Source-linked AI summary
SCMA for Downlink Multiple Access of 5G Wireless Networks
Hosein Nikopour, Eric Yi, Alireza Bayesteh, Kelvin Au, Mark Hawryluck, Hadi Baligh, Jianglei Ma
TL;DR
The paper addresses downlink efficiency and robustness under heavily and lightly loaded network conditions. It develops MU-SCMA with pairing, power sharing, rate adjustment, and scheduling, and evaluates SCMA spreading in both settings. MU-SCMA achieves 28% throughput and 36% coverage gains over OFDMA, while lightly loaded SCMA achieves 56% throughput and 26% coverage gains.
Problem
Downlink networks need improved throughput and robust link adaptation under heavy loading, while lightly loaded networks face rapidly varying interference across time and frequency.
Method
The paper develops MU-SCMA using code-domain user pairing, power sharing, rate adjustment, and scheduling, and evaluates SCMA's spreading and interference averaging in downlink scenarios.
Results
MU-SCMA provides 28% throughput and 36% coverage gains over OFDMA, while SCMA provides 56% throughput and 26% coverage gains in a lightly loaded system.
Takeaways & Limitations
MU-SCMA is attractive for downlink networks because it avoids full instantaneous channel knowledge and is reported as robust to channel variation compared with MU-MIMO.
Abstract
from arXiv · showhide
Sparse code multiple access (SCMA) is a new frequency domain non-orthogonal multiple-access technique which can improve spectral efficiency of wireless radio access. With SCMA, different incoming data streams are directly mapped to codewords of different multi-dimensional cookbooks, where each codeword represents a spread transmission layer. Multiple SCMA layers share the same time-frequency resources of OFDMA. The sparsity of codewords makes the near-optimal detection feasible through iterative message passing algorithm (MPA). Such low complexity of multi-layer detection allows excessive codeword overloading in which the dimension of multiplexed layers exceeds the dimension of codewords. Optimization of overloading factor along with modulation-coding levels of layers provides a more flexible and efficient link-adaptation mechanism. On the other hand, the signal spreading feature of SCMA can improve link-adaptation as a result of less colored interference. In this paper a technique is developed to enable multi-user SCMA (MU-SCMA) for downlink wireless access. User pairing, power sharing, rate adjustment, and scheduling algorithms are designed to improve the downlink throughput of a heavily loaded network. The advantage of SCMA spreading for lightly loaded networks is also evaluated.
I. INTRODUCTION
The introduction motivates SCMA as an open-loop alternative for downlink user multiplexing and presents MU-SCMA for both heavily and lightly loaded networks. It links SCMA's sparsity, spreading, and code-domain layers to detection efficiency, link adaptation, and robustness.
- MU-MIMO requires accurate CSI feedback and suffers from channel aging and cross-layer interference when CSI is poorly estimated.
- SCMA maps incoming bits directly to sparse multi-dimensional codewords whose super-imposed layers share time-frequency resources.
- SCMA enables user multiplexing without CSI knowledge of paired users, while MU-SCMA shares downlink power among multiplexed layers using limited CQI.
- Sparse SCMA codewords support low-complexity MPA detection with ML-like performance even when many layers overload the system.
- SCMA improves on LDS and repetition coding by directly mapping bits to multidimensional codewords, providing shaping gain and improved spectral efficiency for larger constellations.
- The paper evaluates interference averaging in lightly loaded networks and develops MU-SCMA pairing, power-sharing, rate-adjustment, and scheduling techniques for heavily loaded networks.
A. Downlink SCMA Model
The SCMA downlink model represents each user's data through sparse codewords multiplexed over shared OFDMA resources. The received signal includes the users' channel effects, allocated power, and receiver noise.
- An SCMA encoder maps log_2(M) coded bits to a K-dimensional complex codebook containing M sparse codewords.
- Each codeword has fewer than K nonzero entries, and all codewords share the same K−N zero dimensions.
- Users may contain separate SCMA layers, and their codewords are multiplexed over K shared orthogonal resources such as OFDMA tones.
- The model accounts for each user's per-tone transmit power, equal power distribution across layers, channel vectors, and ambient noise.
- Because SCMA is nonlinear, the paper uses a linear sparse-sequence model to develop MU-SCMA pairing algorithms and then constructs its MIMO equivalent.
B. MIMO Equivalent of Linear Sparse Sequence
The paper reformulates linear sparse spreading as a MIMO system and derives a capacity expression dependent on SNR, layer count, and signature structure. This establishes link-adaptation flexibility and identifies the performance gap that SCMA's multidimensional modulation addresses.
- A linear sparse sequence spreads each QAM symbol using a signature vector, with signature matrices collecting the layer-specific spreading vectors.
- Stacking all receive antennas yields a MIMO representation whose multiplexed layers span K resources, with overloading factor J/K.
- For a single user, the open-loop capacity is an upper bound determined by the channel, signature matrix, layer power allocation, and noise covariance.
- The sparse-sequence bound is tight for QPSK but diverges at higher constellation sizes, where SCMA recovers performance through multidimensional modulation.
- The rate depends on SNR, number of layers, and signature matrix, providing an additional degree of freedom for SCMA link adaptation.
