Energy-Efficient Resource Allocation in NOMA Heterogeneous Networks

• The proposed NOMA approach targets cross-tier interference mitigation and improved system energy efficiency through resource optimization.Source: §I · p. 2
• NOMA’s shared-band access uses SIC so stronger users can cancel interference from weaker users multiplexed on the same sub-band.Source: §I · p. 2
• The study is motivated by rising wireless traffic and energy consumption, which make balancing data-rate performance against energy use important.Source: §I · p. 3
• The paper studies energy-efficient user scheduling and power allocation for downlink 5G NOMA heterogeneous networks under perfect and imperfect CSI.Source: §I · p. 3

• Macrocell users and small-cell users can share the same sub-frequency band in the overlay-frequency model.Source: §II · p. 3
• Higher data rates can be achieved by allocating more power to UEs, but this incurs greater power consumption.Source: §II · p. 3
• The design objective is to balance system sum-rate improvement against power consumption.Source: §II · p. 3

• SIC lets a higher-channel-gain user cancel interference from a lower-channel-gain user in NOMA.Source: A. NOMA Overview · pp. 3–4
• A prior NOMA resource-allocation scheme limited each sub-band to two users and reported higher sum rate and energy efficiency than traditional OFDMA.Source: A. NOMA Overview · p. 4
• NOMA allows users in a small cell to share a sub-band with macrocell users; the macrocell’s strong users can remove small-cell interference via SIC.Source: A. NOMA Overview · p. 4

• The model has a macro base station and K overlaid small cells, with each small cell occupying one sub-frequency band.Source: B. NOMA Heterogeneous Network Model · pp. 4–5
• Macrocell users can share a frequency band with small-cell users, while users within a small cell can also be multiplexed on the same band.Source: B. NOMA Heterogeneous Network Model · p. 4
• The model neglects interference between small cells and excludes backhaul considerations, assuming wired MBS–SBS links.Source: B. NOMA Heterogeneous Network Model · pp. 5–6
• The resource-allocation objective is to maximize energy efficiency by splitting coupled user-scheduling and power-allocation subproblems.Source: B. NOMA Heterogeneous Network Model · p. 6
• Energy efficiency is defined as system sum rate divided by total power consumption, including circuitry and transmission power.Source: B. NOMA Heterogeneous Network Model · p. 6

• For the scheduling design, the authors initially assume the base stations know the channel state information.Source: §III · p. 6
• The scheduling problem is non-convex mixed-integer programming; exhaustive search can find the global optimum but is computationally infeasible in practice.Source: §III · p. 6
• The proposed sub-optimal scheduler uses matching theory to assign users to sub-bands, with at most a set number of users multiplexed on each sub-band to reduce receiver decoding complexity.Source: §III · p. 7
• The worst-case complexity of the sub-optimal scheme is O(N^2M), and the paper reports it is much lower than exhaustive search.Source: §III · p. 7
• The system energy-efficiency objective is constrained by maximum transmission power, minimum user data rate, and the maximum number of users per sub-band.Source: §III · p. 7

A. User Scheduling

• The scheme schedules small-cell and macro-cell users separately, then co-multiplexes them on occupied sub-bands, subject to a maximum number of users per sub-band.Source: A. User Scheduling · p. 7
• Small-cell scheduling ranks small cells on each sub-band by channel gains between the small-cell base station and its users, assuming equal power allocation among those users.Source: A. User Scheduling · p. 7
• For each sub-band, the scheduler selects the small cell with the highest channel gain and calculates its energy efficiency assuming equal power allocation.Source: A. User Scheduling · p. 7
• Macro-cell users are preferred by channel gain, with energy efficiency used to choose among users when a sub-band has reached its multiplexing limit.Source: A. User Scheduling · pp. 7–8
• Optimal user-to-sub-band matching requires searching all combinations and is impractical; the proposed sub-optimal scheduling has lower complexity.Source: A. User Scheduling · p. 8

B. Power Allocation

• The energy-efficiency optimization is non-convex; DC programming and time-sharing are noted as ways to transform the problem into a convex one.Source: B. Power Allocation · p. 9
• The paper uses fractional transmit power allocation to allocate power among users and across sub-bands, replacing equal power allocation to further improve system energy efficiency.Source: B. Power Allocation · p. 9
• The energy-efficiency-versus-small-cell-count plot uses a 10 MHz bandwidth, 20 W system peak power, and 0.1 W circuit power per sub-band.Source: Figure 3 · p. 9, B. Power Allocation · p. 9
• NOMA HetNet outperforms OFDM HetNet in energy efficiency, and energy efficiency increases as the number of small cells increases.Source: B. Power Allocation · p. 9
• The NOMA HetNet user-average-data-rate CDF is below the OFDM HetNet CDF under equal power allocation.Source: B. Power Allocation · p. 10

IV. ENERGY EFFICIENT RESOURCE OPTIMIZATION FOR NOMA HETEROGENEOUS NETWORK WITH IMPERFECT CSI

• Scheduling uses estimated CSI and begins with equal power; the practical sub-optimal algorithm avoids the exhaustive search over possible user combinations.Source: §IV · p. 11
• The imperfect-CSI problem includes an outage-probability requirement, maximum transmitted power, and a limit on multiplexed users per sub-band.Source: §IV · pp. 10–11
• The outage constraint is transformed into minimum-power constraints, allowing user scheduling and power allocation to be treated separately.Source: §IV · p. 11
• The scheduling scheme assigns each small cell a sub-band, selects high-channel-gain MUEs, limits multiplexed users, and uses FTPA for their power allocation.Source: §IV · p. 11
• For fixed scheduling, the power problem is convex under the system power constraint, yielding a unique optimum via gradient-assisted binary search.Source: §IV · p. 11
• With estimation-error variance 0.05, energy efficiency increases with small-cell count; NOMA HetNet exceeds OFDM HetNet, and FTPA exceeds equal-power allocation.Source: §IV · p. 11, Figure 5 · p. 12
• With different estimation-error variances, energy efficiency deteriorates as errors increase, while NOMA HetNet always outperforms OFDM HetNet.Source: §IV · p. 12
• Energy efficiency of NOMA HetNet with FTPA power allocation is 27% more than that of OFDM HetNet with equal-power (EQ) allocation under imperfect channel CSI.Source: §IV · p. 12

V. OPEN ISSUES, AND CHALLENGES

• Combining NOMA with massive MIMO raises resource-allocation challenges because beamforming and differing macrocell and small-cell antenna configurations must also be considered.Source: §V · p. 12
• Energy harvesting from solar, wind, or thermoelectric sources remains an open issue because those sources are intermittent over time and space.Source: §V · p. 13
• User pairing is identified as a key research direction, with matching games proposed as a possible way to model and optimize pairing.Source: §V · p. 13
• Open, closed, and hybrid small-cell access modes require different energy-saving methods in NOMA networks.Source: §V · p. 13

VI. CONCLUSION

• The paper examines energy-efficient user scheduling and power allocation for 5G NOMA HetNets under both perfect and imperfect CSI.Source: §VI · p. 13
• The resource-optimization problem is addressed with convex optimization and a two-step user-scheduling and power-optimization scheme.Source: §VI · p. 13
• Simulations compare the schemes with an existing scheme in energy efficiency; the conclusion envisions joint scheduling and power allocation as a way to further enhance overall system energy efficiency.Source: §VI · pp. 13–14