Latency-Aware Uplink NOMA-MEC with PASS: A Feasibility-First Resource Allocation Framework

Abstract

Meeting hard task-completion deadlines for many simultaneously offloading users is the binding challenge in mobile-edge computing (MEC). As the user count grows, the transmit power required to satisfy every deadline within per-user budgets rises until no feasible allocation exists. We take a \emph{feasibility-first} view of this $K$-user problem in an uplink non-orthogonal multiple access (NOMA) MEC system served by a pinching-antenna system (PASS), whose radiating points establish strong line-of-sight (LoS) links that lower the received-power requirement. We formulate a latency-constrained joint optimization of transmit powers, CPU frequencies, and antenna configuration. We then derive a \emph{closed-form least-power feasibility test}: the componentwise-least allocation meeting every deadline. It certifies feasibility with the largest attainable margin to the power budgets and defines a practical \emph{quality-of-service (QoS)-feasibility outage} metric, the probability that no allocation meets all deadlines within budget. We solve the problem with a three-block alternating-optimization algorithm that is provably monotone-convergent and has polynomial complexity. Numerical results show that the proposed scheme sustains deadline feasibility at user loads where equal-power, orthogonal multiple access (OMA), and single-antenna baselines fail or degrade sharply, each by a mechanism predicted directly from the system model, whereas the proposed design degrades gracefully. Stress tests further show that feasibility is sustained under order-of-magnitude budget reductions, and that the pinching-antenna array itself enables constrained-budget operation. The result is a scalable, closed-form-initialized design for deadline-feasible offloading.