[Paper Review] Characterization of Coded Random Access with Compressive Sensing based Multi-User Detection

Yalei Ji, Čedomir Stefanović+3 more|arXiv (Cornell University)|Apr 8, 2014

Indoor and Outdoor Localization Technologies16 references4 citations

TL;DR

This paper proposes a novel integration of coded slotted ALOHA with compressive sensing-based multi-user detection (CS-MUD) to enhance random access performance in massive machine-type communication (mMTC). By leveraging the capture effect and joint activity-data detection via CS-MUD, the scheme achieves significantly higher throughput and user resolution than conventional methods, with optimal performance at low user-to-slot ratios and high SNR.

ABSTRACT

The emergence of Machine-to-Machine (M2M) communication requires new Medium Access Control (MAC) schemes and physical (PHY) layer concepts to support a massive number of access requests. The concept of coded random access, introduced recently, greatly outperforms other random access methods and is inherently capable to take advantage of the capture effect from the PHY layer. Furthermore, at the PHY layer, compressive sensing based multi-user detection (CS-MUD) is a novel technique that exploits sparsity in multi-user detection to achieve a joint activity and data detection. In this paper, we combine coded random access with CS-MUD on the PHY layer and show very promising results for the resulting protocol.

Motivation & Objective

• To address the scalability and efficiency challenges of traditional random access in massive machine-to-machine (M2M) communication systems.
• To model and incorporate the capture effect in coded slotted ALOHA (CSA) systems with compressive sensing-based multi-user detection (CS-MUD).
• To analyze the asymptotic performance of frameless ALOHA when combined with CS-MUD, considering both intra- and inter-slot interference cancellation.
• To optimize system parameters such as user degree and slot degree for maximum throughput and user resolution under realistic fading conditions.

Proposed method

• Adapts and-or tree evaluation to model the iterative decoding process of coded slotted ALOHA, incorporating capture probabilities derived from CS-MUD.
• Uses compressive sensing (CS) to jointly detect active users and their data in a sparse multi-user environment, enabling low-overhead detection.
• Employs numerical evaluation of capture probabilities under Rayleigh fading and varying SNR to model realistic receiver performance.
• Applies the framework to frameless ALOHA, a variant of coded slotted ALOHA, to analyze throughput and user resolution gains.
• Derives key performance metrics such as maximum user resolution probability $P^{ ext{*}}_{\text{R}}$, maximum expected throughput $T^{ ext{*}}$, and optimal average slot degree $\beta^{\text{*}}$.
• Integrates the capture effect into the iterative decoding process by modifying the probability of successful recovery in collision slots based on SNR and number of interferers.

Experimental results

Research questions

• RQ1 How does the capture effect influence the performance of coded slotted ALOHA when combined with CS-MUD?
• RQ2 What is the asymptotic user resolution probability and maximum expected throughput of frameless ALOHA with CS-MUD under different SNR and user-to-slot ratios?
• RQ3 How do the optimal system parameters (e.g., average slot degree $\beta^*$) vary with SNR and user density?
• RQ4 To what extent do intra-slot and inter-slot interference cancellation contribute to throughput gain under high SNR conditions?
• RQ5 How does the performance of the proposed CS-MUD-based scheme compare to conventional CSA without capture or with narrowband capture?

Key findings

• At 10 dB SNR, the maximum expected throughput $T^*$ reaches approximately 24 when $M/N \approx 0.0234$, indicating high spectral efficiency.
• For $10$ dB SNR, the optimal average slot degree $\beta^*$ is about 37, significantly higher than in non-capture or narrowband systems, reflecting the benefit of dense collisions.
• At the critical $M/N$ ratio, the expected number of replicas per user is only 0.87 at 10 dB, indicating that most throughput gain comes from intra-slot interference cancellation.
• At 5 dB SNR, the expected user degree at the critical $M/N$ is 1.83, showing that both intra- and inter-slot IC contribute to performance, unlike at higher SNR.
• The capture probability increases with SNR, widening the range of interfering users $t$ for which successful recovery is likely, especially at $10$ dB.
• The maximum user resolution probability $P^*_{\text{R}}$ increases sharply at first with $M/N$ and then saturates, mirroring the behavior of iterative BP erasure decoding.