Drift-Resilient Predictive Interference Management: Online Adaptation of an Extreme-Value Predictor
Abstract
Predictive interference management supports ultra-reliable low-latency communication (URLLC) by allocating finite-blocklength resources against a high conditional quantile of next-slot interference. Building on an extreme-value-theory/kernel-density-estimation (EVT/KDE) predictor with quantile-defined interference states, this paper makes the predictor resilient to unannounced shifts in the interference distribution. Its cumulative refitting on all past observations does not specify how the state partition and retained samples should evolve after a regime change. We show that drift first makes the quantile partition inefficient: cumulative refitting preserves tail support and coverage but cannot recover conditional resolution, and pays for it in channel use, whereas discarding history starves the tail estimator. We therefore propose event-triggered quantile-space adaptation. A nonparametric cumulative sum (CUSUM) on the predictor's level stream detects change. At each confirmed event, the quantile partition is re-derived from post-change observations, retained history is re-binned under it, and each state is refit from a bounded per-state memory, capped to balance stale-regime bias against tail-estimation variance, with a closed-form L-moment tail estimator. At a nominal coverage target η = 0.99, relative to matched cumulative refitting with fixed states, the proposed method reduces channel use by $5.7−−13.2\%$ and quantile pinball loss by $25−−50\%$ across abrupt, incremental, compound, downward, and reversal drift, with coverage changing by −0.5 to +1.0 percentage points. At matched achieved coverage, it saves $4−−18\%$ of channel uses, while resource use in load-only and stationary cases changes by less than 0.5%. With K samples per state and L states, predictor memory is bounded at O(KL) rather than growing as O(t).
Related
- How Wrong Can a Good Predictor Be? Diverging Updates with Vanishing Predictive KL
- Comparing SGLD and a fixed-noise predictor-corrector adaptation in canonical joint energy-based models on CIFAR-10
- PRB-WM: Physics-Regularized Beta World Models for Predictive Link Adaptation on Real Time-Varying Underwater Acoustic Channels
- Predictor-Based Stabilization of a Delayed Hetero-Dimensional Shoreline--Bathymetry System
- Scalable Variable Selection under Predictor Dependence with Adaptive Virtual Dummies
BibTeX
@misc{mahmood2026driftresilient,
title = {{Drift-Resilient Predictive Interference Management: Online Adaptation of an Extreme-Value Predictor}},
author = {Aamir Mahmood and Fateme Salehi and Nurul Huda Mahmood and Song Han},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.08132},
eprint = {2611.08132},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.08132},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Drift-Resilient Predictive Interference Management: Online Adaptation of an Extreme-Value Predictor
AU - Aamir Mahmood
AU - Fateme Salehi
AU - Nurul Huda Mahmood
AU - Song Han
PY - 2026
DA - 2026/11/05
PB - arXiv
DO - 10.48550/arxiv.2611.08132
UR - https://arxiv.org/abs/2611.08132
AB - Predictive interference management supports ultra-reliable low-latency communication (URLLC) by allocating finite-blocklength resources against a high conditional quantile of next-slot interference. Building on an extreme-value-theory/kernel-density-estimation (EVT/KDE) predictor with quantile-defined interference states, this paper makes the predictor resilient to unannounced shifts in the interference distribution. Its cumulative refitting on all past observations does not specify how the state partition and retained samples should evolve after a regime change. We show that drift first makes the quantile partition inefficient: cumulative refitting preserves tail support and coverage but cannot recover conditional resolution, and pays for it in channel use, whereas discarding history starves the tail estimator. We therefore propose event-triggered quantile-space adaptation. A nonparametric cumulative sum (CUSUM) on the predictor's level stream detects change. At each confirmed event, the quantile partition is re-derived from post-change observations, retained history is re-binned under it, and each state is refit from a bounded per-state memory, capped to balance stale-regime bias against tail-estimation variance, with a closed-form L-moment tail estimator. At a nominal coverage target η = 0.99, relative to matched cumulative refitting with fixed states, the proposed method reduces channel use by $5.7−−13.2\%$ and quantile pinball loss by $25−−50\%$ across abrupt, incremental, compound, downward, and reversal drift, with coverage changing by −0.5 to +1.0 percentage points. At matched achieved coverage, it saves $4−−18\%$ of channel uses, while resource use in load-only and stationary cases changes by less than 0.5%. With K samples per state and L states, predictor memory is bounded at O(KL) rather than growing as O(t).
ER -