Neutralizing Schedulability Asymmetry Through Period Decomposition: Configured Grant Scheduling for Periodic Reservations in NR Sidelink Mode 1

초록

Periodic resource reservation underpins vehicular sidelink traffic, yet its compatibility is inherently asymmetric. In New Radio (NR) sidelink Mode 1, two configured grant (CG) streams collide if and only if the greatest common divisor (GCD) of their periods divides their offset difference, so coprime periods cannot be separated by any offset on a single subchannel. Admission opportunity is thus governed by the arithmetic of the period set, not by load alone-a disparity we call schedulability asymmetry. We propose Divisor-Aligned Period Decomposition (DAPD), which fixes a common divisor d and reduces coexistence to a closed-form GCD test: periods that are multiples of d are admitted directly, and incompatible ones are reshaped into a d-aligned substitute within the delay budget rather than rejected. This levels the schedulability asymmetry at its source for any catalog period whose delay budget admits a d-compatible substitute: once every request is represented on a common divisor, no period is structurally disadvantaged, so such a period can be admitted regardless of its arithmetic relationship to the others. Under a collision-only scheduler model that isolates deterministic reservation collisions from physical-layer effects, every admitted CG is collision-free by construction. An extension, DAPD+, salvages the resulting slack with dynamic grants that never overlap an admitted CG. In simulation over 1000 seeds across vehicle densities up to subchannel saturation, DAPD+ attains the highest overall packet delivery ratio among all evaluated schedulers, including an oracle offset optimizer and a recent multi-configuration CG scheduler, while holding admitted-CG delivery at one and keeping far more per-vehicle delivery ratios above target, where the reactive baseline's per-vehicle reliability collapses with density. DAPD+ thus serves the largest volume of periodic traffic at high reliability without sacrificing aggregate throughput, equalizing admission opportunity across heterogeneous periods.

키워드