Bandwidth-Aware Quantized-Event Distributed State Estimation for V2X Networks

Vehicle-to-everything (V2X) networks enable cooperative perception and control, yet practical deployments are constrained by tight bandwidth budgets, stochastic delays, and packet losses. We propose a bandwidth-aware quantized-event distributed state estimator that integrates local EKF-based filtering, normalized-innovation-squared (NIS) event triggering with multi-level innovation quantization, delay-aware reception (de-duplication and outdated-packet screening), and covariance-intersection (CI) fusion without exchanging cross-covariances. We establish uniform mean-square boundedness and prove monotonic accuracy–communication trade-offs with respect to quantization resolution, triggering thresholds, link reliability, and delay bounds. Simulations in cooperative adaptive cruise control (CACC) and intersection scenarios improve the RMSE–bandwidth Pareto frontier over periodic unquantized EKF+CI, triggering-only, and quantization-only baselines; under a strict per-agent budget (e.g., Bcap= 9 bits/step/agent), admitted traffic remains bounded and budget-aware retuning mitigates scalability-induced RMSE degradation as fleet size increases. Stress tests show enhanced robustness: innovation gating with covariance inflation reduces the divergence rate to zero in the evaluated setting. Profiling indicates that EKF updates and CI fusion dominate runtime with worst-caseO(n3) complexity in the state dimension.