⚡ electrical engineering

Airtime- and QoS-Penalty-Aware Measurement Coordination in Multi-Router Wi-Fi Sensing Networks

This paper proposes AQ-MC, a controller framework for multi-router Wi-Fi sensing networks that dynamically coordinates measurement rates and active node subsets to ensure sensing activities remain within strict airtime and quality-of-service (QoS) penalty budgets, thereby enabling feasible coexistence with ordinary WLAN traffic.

Original authors: Volodymyr Pavlenko

Published 2026-07-09

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Volodymyr Pavlenko

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a busy, multi-lane highway where two types of vehicles are trying to drive: regular cars (your normal Wi-Fi traffic like streaming videos or browsing the web) and sensing drones (specialized Wi-Fi signals sent out just to detect if someone is moving in a room).

Usually, these "sensing drones" just fly around as much as they want, hoping they don't crash into the regular cars. But in a crowded network, if too many drones fly at high speed, they clog the road, causing traffic jams for everyone.

This paper introduces a new Traffic Controller called AQ-MC (Airtime- and QoS-Penalty-Aware Measurement Coordination). Instead of letting the drones fly wild, AQ-MC acts like a smart air traffic controller that decides exactly which drones can fly, how fast they should go, and when they must land, all to make sure the regular cars keep moving smoothly.

Here is how it works, broken down into simple concepts:

1. The Problem: Too Many Drones, Not Enough Sky

In a house with multiple Wi-Fi routers, each router can act as a "drone" to sense motion. If you turn on all of them at their fastest speed, they use up all the "airtime" (the time the wireless signal is active). This leaves no room for your Netflix or Zoom calls, causing them to lag or drop.

2. The Solution: The Smart Traffic Controller (AQ-MC)

AQ-MC doesn't just guess; it uses a strict set of rules (a budget) to decide what is allowed. It looks at two main things before letting a sensing drone take off:

The Airtime Budget: "How much of the road will this drone use?"
The QoS Penalty (The 'Traffic Jam' Score): "How bad will the traffic get if we add this drone right now?"

The Analogy: Imagine you are at a party.

Airtime is how loud you are speaking.
QoS Penalty is how crowded the room is.
If the room is empty (low load), you can shout a little louder (higher sensing rate) without bothering anyone.
If the room is packed (high load), even a whisper might disturb the conversation. AQ-MC calculates this "crowdedness" and tells the sensing drones to be quiet or stay home if the room is too full.

3. How the Controller Makes Decisions

The controller looks at every router and gives it a "score" based on how useful it is right now (e.g., is the signal changing because someone is moving?).

High Score + Low Traffic: The router gets the "High Speed" tier (10 times per second).
Medium Score: It gets the "Medium Speed" tier.
Low Score or High Traffic: It gets "Low Speed" or is told to turn off completely.

Crucially, the controller checks its Budget before saying "Go." If turning on a high-speed drone would break the budget (use too much airtime or cause too much traffic pressure), the controller says "No" and rejects that configuration. It doesn't try to fix the problem later; it prevents the problem before it starts.

4. What the Paper Actually Found

The author tested this system in a lab with real Wi-Fi routers and simulated traffic. Here are the key takeaways from the results:

It Works as a Gatekeeper: The system successfully admitted 1,483 decisions where the sensing worked without breaking the budget.
It Saves Space: In one test, the controller decided to turn off one of three routers (a 33% reduction in active sensors) because keeping all three on would have been too expensive for the network.
It Keeps Traffic Flowing: When the system was under heavy load, the controller kept the sensing "airtime" very low (around 0.15% to 0.22% of the total time), ensuring it didn't interfere with regular internet use.
The "All-High" Mistake: The researchers tried a "dumb" approach where they tried to keep all routers on at maximum speed. They found that this "dumb" approach would have failed (caused traffic jams) in 74.71% of the scenarios. AQ-MC successfully rejected those bad ideas before they could happen.

5. What This Is NOT

It is important to understand what this paper doesn't claim:

It does not claim to make the motion detection more accurate (it doesn't say you can see people better).
It does not claim to make your internet faster (it just promises not to slow it down).
It is not a new Wi-Fi standard (like 802.11bf); it is a software "manager" that runs on top of existing routers.

Summary

Think of AQ-MC as a smart bouncer for a Wi-Fi network. Its only job is to check the ID of every sensing signal, calculate if the club (the network) is too crowded, and decide who gets in and how loud they can be. If the club is full, the bouncer turns away the extra sensing signals to ensure the regular guests (your video calls and downloads) can still enjoy the party.

Technical Summary: Airtime- and QoS-Penalty-Aware Measurement Coordination in Multi-Router Wi-Fi Sensing Networks

Problem Statement
Wi-Fi sensing utilizes Channel State Information (CSI), Received Signal Strength Indicator (RSSI), and packet-level descriptors to detect motion, presence, and environmental changes without attached sensors. While multi-router (multi-access point) systems can improve sensing robustness by fusing data from multiple radio paths, a critical operational gap exists: how to coordinate the active subset of sensing nodes and their measurement rates without degrading ordinary Wireless Local Area Network (WLAN) traffic.

