Source-linked AI summary
Coexistence Mechanism between eMBB and uRLLC in 5G Wireless Networks
Anupam Kumar Bairagi, Md. Shirajum Munir, Madyan Alsenwi, Nguyen H. Tran, Sultan S Alshamrani, Mehedi Masud, Zhu Han, Choong Seon Hong
TL;DR
Coexisting eMBB and uRLLC traffic creates a scheduling dilemma because eMBB seeks high rates while uRLLC requires stringent latency and reliability. The paper formulates and decomposes a puncturing-based MEAR optimization, then evaluates its scheduling algorithms against baselines using MEAR and fairness. The conclusion reports a notable performance gain for the proposed approach.
Problem
Sharing radio resources between eMBB and uRLLC requires balancing eMBB MEAR against uRLLC traffic provisions under competing service requirements.
Method
The paper decomposes a puncturing-based coexistence optimization into eMBB and uRLLC scheduling sub-problems, using PSUM, TM, and a lower-complexity heuristic.
Results
The proposed approach is evaluated against baseline approaches using eMBB MEAR and fairness, with the conclusion reporting a notable performance gain.
Takeaways & Limitations
The study provides a puncturing-based coexistence mechanism for enabling uRLLC and eMBB traffic to share the same radio resources.
Abstract
from arXiv · showhide
uRLLC and eMBB are two influential services of the emerging 5G cellular network. Latency and reliability are major concerns for uRLLC applications, whereas eMBB services claim for the maximum data rates. Owing to the trade-off among latency, reliability and spectral efficiency, sharing of radio resources between eMBB and uRLLC services, heads to a challenging scheduling dilemma. In this paper, we study the co-scheduling problem of eMBB and uRLLC traffic based upon the puncturing technique. Precisely, we formulate an optimization problem aiming to maximize the MEAR of eMBB UEs while fulfilling the provisions of the uRLLC traffic. We decompose the original problem into two sub-problems, namely scheduling problem of eMBB UEs and uRLLC UEs while prevailing objective unchanged. Radio resources are scheduled among the eMBB UEs on a time slot basis, whereas it is handled for uRLLC UEs on a mini-slot basis. Moreover, for resolving the scheduling issue of eMBB UEs, we use PSUM based algorithm, whereas the optimal TM is adopted for solving the same problem of uRLLC UEs. Furthermore, a heuristic algorithm is also provided to solve the first sub-problem with lower complexity. Finally, the significance of the proposed approach over other baseline approaches is established through numerical analysis in terms of the MEAR and fairness scores of the eMBB UEs.
I. INTRODUCTION
The introduction frames eMBB–uRLLC coexistence as a scheduling dilemma caused by competing rate, latency, reliability, and spectral-efficiency requirements. It proposes a puncturing-based optimization and decomposed scheduling approach, evaluated against baseline methods.
- eMBB requires high data rates, while uRLLC targets 99.999% reliability and 0.25–0.30 ms/packet latency.
- Allocating dedicated resources to sporadic uRLLC traffic can under-utilize radio resources, motivating effective multiplexing with eMBB.
- 3GPP-recommended puncturing reduces control-channel overhead relative to short-TTI operation but requires mechanisms to recognize and heal punctured transmissions.
- With puncturing, eMBB is scheduled at slot beginnings, while uRLLC can override overlapping eMBB transmissions on the same physical resources.
- The paper formulates MEAR maximization for eMBB UEs, decomposes scheduling into eMBB and uRLLC sub-problems, and applies PSUM, TM, and a heuristic algorithm.
- The proposed scheduling approach is experimentally compared with PS, MUPS, RS, EDS, and MBS using eMBB MEAR and fairness scores.
II. LITERATURE REVIEW
The literature review covers analytical models, architectures, slicing, dynamic multiplexing, puncturing, and specialized schedulers for 5G traffic coexistence. It highlights differing trade-offs involving eMBB performance, uRLLC latency, reliability, spectral efficiency, and resource utilization.
