Probabilistic Constellation Shaping for OFDM ISAC Signals Under Temporal-Frequency Filtering

Basic Information

• Paper ID: 2510.12204
• Title: Probabilistic Constellation Shaping for OFDM ISAC Signals Under Temporal-Frequency Filtering
• Authors: Zhen Du, Jingjing Xu, Yifeng Xiong, Jie Wang, Musa Furkan Keskin, Henk Wymeersch, Fan Liu, Shi Jin
• Category: eess.SP (Signal Processing)
• Publication Date: October 14, 2025 (arXiv preprint)
• Paper Link: https://arxiv.org/abs/2510.12204v1

Abstract

Integrated sensing and communication (ISAC) is considered an innovative technology for sixth-generation (6G) wireless networks, where utilizing orthogonal frequency-division multiplexing (OFDM) communication signals for sensing provides a cost-effective solution for ISAC implementation. However, the sensing performance of matched filtering and mismatched filtering schemes significantly deteriorates due to the signal randomness introduced by finite-alphabet modulation (such as QAM constellations) with non-constant envelope. Therefore, improving sensing performance without significantly compromising communication capability remains a challenging task. To address this, the paper proposes a unified probabilistic constellation shaping (PCS) framework that maximizes communication rate under mean squared error (MSE), power, and probability distribution constraints on the sensing channel state information (CSI), compatible with both matched and mismatched filtering schemes.

Research Background and Motivation

Problem Background

1. ISAC Technology Requirements: 6G wireless networks need to support application scenarios such as smart cities, intelligent transportation, and low-altitude economy, which require ultra-reliable and high-precision localization sensing services.
2. Dual Challenges of OFDM Signals:
   • Communication systems typically employ QAM-modulated OFDM signals to achieve greater Euclidean distance and improve communication reliability
   • Sensing systems prefer flat signal spectra (such as PSK modulation) to achieve high-resolution mainlobes and low sidelobes
3. Limitations of Existing Methods:
   • Existing PCS methods (such as reference 17) are only applicable to matched filters and cannot be compatible with mismatched filtering schemes
   • Lack of unified sensing performance evaluation metrics
   • Insufficient research on transmitter-side optimization, primarily focused on receiver design

Research Motivation

This paper aims to address how to mitigate the negative impact of OFDM signal randomness on sensing performance while maintaining communication capability, and achieve compatibility with different filtering schemes.

Core Contributions

1. Establishes Relationships Between Sensing Performance Metrics: Derives mathematical relationships between MSE of sensing CSI, output signal-to-noise ratio (SNRout), and integrated sidelobe ratio (ISLR), proving that MSE is a more comprehensive sensing performance evaluation metric.
2. Proposes Unified PCS Framework: Constructs a unified PCS optimization framework compatible with matched filtering (MF), reciprocal filtering (RF), and Wiener filtering (WF), employing MSE and achievable information rate (AIR) as sensing and communication performance metrics.
3. Enables Flexible Sensing-Communication Tradeoff: Through optimization of constellation input distribution, achieves customizable sensing-communication performance tradeoffs according to actual QoS requirements.
4. Validates Practical Applicability: Verifies the effectiveness and credibility of the proposed PCS method in practical 6G applications through simulations and field experiments.

Methodology Details

Task Definition

Input: OFDM ISAC signals containing N subcarriers and M symbols Output: Optimized constellation probability distribution achieving sensing-communication performance tradeoff Constraints: MSE constraint, power constraint, probability distribution constraint

System Model

1. Transmitted ISAC Signal Model

The OFDM ISAC signal is expressed as:

s(t) = (1/√M) ∑[m=0 to M-1] s_m(t - mT_sym)

where s_m(t) is the m-th OFDM symbol and x_{n,m} is the frequency-domain data symbol.

2. Sensing Signal Model

Reflected echo signal:

y_s(t) = ∑[q=1 to Q] α_q s(t-τ_q)e^{j2πν_q t} + z(t)

After symbol synchronization and cyclic prefix removal, the observation data matrix is:

Y_s = H ⊙ X + Z

3. Filtering Schemes

Filtering matrices G for three filtering schemes:

• Matched Filtering (MF): G = X*
• Reciprocal Filtering (RF): G = 1/X
• Wiener Filtering (WF): G = X*/(|X|² + SNR_in^{-1})

Sensing Performance Metrics Analysis

MSE Derivation

The MSE of sensing CSI is defined as:

E{ε²} = E{||Ĥ - H||²_F}

Through derivation, it is obtained:

E{ε²} = σ²_α E{∑_{n,m} (χ_{n,m} - 1)²} + σ² E{∑_{n,m} |g_{n,m}|²}

Key Theorem

Theorem 1: The relationship between MSE, SNRout, and ISLR:

σ²_α [ISLR + (NME{(1-r(0,0)/√NM)²})/E{r²(0,0)}] + 1/SNRout = E{ε²}/E{r²(0,0)}

This indicates that MSE comprehensively considers sidelobe response, mainlobe energy loss, and noise amplification effects.