III. DOWNLINK MU-SCMA ALGORITHMS
Downlink MU-SCMA requires coordinated user pairing, power allocation, and rate adjustment under a total transmit-power constraint. Pairing choices are evaluated against adjusted rates, but exhaustive search may be impractical.
- MU-SCMA selects paired users from a user pool and splits the transmitter's total power according to their channel conditions.
- After power allocation, each user's adjusted rate depends on both the paired user and the power-sharing strategy.
A. User Pairing to Maximize Wiegthed Sum-Rate
The pairing design seeks to maximize weighted sum-rate by evaluating user combinations and their adjusted rates after power sharing. Exhaustive search is potentially impractical, motivating a greedy alternative.
- Pairing targets maximization of the weighted sum-rate for candidate user combinations.
- Adjusted user rates depend jointly on the paired users and the power-sharing strategy.
- Exhaustive search checks all U(U − 1) pairing options and single-user options, but its complexity grows on the order of U^2.
- Greedy scheduling first selects a user using the single-user criterion, then pairs another user with it.
- A greedy pairing result is accepted only when the stated validity condition is satisfied.
B. Rate Adjustment and Detection Strategy
The detection strategy develops paired-user operation by modeling a degraded channel and assigning different decoding roles to high- and low-quality users. Capacity-region operating points determine feasible rate and power-sharing choices.
- The analysis assumes γ1 > γ2 and transforms the broadcast channel into an approximated degraded model for tractable design.
- In the degraded OFDMA model, user 1 is the good-quality user with a higher instantaneous rate than user 2.
- User 1 can decode user 2 before successive interference cancellation, reducing its own detection problem to a single-user form.
- The SIC strategy corresponds to point C in the capacity region, where user 1 receives its maximum achievable rate.
- User 2 treats user 1 as co-paired interference and requires a tighter rate condition for detecting its intended signal.
- The shaded capacity-region area represents operating points where both users can detect their intended streams under the selected power-sharing factor.
- When user weights satisfy R1 > R2, point B is maximized only at α = 0, reducing scheduling to the single-user case.
C. OFDMA Power Sharing Optimization
For OFDMA, power sharing is optimized at the paired-user operating point A. The solution is constrained to valid power fractions, and user separability improves when long-term rate imbalance is sufficiently large.
- The OFDMA design optimizes the power-sharing factor α while paired users operate at point A of the capacity region.
- The optimum α is obtained by solving the weighted sum-rate optimization for point A.
- The optimal solution α* is valid only when α* lies in (0,1), with a smaller range optionally imposed to facilitate detection.
- The optimized detection margin shows that two paired users are easily separated when their long-term-rate ratio is sufficiently large.
D. SCMA Power Sharing Optimization
The SCMA extension derives adjusted rates for both paired users and optimizes their power split through the resulting weighted sum-rate expression. Valid solutions are selected from real power-sharing roots within the feasible interval.
- The adjusted rate of the first paired user is derived by extending the OFDMA detection strategy through the SCMA rate expression.
- The adjusted rate of the second paired user is likewise expressed using the SCMA rate and paired-user channel model.
- The SCMA rate expression can be decomposed using the signature matrix and its eigenvalue-related diagonal terms.
- The optimum SCMA power-sharing factor is obtained by solving the derivative condition for the point-A weighted sum-rate.
- Among multiple roots, the selected α* is real, lies in (0,1.0), and maximizes the point-A weighted sum-rate.
IV. NUMERICAL RESULTS
System-level simulations compare OFDMA, SCMA, and MU-SCMA under heavily and lightly loaded network conditions. SCMA and MU-SCMA improve throughput and coverage, while MU-SCMA also enables a coverage-throughput trade-off.
- Simulation setup: Simulations use 3GPP-based assumptions with four-dimensional SCMA codewords, two nonzero elements per codeword, and up to six detected layers.SCMA layers use a near-optimal MPA detector; MU-SCMA detection follows the paper’s proposed strategy.
- Heavily loaded network: The full-buffer scenario models a heavily loaded 10 MHz LTE network with average 10 users per cell and wideband proportional-fair scheduling.Results compare cell aggregate throughput and 5 percentile coverage rate for OFDMA, SCMA, and MU-SCMA.
- Coverage-throughput trade-off: More than 51% cell-throughput gain is obtained with MU-SCMA over OFDMA at a 700 kbps coverage rate.The modified proportional-fair metric controls the coverage-throughput trade-off.
- Lightly loaded network: SCMA achieves 56% throughput and 26% coverage gain over OFDMA with 50% resource utilization and subband scheduling.SCMA’s spreading and slower transmit-power variation average interference across frequency in the lightly loaded scenario.
V. CONCLUSION
The paper introduces MU-SCMA to increase downlink spectral efficiency in 5G cellular networks. It develops the approach for single-transmit-point SIMO channels and identifies multi-transmit-point and MIMO extensions as future work.
- Conclusion: MU-SCMA increases downlink spectral efficiency without requiring full knowledge of users’ instantaneous channels.The paper contrasts this open-loop code-domain multiplexing with MU-MIMO’s channel-aging and feedback-overhead issues.
- Conclusion: The developed MU-SCMA techniques target single-transmit-point and SIMO channels, while multiple-transmit-point and MIMO extensions remain future research.The stated scope is limited to the single-TP SIMO setting.