✗ Failed on benchmark 2026

AoI Water-Filling for Neural Data Refresh

Implementation & benchmark of arXiv:2608.21596 — Age-Optimal Target Wake Time: Provably Good Wake Schedules for Energy-Constrained Wi-Fi Status Updating

Usefulness7/10

Difficulty4/10

Novelty7/10

Optimization Accuracy Scalability Speedup Federated World model Inference speedup Stochastic processes Probability Convex analysis Optimization

Source paper: Age-Optimal Target Wake Time: Provably Good Wake Schedules for Energy-Constrained Wi-Fi Status Updating arXiv:2608.21596 ⓘ · analyzed Sep 1, 2026

AI-generated research hypothesis, automatically tested. Not peer-reviewed.

Idea description

Use the renewal Age of Information model to schedule refreshes from heterogeneous federated clients, sensors, retrieval indexes, or world-model observation streams. Sources with high downstream importance and reliable, cheap updates receive shorter refresh periods, while unreliable or expensive sources are refreshed less often. Pack the resulting requests into a non-overlapping communication schedule.

Formulas

$$\bar{A}_i=\delta_i+T_i\left(\frac{1}{p_i}-\frac{1}{2}\right),\qquad \bar{A}^{\mathrm{peak}}_i=\delta_i+\frac{T_i}{p_i}$$

$$p_i=1-(1-q_i)^{k_i},\qquad d_i=k_is_i+c_i$$

$$J^*=\min\left\{\sum_i a_iT_i:\sum_i\frac{d_i}{T_i}\leq 1,\;T_i\geq\max(T_i^{\min},d_{\max})\right\}$$

$$T_i(\lambda)=\max\left(T_i^{\min},d_{\max},\sqrt{\frac{\lambda d_i}{a_i}}\right)$$

Mathematical statement

For source i, T_i is its refresh interval, p_i in (0,1] is the probability that a scheduled refresh succeeds, delta_i is its mean delivery delay, a_i > 0 is its neural-task importance weight, and d_i is the duration of one refresh. Proposition 1 gives the stationary time-average age bar A_i = delta_i + T_i(1/p_i - 1/2), using a geometric number G_i of intervals until success with E[G_i] = 1/p_i and E[G_i^2] = (2-p_i)/p_i^2. Since delta_i and p_i are fixed during an allocation step, minimizing weighted age reduces to minimizing sum_i a_i T_i. The paper relaxes non-overlapping scheduling to the density constraint sum_i d_i/T_i <= 1, with T_i >= L_i and L_i = max(T_i^min,d_max), where T_i^min is an application or energy floor and d_max = max_j d_j. Lemma 3 gives T_i(lambda) = max(L_i,sqrt(lambda d_i/a_i)); lambda is found by bisection so the density constraint is tight when it binds. For k_i independent attempts with per-attempt success probability q_i, p_i = 1-(1-q_i)^k_i. If an attempt takes duration s_i and has fixed overhead c_i, d_i = k_i s_i + c_i.

Implementation notes

(1) Integration point: add the allocator between the training server and communication or observation sources. Start with federated learning, where source i is client i and a refresh is a client update. Do not modify the model or optimizer; modify only which clients are contacted at each round. (2) Maintain exponential moving averages of each client's success rate q_i, response delay delta_i, update duration s_i, and importance a_i. Estimate a_i by periodically withholding that client's newest update and measuring the increase in a held-out validation loss; clip and normalize the values. Choose k_i and c_i, compute p_i = 1-(1-q_i)^k_i and d_i = k_i s_i+c_i, and set L_i = max(T_i^min,max_j d_j). Find lambda with bisection: compute T_i(lambda) = max(L_i,sqrt(lambda d_i/a_i)); evaluate D(lambda)=sum_i d_i/T_i(lambda); increase lambda if D(lambda)>1 and decrease it otherwise. Schedule client i every T_i time units. (3) Convert intervals to integer round numbers. To avoid collisions, round periods upward to a power-of-two multiple of a slot width g, sort clients by increasing period, and assign each occurrence to the earliest free slot. If placement fails, increase that client's period to the next quantized value. Recompute every M rounds. The formulas are exact for the estimated parameters; q_i, delta_i, s_i, and a_i are measured online. (4) First experiment: use a 20-client FEMNIST or CIFAR-10 federated CNN with heterogeneous client reliability and update sizes. Compare uniform polling, data-volume-greedy polling, and this allocator at equal communication bytes. Track accuracy, wall-clock time, weighted update age, and successful updates. Success means lower weighted age and either higher accuracy at equal bytes or equal accuracy with fewer communication rounds. Ablate reliability weighting, importance weighting, water-filling, and slot packing.