Unified PCS Optimization Framework

Optimization Problem P2

max_{p_x} max_{q_{x|y_c}} F(p_x, q_{x|y_c})
s.t. C1: E{ε²} ≤ c_0
     C2: ∑_x p(x)|x|² = 1
     C3: ∑_x p(x) = 1  
     C4: 0 ≤ p(x) ≤ 1

Modified Blahut-Arimoto Algorithm

1. Update Posterior Probability:

q^{(ℓ)}(x|y_c) = p^{(ℓ)}(x)p(y_c|x) / ∑_{x'} p^{(ℓ)}(x')p(y_c|x')

2. Update Input Distribution:

p^{(ℓ+1)}(x) = e^{∑_{y_c} p(y_c|x) log q^{(ℓ)}(x|y_c) - λ₁f(x) - λ₂|x|²} / Z

where f(x) corresponds to the MSE constraint function for different filtering schemes.

Experimental Setup

Simulation Parameters

• OFDM Parameters: N=64 subcarriers, M=32 symbols
• Modulation Schemes: 64-QAM, 64-PSK
• Filtering Schemes: MF, RF, WF
• SNR Range: -10dB to 25dB

Field Experiments

• Equipment: USRP prototype system
• Parameters: Carrier frequency 3GHz, bandwidth 100MHz, transmit power 20dBm
• Scenario: Static corner reflector, distance 15 meters
• OFDM Configuration: 100 symbols, 1024 subcarriers

Evaluation Metrics

1. Sensing Metrics: Dynamic range (DR), MSE, NMSE, detection probability
2. Communication Metrics: Achievable information rate (AIR)
3. Integrated Metrics: Sensing-communication tradeoff curves

Experimental Results

Main Findings

1. Performance Under Different SNR Conditions

• Low SNR (-10dB): MF and WF perform similarly, superior to RF; PCS optimization significantly improves RF performance
• Medium SNR (4dB): PCS provides 2-4dB performance gain
• High SNR (20dB): RF and WF perform similarly, superior to MF; PCS improves all schemes

2. Filtering Scheme Comparison

Table 1 summarizes the performance metrics of three filtering schemes:

3. Sensing-Communication Tradeoff

• PCS achieves flexible sensing-communication performance tradeoff
• WF approaches the sensing-communication performance boundary under medium SNR conditions
• All schemes outperform time-division strategies

Field Experiment Validation

Field experiment results confirm:

• PCS optimization provides 2-3dB sensing gain
• Both range and velocity profiles are improved
• WF performs best in practical environments

Related Work

ISAC Technology Development

1. Communication-Centric Design: Directly utilizing existing communication signals for sensing, compatible with 5G NR protocols
2. Sensing-Centric Design: Specifically designing sensing waveforms
3. Joint Design: Simultaneously optimizing sensing and communication performance

PCS Technology Evolution

1. Traditional PCS: Primarily used to minimize the gap between AIR and Shannon capacity
2. PCS in ISAC: Reference 17 first applied PCS to ISAC, but limited to matched filtering
3. This Paper's Contribution: Extended to mismatched filtering schemes, providing a unified framework

Conclusions and Discussion

Main Conclusions

1. MSE is a more comprehensive sensing performance metric, comprehensively considering sidelobes, mainlobe loss, and noise amplification
2. Unified PCS framework successfully accommodates MF and MMF schemes
3. Wiener filtering performs optimally under medium SNR conditions
4. Field experiments validate the practical applicability of the PCS method

Limitations

1. Idealized Assumptions: Neglects range migration effects and inter-carrier interference
2. Computational Complexity: Convergence and computational overhead of the MBA algorithm
3. Practical Deployment: Requires consideration of channel estimation errors and hardware imperfections

Future Directions

1. Extension to MIMO-OFDM Systems
2. Consideration of High-Mobility Environments with Coherent Processing
3. Integration with Machine Learning for PCS Optimization
4. Practical 6G Network Deployment Verification

In-Depth Evaluation

Strengths

1. Significant Theoretical Contributions: First establishes mathematical relationships between MSE, SNRout, and ISLR, providing new perspectives for sensing performance evaluation
2. Strong Method Innovation: The proposed unified PCS framework overcomes the limitation of existing methods applicable only to matched filtering
3. Comprehensive Experimental Validation: Combines simulations and field experiments to demonstrate method effectiveness and practicality
4. Rigorous Mathematical Analysis: Detailed derivation process and complete theorem proofs

Weaknesses

1. Idealized Assumptions: Neglects some non-ideal factors in practical systems
2. Insufficient Computational Complexity Analysis: Lacks detailed analysis of computational complexity and convergence conditions of the MBA algorithm
3. Limited Comparison Methods: Primarily compares with uniform constellations, lacking comparison with other ISAC optimization methods

Impact

1. Academic Value: Provides new theoretical frameworks and optimization tools for ISAC system design
2. Practical Value: Can be precomputed offline and deployed online, suitable for practical 6G network applications
3. Reproducibility: Provides detailed algorithm steps and experimental parameters

Applicable Scenarios

1. 6G Wireless Networks: Scenarios requiring integrated sensing-communication such as smart cities and intelligent transportation
2. Radar Communication Systems: Dual-functional systems requiring balanced sensing and communication performance
3. Internet of Things Applications: Efficient sensing-communication solutions in resource-constrained environments

References

The paper cites 38 relevant references covering key works in ISAC technology, OFDM signal processing, probabilistic constellation shaping, and other critical domains, providing a solid theoretical foundation for the research.

Overall Assessment: This is a high-quality academic paper that makes important contributions to ISAC system optimization. The theoretical analysis is in-depth, experimental validation is comprehensive, and it possesses strong academic value and practical potential.