⚡ electrical engineering

ISAC-Enabled Grant-Free Uplink via Artificial-Path Delay Modulation

This paper proposes a low-complexity, SIC-free grant-free uplink framework for ISAC systems where user equipment conveys data by modulating the delay of a weak artificial path within the cyclic prefix, enabling reliable uplink-downlink coexistence with minimal degradation to scheduled downlink users.

Original authors: Ruiqi Kong, He Chen

Published 2026-06-01

📖 5 min read🧠 Deep dive

CC0 1.0

Original authors: Ruiqi Kong, He Chen

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ 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 highway where a main truck (the Scheduled User) is delivering a heavy load of data to a warehouse (the Access Point). Usually, if a small delivery van (the Grant-Free User) wants to send a message, it has to wait for a traffic light to turn green, or it has to drive right next to the truck and shout its message over the engine noise, which causes a lot of confusion and requires a complex noise-canceling system to hear it.

This paper proposes a clever new way for that small van to send a message without stopping traffic or shouting. Instead of driving its own car, the van acts like a smart mirror that briefly tilts the light beam coming from the main truck.

Here is the breakdown of how this "Artificial-Path Delay Modulation" works, using simple analogies:

1. The Core Idea: The "Echo" Trick

In this system, the small van (Grant-Free User) doesn't generate its own radio signal. Instead, it catches the signal the main truck is already sending and bounces it back to the warehouse.

The Twist: The van doesn't just bounce it back immediately. It uses a special device to delay the bounce by a tiny, precise amount of time.
The Message: The amount of delay represents the message. If the van delays the echo by a tiny bit, it means "0". If it delays it by a slightly larger bit, it means "1", and so on.
The Result: The warehouse (Access Point) listens to the main truck's signal, but it also hears a faint "echo" that arrives a split-second later. By measuring exactly when that echo arrives, the warehouse can read the van's message.

2. Why This is Better Than Old Methods

The Old Way (Power-Domain NOMA):
Imagine the van trying to talk by turning up its volume to match the truck. The warehouse has to use a complex "noise-canceling" headset (called Successive Interference Cancellation or SIC) to subtract the truck's voice to hear the van. This is hard to do, uses a lot of battery, and if the headset fails, both messages get garbled.

The New Way (Delay-Domain Sensing):
The van stays quiet and just tweaks the timing of the echo.

For the Main Truck: The echo is so weak and arrives so quickly (within the "guard time" of the signal) that the truck doesn't even notice it. It's like a whisper arriving just before the next sentence starts; it doesn't ruin the conversation. The truck can keep driving at full speed without needing a noise-canceling headset.
For the Warehouse: The warehouse is equipped with a super-sensitive "sensing radar." It doesn't just listen for volume; it listens for timing. It strips away the main truck's signal and looks at the tiny "ripple" left behind by the van's echo to decode the message.

3. The "Cyclic Prefix" Safety Zone

In wireless communication, there is a safety buffer called the Cyclic Prefix (CP). Think of it as a "grace period" at the end of a sentence where you can pause without the listener getting confused.

The paper ensures the van's echo always happens inside this grace period.

If the echo is too late: It spills into the next sentence, causing a crash (interference).
If the echo is just right (inside the CP): It is invisible to the main truck but perfectly visible to the warehouse's sensing radar.

4. The Challenges and Solutions

The researchers had to solve two main problems:

Problem A: The "Off-Grid" Blur.
Imagine trying to measure a delay that falls exactly between two ticks on a clock. If you only look at the ticks, the measurement looks blurry and spreads out.
- Solution: They created a "calibration" step. Before sending data, the van sends a test signal so the warehouse can learn exactly where the "zero point" is. Then, they use a special mathematical filter (a Normalized Matched-Filter) that can measure delays with extreme precision, even if they don't land perfectly on a clock tick.
Problem B: The Volume Balance.
If the van bounces the signal too loudly, it disturbs the main truck. If it bounces too quietly, the warehouse can't hear it.
- Solution: The paper found a "sweet spot." Even if the van's echo is 15 decibels quieter than the main truck's signal (very faint), the warehouse can still decode it better than the main truck decodes its own data!

5. The Bottom Line

This paper presents a system where:

No waiting: Small devices can send data instantly without asking for permission (Grant-Free).
No interference: The main user doesn't need complex hardware to cancel out the small user's signal.
High efficiency: By using the timing of a reflection rather than the volume, they can send data reliably even when the reflection is very weak.

In short, instead of fighting for space on the radio highway, the small user simply taps the main truck on the shoulder with a perfectly timed "tap-tap" that only the warehouse can hear.