Verification

Failed on benchmark

Stage 1 · Toy mechanism gate: Passed ✓

Stage 2 · Mechanism transferred to benchmark: Not reproduced ✗

Stage 2 · Practical benchmark result: Significantly worse ✗

Methodology: 8 paired seeds, permutation test (p<0.05) against a learning-rate-tuned baseline. How verification works

Stage 1 — Mechanism check agent confidence 9/10

Built a readable AoI water-filling allocator with bisection, renewal-age simulation, and duration-aware power-of-two slot packing in aoi_experiment.py. The mechanism manifested: the renewal formula matched Monte Carlo with 0.85% mean relative error, importance scaling matched exactly, and the floor-density transition matched all five sweep cases. In the heterogeneous toy comparison, water-filling reduced theoretical weighted age from 9.702 to 9.152 (5.7%) at density 1.0; this is a mathematical scheduling signal, not evidence of a neural accuracy win.

Agent confidence: 9/10

Baseline: Uniform polling: T=1.33 for all six sources, weighted theoretical age=9.70175.
Idea: Water-filling: T=[1.5033,1.4062,1.5033,1.0270,1.1364,1.7628], lambda=11.29986, density=1.0000000000000002, weighted theoretical age=9.15164; 5.67% lower than uniform. Renewal sweep mean relative error=0.00850; importance objective ratios observed [0.25,0.5,2,4] exactly; floor-transition predictions matched 5/5.

Limitations: Only a synthetic six-source relaxed-density experiment was tested; no FEMNIST/CIFAR federated training, learned importance estimates, wall-clock communication, dynamic nonstationary reliability, or accuracy impact was evaluated. The packed schedule is a small greedy demonstrator and its quantized periods are more conservative than the continuous water-filling solution.

How to run: python3 aoi_experiment.py > results.json

Stage 2 — Benchmark vs. tuned baseline (arXiv:2608.21596) Sep 2, 2026

Significantly WORSE than the tuned baseline (+37.2%, p=0.0081, wins 0 / 8 paired seeds).

Benchmark: Tabular regression (Friedman)
Model: mlp_tiny
Paired seeds: 8
Baseline mean: 14.3738
Idea mean: 19.7252
Effect (Δ): 5.3514 (+37.2%; negative = idea better)
Wins: 0 / 8 paired seeds
p-value: 0.0081 (permutation test, 20 000 shuffles)
Smallest detectable effect: ±33.7%
Mechanism: Not confirmed ✗
Practical effect: Hurts
Baseline tuning: swept over 6 configs

Limitations:

This tested the registered tabular proxy rather than a true federated dataset because no built-in track directly represents client refresh. Real networking bytes, wall-clock communication, online EMA updates, and collision-free packed schedules were not tested. The signature was not confirmed: the integer period conversion and density accounting need correction before claiming quantitative mechanism transfer.

How to run: python3 aoi_bench.py

Verdict computed by deterministic test code from paired-seed statistics — not by the language model.

Stage 2 — Benchmark Sep 1, 2026 latest Failed ✗

Agent confidence: 9/10

The AoI water-filling scheduler was implemented as a client-selection intervention with the shared tabular MLP/FedAvg system and evaluated using the registered bench tabular track. It performed significantly worse than the tuned baseline: paired delta +5.3514 MSE with permutation p=0.0081, so the bench verdict is idea worse (significant) and worked=false.

Baseline: Tuned baseline mean test MSE 14.373831629753113 over 8 seeds; best configuration lr=0.01, local_epochs=2, rounds=12.
Delta: Idea minus baseline = +5.351370453834534 MSE; idea wins 0/8 seeds; permutation p=0.0081.
Idea: AoI mean test MSE 19.725202083587646 over 8 seeds; configuration lr=0.01, local_epochs=2, rounds=12, k=2, tmin=1.0.

Benchmark result