/receivers, different from linear MMSE receiver, can work with more rank efficient scenarios and improve WiFi capacity. Moreover, there is duality of receiver side and transmit side which implies by applying all types of iterative receivers, not so stringent design required. Prof. Choi: the reason of NoMA s popularity is due to its nature as generalization of OMA, it can be applied to WiFi or any other communication system. Kishiyama-san: cellular network favors NoMA application due to its more accurate power control/link adaptation, but in WiFi, these conditions are not the same, which needs to be considered. Page 11

12 Open Questions Discussion Open minded discussion towards future study What are your most interested application scenarios for NoMA? What are the most important metrics to evaluate NoMA in your mind? What are your most interested features for NoMA design? Kishiyama-san: massive connectivity is the most beneficial scenario; as operator, performance is more important than complexity; NoMA + full duplex/dynamic TDD is of interest to deal with the inter-user interference Prof. Hanzo: varied services, delay-sensitive/non-sensitive, power domain superposition at different SNR level with precise power control, PIC with iterative information exchange may make life easier than hard SIC in the sense of ordering users and reduce latency Prof. Ding: in general, like to think about applications that haven t been discussed; the current main scenarios such as NoMA + MIMO are mainly to show NoMA are compatible; different scenarios shall have different metrics (delay, thp, connectivity); would like to reverse the question as what other features defined outside NoMA can facilitate NoMA design jointly. Some examples of new applications V2X worth considering. Dr. Ma: NoMA supposed to be used in all scenarios such as embb, URLLC, and mmtc. In particular interest, NoMA + grantfree, due to benefit of signaling overhead/energy consumption/latency reduction, which are important for machine type devices. For DL, UE experience in multi-cell is important and reduce handover. In general, SE is still the key for all scenarios, but KPI is also related to scenarios. For example, UE drop rate given UE service requirement can be a measure for DL and URLLC, for UL, the overall connectivity in a system given bandwidth/time, etc. About features, spatial multiplexing and full duplex should be the features jointly considered with NoMA. Joint optimization is interesting topic. Prof. Choi: two key words in feature design, i.e., power v.s. energy domain noma + interference. Two types of interferences, strong (in power domain noma) receiver can detect and cancel no impact in theory and thus better than weak interference. HARQ process design can be considered. Page 12

13 Open Questions Discussion Open minded discussion towards future study Grant-free access is one important scenario for NoMA, but it seems to be contradictory with high capacity. How can we balance grant-free access and high capacity. For instance, for 10 year design, how to balance the two. Prof. Ding: this should depends on scenarios. For URLLC, delay and reliability is more important than capacity; for IoT, connectivity is more important than SE. Multiple modes can be supported in the future and applied with network slicing as a way to accommodate different schemes for different scenarios. Dr. Ma: For grant-based, if you have ideal scheduler, then it is always the best, however, in practice it costs resources to do scheduling. If the packet size is small, then the ratio of overhead in the scheduling overhead is large. Instead, grant-free NoMA can solve this with robust transmission and the efficiency can be achieved by multiplexing a group of users instead of for each user to achieve high SE. For NoMA, different modes shall be supported. Prof. Choi: wireless communications need to combat the fading, and for unknown environment, it takes time/resource to understand environment. Prof. Hanzo: scheduling based can give best user more resource to achieve total high capacity, but for fairness and delay limited services, it is not. Depending on the service type, e.g., uplink buffer-aided solutions should be investigated. Page 13

14 Open Questions Discussion Open minded discussion towards future study NoMA can be considered as sending data simultaneous. What if the data is correlated? What is the theoretical limits how good NoMA can do in this scenario? Prof. Choi: there are some papers addressing NoMA for multi-cast. And we are generalizing the problem which may cover the issue. Another relative example is in sensor networks, which is under investigation. Prof. Hanzo: there is discussion on Research Gate. Page 14

15 Copyright 2014 Huawei Technologies Co., Ltd. All Rights Reserved. The information in this document may contain predictive statements including, without limitation, statements regarding the future financial and operating results, future product portfolio, new technology, etc. There are a number of factors that could cause actual results and developments to differ materially from those expressed or implied in the predictive statements. Therefore, such information is provided for reference purpose only and constitutes neither an offer nor an acceptance. Huawei may change the information at any time without notice.