- Prior studies analyze heterogeneous 5G service multiplexing through slicing, physical-layer design, queueing models, and dynamic time–frequency mechanisms.
- Dynamic mechanisms were proposed because static bandwidth partitioning is inefficient for eMBB and uRLLC traffic.
- Puncturing-based studies improve resource utilization or uRLLC latency, but some approaches substantially deteriorate eMBB performance.
- MUPS combines multi-user pairing with punctured scheduling to improve spectral efficiency, while its feasibility and uRLLC reliability depend on pairing and interference conditions.
- Other proposals use null-space spatial degrees of freedom or CVaR to serve uRLLC traffic and limit uncertainty in eMBB transmissions.
III. SYSTEM MODEL AND PROBLEM FORMULATION
The model multiplexes eMBB and uRLLC traffic over licensed resource blocks, with eMBB allocations made per time slot and uRLLC allocations per mini-slot. It formulates a fairness-oriented optimization that protects uRLLC latency and reliability while maximizing eMBB achievable rates.
- System model: The system models one gNB serving eMBB and uRLLC user equipments over licensed resource blocks, with uRLLC arrivals modeled by a Gaussian distribution.Time slots are divided into mini-slots, and each uRLLC request has payload size L_m,t.
- Traffic and rate model: Because uRLLC payloads are short and interference threatens reliability, their achievable rate is approximated using a finite-blocklength formulation rather than Shannon’s data-rate expression.The approximation accounts for SINR and channel dispersion when uRLLC traffic overlaps eMBB traffic.
- Puncturing mechanism: Puncturing assigns zero power to eMBB UEs in punctured mini-slots, preventing eMBB interference from affecting uRLLC traffic.The resulting puncturing causes throughput loss for the affected eMBB UEs, estimated with a linear model.
- Problem formulation: The optimization maximizes the actual achievable rates of eMBB UEs across time slots while serving uRLLC requests and preserving their latency and reliability constraints.Max-Min fairness is applied to provide fairness-oriented service quality and spectral-efficiency objectives.
- Resource allocation: The gNB allocates resource blocks orthogonally among eMBB UEs at each time slot, while uRLLC UEs receive blocks immediately on a mini-slot basis.uRLLC traffic can overlap with ongoing eMBB traffic within a time slot.
- Problem formulation: The formulation is a chance-constrained combinatorial programming problem that is NP-hard because of its combinatorial structure.The constraints also encode orthogonal resource allocation, minimum resource assignment, resource limits, and binary decision variables.
IV. DECOMPOSITION AS A SOLUTION APPROACH FOR PROBLEM (18)
Problem (18) is decomposed into eMBB and uRLLC resource-allocation sub-problems while retaining the original objective. eMBB scheduling operates per time slot, whereas uRLLC scheduling operates per mini-slot, with PSUM used for the eMBB sub-problem.
- Problem (18) is divided into eMBB and uRLLC resource-allocation sub-problems while preserving the original objective.
- eMBB resources are scheduled on a time-slot basis, while uRLLC resources are scheduled on a mini-slot basis.
- At the next time slot, the gNB compensates eMBB users according to their prior losses and seeks to equalize throughput loss.It allocates more RBs to user 2 and fewer to user 3 after their respective losses.
- The resulting per-slot balancing is intended to reach the objective of (18) over the long run.
- A. PSUM as a Solution of the Sub-Problem (19): PSUM approximately solves sub-problem (19) because its NP-hardness makes globally optimal computation expensive.The algorithm relaxes binary variables, adds a penalty term, and uses successive upper-bound minimization with rounding.
B. Solution of Sub-Problem (20) through TM
Sub-problem (20) is difficult because of its chance constraint and combinatorial variable. The paper reformulates it as a transportation problem and obtains an optimal allocation using MCC initialization followed by MODI optimization.
- The chance constraint and combinatorial variable make sub-problem (20) difficult to solve directly.
