NOMA-Assisted Multi-BS MEC Networks for Delay-Sensitive and Computation-Intensive IoT Applications

The burgeoning and ubiquitous deployment of the Internet of Things (IoT) landscape struggles with ultra-low latency demands for computation-intensive tasks in massive connectivity scenarios. In this paper, we propose an innovative uplink non-orthogonal multiple access (NOMA)-assisted multi-base station (BS) mobile edge computing (BS-MEC) network tailored for massive IoT connectivity. To fulfill the quality-of-service (QoS) requirements of delay-sensitive and computation-intensive IoT applications, we formulate a joint task offloading, user grouping, and power allocation optimization problem with the overarching objective of minimizing the system’s total delay, aiming to address issues of unbalanced subchannel access, inter-group interference, computational load disparities, and device heterogeneity. To effectively tackle this problem, we first reformulate task offloading and user grouping into a non-cooperative game model and propose an exact potential game-based joint decision-making (EPG-JDM) algorithm, which dynamically selects optimal task offloading and subchannel access decisions for each IoT device based on its channel conditions, thereby achieving the Nash Equilibrium. Then, we propose a majorization-minimization (MM)-based power allocation algorithm, which transforms the original subproblem into a tractable convex optimization paradigm. Extensive simulation experiments demonstrate that our proposed EPG-JDM algorithm significantly outperforms state-of-the-art decision-making algorithms and classic heuristic algorithms, yielding performance improvements of up to 19.3% and 14.7% in terms of total delay and power consumption, respectively.

💡 Research Summary

The paper addresses the pressing challenge of supporting delay‑sensitive and computation‑intensive (DSCI) Internet‑of‑Things (IoT) applications in massive connectivity scenarios. It proposes an uplink non‑orthogonal multiple access (NOMA)‑assisted multi‑base‑station (BS) mobile edge computing (MEC) architecture, where each BS is co‑located with an MEC server and a set of orthogonal sub‑channels is allocated to avoid inter‑BS interference. IoT devices, each equipped with a single antenna, offload their tasks to the nearest MEC server via NOMA, allowing multiple devices to share the same sub‑channel in the power domain while successive interference cancellation (SIC) is performed at the receiver.

The core research problem is a joint optimization of three coupled decisions: (1) task offloading (selection of the target BS), (2) user grouping (assignment of devices to sub‑channels), and (3) transmit power allocation. The objective is to minimize the total system delay, defined as the sum of uplink transmission delay and MEC processing delay, subject to constraints on device power budgets, sub‑channel capacity (maximum number of users per channel), MEC computational capacity, and individual task deadlines. This problem is NP‑hard due to the mixture of discrete (offloading, grouping) and continuous (power) variables and the non‑convex nature of the SINR‑based rate expressions.

To make the problem tractable, the authors decompose it into two sub‑problems and solve them iteratively.

1. Joint Offloading and Grouping Sub‑problem – Modeled as a non‑cooperative game where each device is a player. The strategy of a player consists of a pair (target BS, sub‑channel). The utility is the negative of the device’s experienced delay, which depends on the strategies of other devices through inter‑group interference. The authors prove that this game is an exact potential game; the potential function coincides with the total system delay. Consequently, any unilateral best‑response update monotonically decreases the potential, guaranteeing convergence to a Nash equilibrium. Based on this property, they develop the Exact Potential Game‑Joint Decision‑Making (EPG‑JDM) algorithm, which iteratively lets each device update its decision according to current channel conditions and the decisions of others. The algorithm is distributed, requires only local channel state information (CSI) and limited signaling, and converges in a few iterations.

2. Power Allocation Sub‑problem – With offloading and grouping fixed, the remaining problem is to allocate transmit powers to minimize the total delay. The original formulation is non‑convex because the achievable rate is a logarithmic function of SINR, which itself is a fractional function of the powers. The authors apply the Majorization‑Minimization (MM) technique: at each iteration they construct a convex surrogate (majorizer) of the original objective around the current power vector, solve the resulting convex problem (e.g., via interior‑point methods), and update the powers. This process yields a sequence of decreasing objective values and converges to a stationary point of the original problem.

The two sub‑problems are solved in an alternating fashion: EPG‑JDM provides a new offloading/grouping configuration, then the MM‑based power allocation refines the power vector. The loop repeats until the total delay change falls below a predefined threshold.

Performance Evaluation – Simulations are conducted under realistic 5G parameters with varying numbers of devices (50–200), sub‑channels (5–10), heterogeneous task sizes, and deadline requirements. Baseline schemes include Max‑Min based grouping, Gale‑Shapley matching, nearest‑BS offloading, and computing‑capacity based offloading. Results show that the proposed framework reduces total delay by up to 19.3 % and power consumption by up to 14.7 % compared with the best baseline. Moreover, the algorithm achieves higher MEC utilization and better load balancing across BSs, mitigating the “hot‑spot” problem common in single‑BS MEC deployments.

Critical Observations and Limitations –

• The analysis assumes perfect CSI and ideal SIC; in practice, channel estimation errors and residual interference can degrade performance.
• The orthogonal allocation of sub‑channels among BSs eliminates inter‑BS interference but may underutilize spectrum; more aggressive spectrum sharing could be explored with coordinated interference management.
• Convergence speed of the game‑theoretic component is acceptable in simulations, yet real‑time signaling overhead for large‑scale IoT deployments needs quantification.
• The MM algorithm converges to a local optimum; global optimality is not guaranteed, though empirical results are promising.

Future Directions – Extending the framework to incorporate imperfect CSI, robust SIC models, and learning‑based prediction of traffic patterns would increase resilience. Joint design of downlink NOMA for result dissemination, integration of reconfigurable intelligent surfaces (RIS) for channel shaping, and federated learning for distributed decision making are promising avenues.

In summary, the paper makes a substantial contribution by marrying NOMA with multi‑BS MEC, formulating a realistic joint optimization problem, and solving it with a novel combination of exact potential games and majorization‑minimization. The proposed solution demonstrably improves latency and energy efficiency for massive DSCI IoT applications, offering a solid foundation for future 5G/6G edge‑computing research.