Technical Summary: A Cancellation Mechanism in AFDM Radar Sensing

1. Problem Statement

Integrated Sensing and Communication (ISAC) for sixth-generation (6G) networks faces significant challenges in high-mobility scenarios, such as vehicular-to-everything (V2X) and low-altitude unmanned aerial vehicle (UAV) operations. Conventional Orthogonal Frequency Division Multiplexing (OFDM) waveforms suffer from severe inter-carrier interference (ICI) due to large Doppler shifts, degrading both communication reliability and sensing accuracy. While Affine Frequency Division Multiplexing (AFDM) has been proposed as a multicarrier waveform capable of handling high mobility through a tunable quadratic phase (chirp) and a chirp-periodic prefix (CPP), its fundamental sensing limits remain poorly understood.

Existing literature on AFDM sensing relies heavily on numerical Cramér-Rao bounds (CRB) or focuses on algorithmic design. Crucially, no closed-form Fisher Information Matrix (FIM) analysis has yet revealed how the AFDM chirp structure shapes delay-Doppler estimation accuracy, why AFDM exhibits weaker delay-Doppler coupling than OFDM, or how these advantages scale with the chirp rate. Furthermore, the specific structural role of the CPP—traditionally viewed only as a channel-equalization device—in decoupling sensing parameters has not been analytically characterized.

2. Methodology

The paper employs a rigorous analytical approach to derive exact closed-form expressions for the Fisher Information Matrix (FIM) and Cramér-Rao bounds (CRB) for joint delay and Doppler estimation in AFDM systems.

• Signal Model: The analysis utilizes the standard AFDM signal model where the time-domain signal is generated via an inverse Discrete Affine Fourier Transform (IDAFT) with a chirp-periodic prefix (CPP). The sensing channel is modeled as a single-target, doubly dispersive channel with complex gain, delay, and Doppler shift.
• Derivative Decomposition: The core of the methodology involves a detailed decomposition of the partial derivative of the log-likelihood function with respect to the delay parameter ($\tau_0$). The authors identify that this derivative naturally decomposes into four distinct physical paths:
  1. The global phase term ($\Xi$).
  2. The shift in the Dirichlet kernel due to the equivalent delay ($l_{eq}$).
  3. The scaling of the CPP phase jump ($\Phi_{jump}$) by the fractional delay.
  4. The migration of the CPP segment boundaries.
• Cancellation Mechanism: The authors demonstrate that the frequency drift introduced by the chirp modulation (Path 2) is exactly compensated by the phase jump of the CPP (Path 3a). This cancellation reduces the magnitude of the combined term from $O(c_1 N)$ to $O(1)$, leaving only a small residual quantization error.
• FIM Derivation: Exploiting this cancellation, the authors derive the FIM elements. They show that the three dominant terms in the expansion of the delay self-information ($I_{\tau\tau}$) cancel exactly, collapsing the multi-parameter dependence into a single scalar inflation factor, $\eta$.
• Frameworks: The analysis is conducted under two frameworks:
  1. Channel-Known: Assuming the complex channel gain is known.
  2. Phase-Profiled: Treating the channel phase as unknown and applying Schur complementation to eliminate it.

3. Key Contributions

A. Identification of the Cancellation Mechanism

The paper identifies a structural cancellation in the AFDM delay score function. The frequency drift caused by the chirp modulation is precisely neutralized by the phase jump inherent in the CPP. This mechanism reduces the delay derivative to a single residual term of order $1/C$ (where $C$ is the number of chirp wrapping cycles). To the authors' knowledge, this is the first identification of such a structural cancellation in a chirp-based multicarrier waveform.

B. Exact Closed-Form Fisher Information Matrix

Leveraging the cancellation, the authors derive an exact closed-form $2 \times 2$ FIM for joint delay-Doppler estimation. The entire dependence on the AFDM chirp structure ($c_1, c_2, N$) collapses into a single scalar parameter, $\eta \ge 1$.

• Delay Information ($I_{\tau\tau}$): Scales with $\eta$.
• Doppler Information ($I_{ff}$): Remains independent of the chirp rate $c_1$, identical to the classical OFDM result.
• Cross-Coupling ($I_{\tau f}$): Determined by the residual structure.

C. Proof of Structural Decoupling

The paper proves that for any non-zero chirp rate, the delay-Doppler coupling coefficient $\rho$ of AFDM is strictly smaller than that of OFDM ($\rho_{AFDM} = 3/(4\sqrt{\eta}) < 3/4 = \rho_{OFDM}$). This advantage is structural, not a result of parameter tuning, and is fully captured by the scalar $\eta$.

D. Closed-Form Cramér-Rao Bounds

The authors obtain closed-form CRBs for both frameworks:

• Delay CRB: Improves quadratically with the chirp rate (scaling approximately as $c_1^{-2}$ in the asymptotic regime).
• Doppler CRB: Is unaffected by the chirp rate.
• Continuity: The framework reduces continuously to the classical OFDM CRB as the chirp rate vanishes ($c_1 \to 0$), certifying AFDM as a strict generalization of OFDM sensing theory.

E. Exact Decoupling via Phase Profiling

In the phase-profiled framework, the authors show that Schur complementation yields an exact delay-Doppler decoupling ($I_{\tau f}^{eff} = 0$). This eliminates the coupling penalty entirely, regardless of the chirp rate.

4. Results and Validation

• Numerical Verification: Extensive numerical experiments using realistic V2X and low-Earth-orbit (LEO) parameters validate all closed-form expressions.
• Cancellation Verification: Numerical evaluation confirms that the cancellation mechanism reduces the energy of the delay derivative by approximately 76% compared to the pre-cancellation chirp-induced drift, and reduces the delay-Doppler derivative correlation from 1.0 (pre-cancellation) to ~0.34.
• Performance Gains:
  • At the communication-optimal configuration ($C = 2\alpha_{max} + 1 = 9$), the AFDM delay CRB is approximately four times better than OFDM.
  • As $C$ increases to 20, the delay CRB improves by a factor of roughly 8.6 compared to OFDM.
  • The Doppler CRB remains constant across all chirp rates, confirming the dimensional separation.
• Scaling Laws: The delay CRB scales as $c_1^{-2}$ for large $C$, while the Doppler CRB is invariant. The paper also analyzes $N$-scaling under fixed $c_1$ (CRB $\propto N^{-2}$) versus fixed $C$ (CRB $\propto N^{-0.2}$).

5. Significance and Claims

The paper claims that its work fundamentally shifts the understanding of AFDM sensing:

1. Role of CPP: It identifies the chirp-periodic prefix (CPP) not merely as a channel-equalization device for communication, but as the structural element responsible for decoupling delay and Doppler in sensing.
2. Analytical Characterization: It demonstrates that AFDM's superior sensing performance can be characterized analytically through closed-form bounds, rather than relying solely on numerical simulations.
3. Design Implications: The results provide ISAC system designers with a "clean dimensional separation": the chirp rate can be tuned purely to improve delay resolution (via $\eta$) without incurring a penalty in Doppler estimation accuracy.
4. Generalization: The framework is presented as a strict generalization of OFDM radar sensing theory, bridging the gap between classical OFDM and advanced chirp-based multicarrier systems.

The authors conclude that the cancellation mechanism is the key to AFDM's ability to maintain favorable sensing limits in high-mobility regimes, offering a theoretical foundation for future waveform design and estimator development.