Short‑interval wireless schedulers must deliver a complete, feasible beam‑and‑resource allocation before each execution deadline; a late or invalid output provides no utility in the current control cycle. Extending the search can increase transmitted traffic but may miss the deadline or leave only an infeasible intermediate solution. We introduce Feasibility‑Preserving Transactional Refinement (FPTR), which always retains a validated feasible allocation as the incumbent. Each refinement stage builds a separate candidate while leaving the incumbent unchanged. A candidate replaces the incumbent only if it is complete, passes the feasibility validator before the stage cutoff, and strictly improves the objective; otherwise it is discarded. Four stages are employed: buffer‑aware resource selection, compatibility‑group (CG) user sharing, beam‑mask rebuilding from remaining demand, and joint reassignment of two resources. On 150 fixed synthetic instances, the full pipeline (Full) raises traffic by 3.83 % on average over a lightweight beam‑first reference and by 1.79 % over its Base constructor. In a separate matched‑input suite, Full outperforms six independent search baselines (ALNS, tabu search, GA, simulated annealing, iterated local search, and GRASP) by 6.68‑8.35 %, where outputs later than 100 ms receive zero utility. All 750 Full runs in the internal suite returned feasible allocations, none exceeding 100 ms. Separating mandatory output construction from optional refinement lets spare computation time improve the allocation without endangering the incumbent. The open‑source implementation supports independent reproduction.
Review: FPTR demonstrates reliable incremental optimization under strict deadlines, highlighting the practical value of transactional refinement.