ScholarWorks@성균관대학교

초록

In high-density device environments, networks must support large-scale simultaneous access under tight latency constraints. In contention-based medium access control (MAC), throughput degrades and delay grows rapidly because the collision probability increases as more devices attempt to access the channel. We consider combining intelligent reflecting surfaces (IRS), which control the propagation via passive metamaterial elements, with non-orthogonal multiple access (NOMA), which multiplexes users in the power domain. Prior studies commonly assume a single IRS serving many devices without MAC-layer contention modeling, which limits performance in dense regimes. We propose a MAC protocol that (i) partitions the IRS into multiple element groups, (ii) independently optimizes each group's reflection coefficients, and (iii) applies NOMA with successive interference cancellation (SIC) during IEEE 802.11 DCF contention. We formulate a throughput maximization with unit-modulus phase constraints and probabilistic interference, and solve it via an alternating optimization framework combining interior-point updates for power-selection probabilities and successive convex approximation for IRS phases. Simulations show that balanced high/low power selection (dH 0.5) and phases near 0 or 2 π maximize the rate and saturation throughput across a wide range of device densities.

키워드