Technical Summary: ISAC-Enabled Grant-Free Uplink via Artificial-Path Delay Modulation

Problem Statement
Current wireless systems rely heavily on Orthogonal Multiple Access (OMA) or Non-Orthogonal Multiple Access (NOMA) schemes. While NOMA offers improved spectrum efficiency through power-domain superposition and Successive Interference Cancellation (SIC), its practical deployment is hindered by high implementation complexity, increased power consumption, and the need for tight receiver coordination. Furthermore, existing 6G physical-layer discussions largely favor OFDM-compatible designs that avoid radical departures from legacy interfaces. Consequently, there is a need for a grant-free uplink mechanism that enables NOMA-like spectrum reuse without requiring multiuser detection at scheduled users or modifying the baseline OFDM downlink waveform.

Methodology
The paper proposes an Integrated Sensing and Communication (ISAC)-enabled grant-free uplink framework based on Artificial-Path Delay Modulation. In this system, a Grant-Free User Equipment (g-UE) does not transmit an independent co-channel waveform. Instead, it reuses the scheduled downlink signal transmitted by an Access Point (AP) to a Scheduled User Equipment (s-UE). The g-UE conveys uplink information by dynamically reflecting the downlink waveform to create a controllable, weak artificial propagation path.

Key technical components include:

Delay-Domain Modulation: The g-UE encodes information by selecting the delay position of a single artificial path. This shifts the uplink-downlink coexistence mechanism from power-domain superposition to the delay-sensing domain.
CP Constraint: The artificial path is constrained to remain within the Cyclic Prefix (CP) duration of the OFDM system. This ensures that the reflection acts as a distinct first-order multipath component, preserving subcarrier orthogonality and causing minimal degradation to the s-UE's standard OFDM reception.
Baseline Calibration and Detection: To address practical challenges such as unknown path gains and off-grid delay leakage (where physical delays do not align with discrete DFT bins), the AP employs a two-step process:
- Calibration: The AP estimates a baseline delay by subtracting a reference natural channel observation from an observation where the g-UE is active with zero delay offset.
- Detection: The AP uses a normalized matched-filter detector over a fine continuous delay grid. This allows the AP to decode the embedded information by identifying the delay hypothesis that maximizes the projection energy, effectively mitigating off-grid leakage without relying on coarse DFT bins.

Key Contributions
The paper outlines four primary contributions:

Downlink-Friendly Mechanism: Introduction of a grant-free uplink scheme where information is embedded in the delay of a controllable artificial path, enabling spectrum reuse via AP-side sensing rather than scheduled-user-side SIC.
CP-Constrained Design: Development of a delay-position modulation design where the feasible delay alphabet is jointly determined by OFDM parameters and the calibrated baseline propagation delay.
Robust Detection Algorithm: Formulation of symbol detection as a normalized matched-filter search over a fine grid, integrated with baseline delay calibration and reference subtraction to handle unknown gains and off-grid effects.
Performance Characterization: Analysis of the Bit Error Rate (BER) trade-offs, demonstrating that reflection power governs the reliability trade-off between the g-UE and s-UE, while the delay step primarily controls the g-UE's reliability-efficiency trade-off with negligible impact on the downlink.

Results
Numerical evaluations using Monte Carlo simulations (with 64 subcarriers and 20 MHz bandwidth) yielded the following findings:

Estimation Accuracy: The normalized matched-filter estimator achieves mean absolute delay estimation errors below $10^{-1}$ sampling periods at -5 dB SNR and reaches $10^{-2}$ levels in medium-to-high SNR regimes.
Reflection Power Trade-off: Increasing the reflection power (from -20 dB to -5 dB relative to the environment signal) significantly improves g-UE reliability (e.g., g-UE BER drops from 0.2255 to 0.0017 at 10 dB SNR). Crucially, when the reflection power is sufficiently low (e.g., -15 dB), the s-UE BER remains nearly identical to the baseline (undisturbed) downlink, confirming the "transparency" of the scheme to scheduled users.
Delay Step Efficiency: Varying the delay step affects g-UE BER but has little impact on s-UE BER. Notably, with a delay step of 0.2 or larger and an SNR of 10 dB or higher, the g-UE achieves a lower BER than the scheduled s-UE (using 16-QAM) while maintaining a modulation order of at least 16-QAM equivalent. This indicates the system can provide competitive spectral efficiency without sacrificing reliability.

Significance and Claims
The authors claim that this framework offers a low-complexity, SIC-free, and downlink-friendly solution for grant-free uplink in ISAC systems. By leveraging the AP's sensing capabilities to decode intentionally induced channel perturbations, the system achieves NOMA-like spectrum reuse without requiring hardware modifications or interference cancellation at the scheduled user. The paper emphasizes that even with an artificial path 15 dB weaker than the scheduled downlink signal, the g-UE can achieve superior reliability compared to the scheduled user, validating the feasibility of delay-domain modulation as a practical alternative to power-domain NOMA for massive access scenarios.