Existing resource-aware sensing methods focus on compression, packet-rate reduction, or specific link selection, but they do not define a controller-level admission rule that explicitly balances sensing utility against two strict constraints: sensing airtime and load-weighted Quality of Service (QoS) penalty. In operational networks, sensing packets share the medium with client traffic; an unmanaged increase in sensing activity can violate service-level agreements or degrade user experience. The paper addresses the specific problem of admitting an active sensing configuration only if it remains within pre-configured airtime and QoS budgets.

Methodology: AQ-MC
The paper proposes Airtime- and QoS-Penalty-Aware Measurement Coordination (AQ-MC), a controller-based admission mechanism for multi-router Wi-Fi sensing. AQ-MC operates as a short admission loop at fixed decision intervals (e.g., every 2 seconds) and functions as follows:

State and Inputs: The controller receives node-side feature windows (RSSI variation, frequency-domain RSSI spread, and node confidence) and the current network load profile.
Utility Scoring and Tier Assignment:
- Nodes are scored using a weighted utility function: $u_{i,t} = 0.45\sigma_{i,t} + 0.35\tilde{\rho}_{i,t} + 0.20\gamma_{i,t}$ .
- Based on utility scores and binary node-side triggers, nodes are assigned a measurement-rate tier: Off, Low (2 Hz), Medium (5 Hz), or High (10 Hz).
Budget Prediction:
- Airtime ( $A$ ): Predicted as a percentage of PHY airtime based on the number of active nodes, their assigned rates, sample size (512 bytes), and PHY rate (e.g., 54 Mbps).
- QoS Penalty ( $D$ ): Calculated as $D(ct) = w(L_t) \times A(ct)$ , where $w(L_t)$ is a dimensionless weight (0.20 to 0.85) corresponding to the current network load (Light to High). This ensures that sensing airtime is penalized more heavily when the network is already under high load.
Admission Decision: The controller checks if the candidate configuration satisfies both $A(ct) \le A_{max}$ $A (c t) \leq A_{ma x}$ and $D(ct) \le D_{max}$ $D (c t) \leq D_{ma x}$ .
- The controller uses a greedy, utility-ordered approach: it sorts candidates by utility and tentatively activates them. If a candidate violates a budget, it is rejected (or downgraded), and the next candidate is tested.
- Only configurations satisfying both budgets are admitted. If no configuration is feasible, sensing is effectively disabled for that interval.

Key Contributions

AQ-MC Framework: A formal method for coordinating active sensing subsets and measurement-rate tiers under explicit airtime and load-weighted QoS-penalty budgets.
Controller Admission Logic: A decision rule that treats sensing as a managed process, outputting admission feasibility, predicted airtime, predicted QoS penalty, and the specific reason for rejection (airtime vs. QoS limit) rather than optimizing for sensing accuracy.
Empirical Evaluation:
- Active-Subset Experiment: Demonstrated a 33.33% reduction in active nodes (selecting 2 of 3) to maintain feasibility.
- Protected-Traffic Experiments: Two experiments (moderate and high load) produced 1,483 accepted decisions with high retention of the selected node set (hold ratio > 99.86%).
- Deterministic Stress Grid: A simulation of 25,088 cases varying node counts, PHY rates, and budgets.
Boundary Analysis: Identification of operational boundaries where resource-oblivious policies fail (e.g., an "all-high" policy was feasible in only 25.29% of stress cases).

Results

Feasibility: In the deterministic stress grid, a resource-oblivious "all-high" multi-node policy was feasible in only 25.29% of cases. AQ-MC successfully rejected the remaining 74.71% of over-budget attempts before actuation.
Protected-Traffic Performance: In experiments with a 10% airtime budget and 5% QoS-penalty budget:
- Moderate Load (2 nodes): Mean predicted airtime was 0.151% (1.51% of budget); Mean QoS penalty was 0.053% (1.06% of budget).
- Higher Load (3 nodes): Mean predicted airtime was 0.227% (2.27% of budget); Mean QoS penalty was 0.125% (2.50% of budget).
- The controller maintained stable configurations with minimal tier changes (0.257–0.409 changes per minute).
Load Sensitivity: Increasing the load weight from "Light" to "High" reduced the feasible share of configurations from 41.76% to 30.61%, demonstrating the necessity of the load-weighted penalty metric.
PHY Rate Impact: At lower PHY rates (e.g., 6 Mbps) and tight budgets (0.2%), feasibility dropped significantly (3.12%), highlighting the method's sensitivity to physical layer constraints.

Significance and Claims
The paper explicitly frames its contribution as a controller-admission method rather than a sensing-classification improvement or a latency-reduction technique.

Scope: The work is motivated by IEEE 802.11bf WLAN sensing but remains a local controller-admission method, not a standards-compliant Medium Access Control (MAC) or Physical Layer (PHY) implementation.
Primary Claim: AQ-MC provides a bounded admission mechanism that enables the coexistence of cooperative Wi-Fi sensing with ordinary WLAN traffic by enforcing explicit airtime and QoS-penalty limits.
Modesty of Claims: The authors state that the reported outcome is controller admission feasibility, not sensing-classifier accuracy, measured latency improvement, cross-domain robustness, or global optimization quality. The method ensures that sensing does not exceed configured resource budgets, trading off sensing coverage (by reducing active nodes) to preserve network service quality.
Limitations: The study acknowledges that greedy admission may miss higher-utility feasible subsets and that the stress simulation is formula-derived, not modeling complex network dynamics like retransmissions or queueing. Future work is suggested to pair AQ-MC with task-specific metrics and measured QoS calibration.