- The paper transforms the relevant formulation into a balanced transportation model by adding slack variables and a dummy demand entry.The modified problem is identified as a BTM, with an added zero-cost row in the cost matrix.
- 1) Determining Initial Feasible Solution by MCC Method: MCC constructs an initial feasible solution by assigning flow to the lowest-cost cells while updating supply and demand.
- 2) MODI Method for Finding an Optimal Solution: MODI improves the MCC solution by evaluating empty cells, selecting the most negative cost indicator, and reallocating along a stepping-stone path.
- 2) MODI Method for Finding an Optimal Solution: The MODI procedure repeats reallocation until all cost indicators satisfy k_u,e ≥ 0, producing the optimal allocation χ.
C. Low-Complexity Heuristic Algorithm for Solving Sub-Problem (19)
Algorithm 1 can solve sub-problem (19) optimally, but its computation time grows rapidly with problem size. The paper therefore proposes a lower-complexity heuristic that may sacrifice optimality.
- Algorithm 1 solves sub-problem (19) optimally, but its computation time grows much faster as problem size increases.
- The large number of eMBB UEs and short decision period motivate a faster heuristic algorithm.
- Algorithm 2 allocates resources equally among eMBB UEs at t = 1.
- In later time slots, Algorithm 2 allocates resources according to proportional loss in the previous time slot.
- Algorithm 2 has complexity dependent on T and E and is designed to accommodate eMBB EAR over the long run.
V. NUMERICAL ANALYSIS AND DISCUSSIONS
The numerical analysis evaluates the proposed scheduler against state-of-the-art baselines using MEAR and fairness for eMBB UEs. Across varying uRLLC arrivals, traffic rates, and loads, the proposed method generally provides the strongest MEAR and comparable or better fairness.
- Small-network comparison: For a small network, the proposed PSUM+TM method has a 4.20% average-MEAR optimality gap and a 0.32% average-fairness optimality gap.The proposed method's MEAR ECDF is around 0.50 and its probability of MEAR being at least 20 Mbps is around 0.70, while fairness reaches at least 0.995 with probability 0.80.
- Varying uRLLC arrivals: MEAR decreases as σ increases because additional uRLLC traffic requires more resource blocks, while the proposed method remains preferred to the compared methods.The performance gap between the proposed method and PS reduces with increased uRLLC arrivals as PS gains more opportunities to adjust users with higher expected achieved rates.
- Average MEAR and fairness: For σ = 1, the proposed method exceeds RS, EDS, MBS, PS, and MUPS in average MEAR by 10.20%, 10.87%, 5.77%, 5.77%, and 18.55%, respectively.For σ = 10, the corresponding improvements are 15.22%, 16.43%, 6.22%, 3.75%, and 70.20%.
- Varying load and traffic: Across varying uRLLC load L and traffic σ, the proposed method has the highest or a comparable fairness score, although fairness generally decreases as these demands increase.For high σ, fairness can increase with L, whereas MUPS fairness decreases as σ increases.
VI. CONCLUSIONS
The paper formulates eMBB–uRLLC coexistence as maximizing eMBB MEAR while satisfying uRLLC traffic requirements. It decomposes scheduling across the two services and reports performance gains over baseline approaches.
- Conclusion: The proposed approach addresses coexistence by maximizing eMBB UEs' MEAR while attending to uRLLC traffic requirements.The problem is framed for sharing the same radio resources between eMBB and uRLLC traffic.
- Conclusion: A decomposition strategy schedules eMBB resources with a PSUM-based algorithm and uRLLC resources through an optimal transportation model using MCC and MODI methods.The eMBB subproblem is resolved in every time slot, and a heuristic algorithm is also provided for efficient PRB scheduling.
- Conclusion: Extensive simulation outcomes demonstrate a notable performance gain over baseline approaches in the considered indicators.The reported indicators are the MEAR and fairness performance of eMBB UEs.