[Paper Review] Compressive Demodulation of Mutually Interfering Signals

Yuejie Chi, Yao Xie+1 more|arXiv (Cornell University)|Mar 15, 2013

Sparse and Compressive Sensing Techniques22 references3 citations

TL;DR

This paper proposes a compressive demodulation framework for detecting sparse, mutually interfering signals in asynchronous multi-user systems using compressive sensing. By unifying two front-end architectures into a single discrete signal model, it enables iterative matching pursuit algorithms that require only O(K log N(τ+1)) samples—significantly fewer than the N(τ+1) samples required by conventional MUD—while achieving robust performance with Kerdock codes outperforming Gabor frames in error rate and sample efficiency.

ABSTRACT

Multi-User Detection is fundamental not only to cellular wireless communication but also to Radio-Frequency Identification (RFID) technology that supports supply chain management. The challenge of Multi-user Detection (MUD) is that of demodulating mutually interfering signals, and the two biggest impediments are the asynchronous character of random access and the lack of channel state information. Given that at any time instant the number of active users is typically small, the promise of Compressive Sensing (CS) is the demodulation of sparse superpositions of signature waveforms from very few measurements. This paper begins by unifying two front-end architectures proposed for MUD by showing that both lead to the same discrete signal model. Algorithms are presented for coherent and noncoherent detection that are based on iterative matching pursuit. Noncoherent detection is all that is needed in the application to RFID technology where it is only the identity of the active users that is required. The coherent detector is also able to recover the transmitted symbols. It is shown that compressive demodulation requires $\mathcal{O}(K\log N(τ+1))$ samples to recover $K$ active users whereas standard MUD requires $N(τ+1)$ samples to process $N$ total users with a maximal delay $τ$. Performance guarantees are derived for both coherent and noncoherent detection that are identical in the way they scale with number of active users. The power profile of the active users is shown to be less important than the SNR of the weakest user. Gabor frames and Kerdock codes are proposed as signature waveforms and numerical examples demonstrate the superior performance of Kerdock codes - the same probability of error with less than half the samples.

Motivation & Objective

• Address the challenge of multi-user detection (MUD) in asynchronous, noncoherent systems with unknown channel state information (CSI), particularly in RFID and wireless networks.
• Overcome the limitations of conventional MUD, which requires N(τ+1) samples for N users and maximal delay τ, by leveraging compressive sensing to reduce sampling requirements.
• Develop a unified signal model that unifies two front-end architectures—random subsampling and generalized matched filters—for compressive demodulation.
• Enable both coherent and noncoherent detection via iterative matching pursuit, with noncoherent detection sufficient for RFID applications where only user identity matters.
• Evaluate and compare signature waveforms—Gabor frames and Kerdock codes—on performance and sample efficiency in sparse recovery under noise.

Proposed method

• Formulate a discrete signal model that unifies random subsampling and generalized matched filter front-ends, reducing the number of required measurements M below N(τ+1).
• Propose iterative matching pursuit algorithms for both coherent and noncoherent detection, enabling recovery of active users and their transmitted symbols.
• Use Gabor frames and Kerdock codes as signature waveforms, with Kerdock codes shown to offer superior performance due to favorable coherence properties.
• Derive performance guarantees for both detection types, showing identical scaling with respect to the number of active users K, independent of power profile.
• Analyze coherence properties of subsampled Gabor and Kerdock frames, proving that subsampled Kerdock frames maintain low average coherence with high probability.
• Apply Hoeffding’s inequality and union bounds to derive probabilistic bounds on the worst-case and average coherence of the subsampled measurement matrix.

Experimental results

Research questions

• RQ1 Can two disparate front-end architectures for compressive demodulation be unified under a single discrete signal model?
• RQ2 How does the number of required samples scale with K (active users), N (total users), and τ (maximum delay) in compressive MUD compared to conventional MUD?
• RQ3 What is the impact of signature waveform design—specifically Gabor frames versus Kerdock codes—on the performance and sample efficiency of compressive demodulation?
• RQ4 How does the signal-to-noise ratio (SNR) of the weakest active user compare in importance to the power profile of stronger users in recovery performance?
• RQ5 Can noncoherent detection via iterative matching pursuit achieve reliable user identity recovery in RFID systems without CSI, and how does it compare to coherent detection?

Key findings

• Compressive demodulation requires only O(K log N(τ+1)) samples to recover K active users, representing a significant reduction from the N(τ+1) samples required by conventional MUD.
• Noncoherent detection via iterative matching pursuit achieves the same performance scaling with respect to K as coherent detection, despite not requiring CSI.
• The performance of the system is more sensitive to the SNR of the weakest active user than to the power profile of stronger users.
• Kerdock codes outperform Gabor frames, achieving the same probability of error with less than half the number of samples due to their superior coherence properties.
• Theoretical analysis shows that subsampled Kerdock frames maintain average coherence ν(X) ≤ 2/M with high probability, enabling stable sparse recovery.
• Numerical results confirm that Kerdock codes provide superior error-rate performance compared to Gabor frames under the same sampling rate and noise conditions.