Non-Orthogonal Multiple Access (NOMA) for 5G Systems


The key challenges of 5G can be addressed by NOMA technology. In this, same time, frequency, and spreading-code resources are shared by the multiple users via allocation of power. The entire bandwidth can be exploited by each user in NOMA for entire communication time due to which latency has been reduced and users’ data rates can be increased. For multiple access, the power domain has been used by NOMA in which different power levels are used to serve different users. 3GPP also included NOMA in LTE-A due to its spectral efficiency and is known as multiuser superposition transmission (MUST) which is two user special case of NOMA. More number of users can be accommodated by NOMA. The key features of NOMA are:

  • Utilization of the power domain for user multiplexing

For user multiplexing in NOMA, the power domain is used. In power domain, the multiplexing of users is done using superposition coding at the transmitter side and by employing successive interference cancellation (SIC), demultiplexing is done at the receiver. The decoding of the messages of the users which are at high power levels is done by treating the information of other users as noise. While, SIC technique has been employed by the users at low power levels by removing the information of other users successively.

  • Utilization of different channel conditions

In NOMA scheme, users can utilize diverse channel conditions. The users’ channel condition is used to determine the coefficients of power allocation. Less transmission power is allocated for a message for user which have strong channel gain and more power is allocated for message for user whose channel conditions are poor.

  • By slightly increasing the complexity of the receiver a controllable interference has been introduced by NOMA so that overloading can be realized which can improve the spectral efficiency and provide massive connectivity.
  • The system capacity and coverage can be improved by using NOMA.
  • In deployment of wide area networks, robust performance, despite of latency provided due to mobility or CSI feedback, can be provided by NOMA.

NOMA system model

The basic functionality of NOMA is shown in the fig. 1 when receiver is employed with SIC. One BS and two UEs are consisted in this figure where UE1 is situated near BS and UE2 is far from BS, called cell-edge user, in DL transmission.


Fig. 1. Basic NOMA employed with SIC at the receiver

Firstly, the cell edge user (CEU) data has been decoded by the UE near the BS and then its corresponding user data has been decoded. At the transmitted side, superposition coding (SC) is used to combine together all user symbols.

NOMA for uplink transmission

NOMA scheme has also been proposed for UL transmission for OFDM systems in which the resource allocation is removed and same subcarrier is shared by multiple users without any coding/spreading redundancy. In each subcarrier, the number of users is limited so that complexity of receiver can be controlled in UL. NOMA technique provides higher spectral efficiency as compared to that of OMA and complexity of receiver is also lower. In the new NOMA scheme, the optimum multiuser detection is implemented at the receiver side so that users’ data can be easily separated. Hence, the spectral efficiency and fairness has been improved by the system-level performance of the proposed NOMA scheme as compared to the OMA scheme. The multiuser capacity region can be achieved by NOMA with minimum mean squared error-based liner filtering followed by the successive interference canceller (MMSE-SIC) in the multiple access channel (MAC) that can enhance the throughput of the user and CEU. The signaling overhead caused due to NOMA can be reduced by employing SIC at BS. Hence, the system-level throughput performance is improved due to significant control of transmission power.

An eNB with interference cancellation receiver can be employed when single carrier non-orthogonal multiple access (SC-NOMA) scheme is used. An overlapped spectra has been allocated to multiple intra-cell UE whose receiving antennas exceed in number. For SC-NOMA, there is an proposal of frequency domain (FD) scheduling on the basis of expected value of cell throughput. First of all, a candidate subband (SB) has been selected by eNB for a piece of UE with the use of metric. After that both the candidate SB calculate the expected cell throughput to a piece of UE and all the SBs allocated previously. The candidate SB has been allocated to the piece of UE by eNB when the expected cell throughput is more as compared to the assigned candidate SB throughput. Hence, there is an increase in the throughput performance of the FD based scheduling in the SC-NOMA by using this method.

Non-orthogonal access in multiple cells is used for an uplink multi-antenna environment. The space resources are shared by the proposed UL-NOMA so that sum capacity can be improved. The interference can be reduced by using the set selection algorithm. To mitigate the interest interference caused due to set selection algorithm, orthogonality between the users channels is used. To maximize the sum capacity of the system, the optimal power control is used.

NOMA for DL transmission

NOMA system level performance

To improve the capacity of the system, NOMA with SIC is used. This can also improve the CEU throughput performance. There is no need for the transmitter to rely on the frequency-selective CQI for the improvement in the performance of the system. Some of the key link adaptation functionalities of long-term evolution (LTE) radio interfaces such as adaptive modulation and coding (AMC), hybrid automatic repeat request (HARQ), time/frequency-domain scheduling, and outer loop link adaptation (OLLA) are considered in NOMA along with the NOMA functionalities such as multiuser power allocation so that higher potential gains can be achieved over OFDMA. For data deduction, the order of UE devices can be determined by using channel gain order. For data transmission, high channel gain and less power is required for the UE situated near the transmitter. To receive data at UE side, low channel gain and more transmission power is required by CEUs. Hence, NOMA has provided superior cell throughput, CEU throughput, and the degree of proportional as compared to that of OMA.

Cloud Radio Access Network (C-RAN)

In C-RAN, all the BSs are connected via a cloud as shown in fig. 2. To connect BSs and supply services to the mobile users, the network services are provided in a cloud.


