LoRaWAN Multi-Channel vs. RLNC Simulation Dataset

This dataset supports the paper "Scalability Boundaries in Dense EU868 LoRaWAN: A Validated Airtime–Recovery Model" (Abdulaziz, 2026). It provides aggregated ns-3 per-run results used to validate the scalability-boundary model, falsifiable predictions F1–F4, and the loss-attribution analysis. Simulator: ns-3.45 with LoRaWAN extensions; Class-A EU868; 1% unified per-device duty-cycle budget; ADR enabled; unconfirmed uplinks; 1200 s runs (100 s warm-up discarded in post-processing). Profiles (5): baseline (single-channel uncoded), MC (eight EU868 channels), NC (RLNC, one channel), MC-RLNC (RLNC, eight channels), Rep (repetition control, R ∈ {2,3,4}, staggered copies, 5 s spacing). Scenario grid: 50, 100, and 200 nodes; traffic low / moderate / heavy (10 s, 5 s, 2 s); RLNC generation size k ∈ {2,4,8} where applicable. Main boundary-validation campaign: 72 scenarios × 20 seeds = 1,440 runs (four core profiles); Rep and confirmatory n>k RLNC runs are reported separately. Files (v2.0):- lora_simulation_results.csv — master per-run log (5,237 runs, all profiles)- nc_comprehensive_results.csv, mc_rlnc_comprehensive_results.csv — optional aggregates- README.md, REPRODUCIBILITY.md — parameters matching Section V of the manuscript- scripts/ — run_mc_scenarios.sh, run_nc_scenarios.sh, run_mc_rlnc_scenarios.sh, run_rep_scenarios.sh, run_rep_stagger_scenarios.sh, run_nc_nk_confirm.sh Metrics (CSV): Profile, density, traffic, ADR, multi-channel flags, RLNC window / repetition count, PDR, average latency (ms in raw CSV), Jain fairness, transmit energy, collision rate, packets sent/received, and related fields. DOI: 10.5281/zenodo.20358885