Fig. 2. System model of DL C-RAN

As compared to OFDMA, a sum rate of upto eight times more is achieved by using NOMA-based C-RAN system model.

Multiuser Beamforming System

The inter-cluster and inter-user interference and effective power allocation scheme can be minimized by using a new clustering algorithm that increases the sum capacity. Two users whose correlation is high and channel gain difference in each cluster is large, are selected by using the proposed clustering algorithm. Both the users are provided with the power based on the clustering algorithm so that a single BF vector supports these users. As a result, it is shown that NOMA-BF sum capacity is greater.

Cooperative NOMA

When three pieces of UE within the cell are considered, cooperative NOMA is used to analyze the system performance. Different cells are consisted of two BSs that serve the CEU along with the users within each cell by cooperating with each other. Three different conditions are used to compare the rate of these three users as:

  • When data is transmitted by both BSs to CEU under cooperation
  • When CEU is served by only one BS
  • When BS, used to send data on the basis of dynamic cell selection, is selected by the CEU.

There is a comparison between individual rates and the sum rate of the three users with respect to the distance between the piece of UE and BSs and with the increase in SNR values. The near and far users are considered by the performance of the system model in the conventional NOMA systems. The interference caused by near users on the CEU is less as less power is required for data transmission and is considered as noise. The effect of intra- and inter-cell interference on the CEU is shown in fig. 3.


Fig. 3. Cooperative NOMA with inter-user and intra-cell interference

At the receiver and transmitter side, interference cancellation techniques are employed in the proposed system model. There is an increase in the receiver complexity with the increase in the number of users. To cancel the interference, the information of other users can be decoded by the users with better channel conditions in the conventional NOMA system. To improve reliability for the poor connection users, the users with better channel conditions are used as relay. The diversity gain achieved by the cooperative NOMA system is the maximum for all the users.

In this scheme, the far user’s signal is canceled by the user near to BS and then the desired signal is detected. On the other hand, the near user’s signal is considered as a noise by the far user and its desired signal is detected.

Frequency reuse for downlink NOMA

The cell edge throughput can be improved when the inter-cell user interference has been reduced. To achieve this, more transmission power for CEU and wider bandwidth to inter-cell users can be allocated. Because of large reuse factor, the interference at the cell-edge can be reduced by using fractional frequency reuse scheme. The users are provided with transmission power of different levels in the proposer FFR scheme. Hence, CEU and cell-average throughput can be improved by using this scheme.

NOMA with single-user MIMO

The NOMA has been combined with the open-loop and closed-loop single-user MIMO ti investigate the performance gain. For cell average throughput, there is 23% performance gain of NOMA for open-loop and closed-loop SU-MIMO. For CEU throughput, there is 33% performance gain of NOMA for open-loop and closed-loop SU-MIMO. The rank optimization methods are also proposed at the transmitter that will improve the performance gain as well.

Receiver complexity

The link-level simulations are used to evaluate and compare the performance of the different receiver designs. For downlink NOMA, new schemes are proposed for transmission and reception. At the transmitter side, joint modulation is applied so that Gray mapping of the superposed signal of different users can be achieved. There is a use of a simple log-likelihood ratio (LLR) calculation method so that the desired signal can be decoded without SIC processing and the complexity can be reduced at the receiver side.

There is a direct detection of the desired signal of the cell center user without any detection of the CEU and SIC processing signal. There is an evaluation and comparison of its performance with other SIC receivers. Much better performance can be achieved by this proposed receiver as compared to the symbol-level SIC even when SIC processing is not needed. For a large range of power allocation ratios, the codeword level SIC and the proposed receiver are used and when the power allocation ratio is assigned to CEUs, the symbol-level SIC is used. There is an allocation of larger power allocation ratio to CEUs when, for cell center users, high-order modulation is applied.

Types of Non-Orthogonal Access Schemes

There are three multiple access schemes that are sparse code multiple access (SCMA), multi-user shared multiple access (MUSA), and pattern division multiple access (PDMA).

To further improve the spectrum efficiency, SCMA scheme is used in which the spatial and code domain are considered. Huawei proposed SCMA which is a frequency-domain NOMA technique. The sparse codebook is used by SCMA so that the spectrum efficiency can be improved by code domain multiple access.

Another 5G multiple access scheme, proposed by Zhongxing Telecommunication Equipment Corporation (ZTE), is MUSA. The spreading sequences are used by MUSA, which can be non-orthogonal, and advanced SIC receivers. First of all, the spreading sequence is used to spread the data of the user. After that, the data of all the users’ is combined and then transmitted. The data of each user can be demodulated and retrieved by using the advanced SIC-based receiver at the receiver side.

Datang proposed the PDMA scheme which is a novel NOMA scheme. At the transmitter’s side, it is based on the joint design of SIC-amenable pattern and at the receiver’s side, low-complexity quasi-ML SIC detection is used. To distinguish between the users, non-orthogonal characteristic patterns in different domains, like power, space, or code domains are used at the transmitter side. At the receiver side, an equivalent diversity degree can be achieved by the multiple users by using the SIC amendable detection. Compared with MUSA and PDMA, better performance in terms of bit-error rate are achieved by using SCMA.

The interleave division multiple access (IDMA) is another type of NOMA technique. IDMA give access to a large number of stations, hence, the efficiency of the system can be improved. Different interleaver patterns are used by IDMA to distinguish the users.