Grant-Free Massive Access via Sparse Signal Recovery

18 concepts17 papers1 proposal

Traditional cellular networks are built on a fundamental promise: when a device wants to transmit data, it must first ask for permission. This human-centric paradigm relies on a multi-step handshake protocol. A device sends a scheduling request, the base station returns a grant allocating an exclusive, orthogonal slice of time and frequency, and only then does the transmission occur.

This approach breaks down in the era of the Internet of Things (IoT) and massive Machine-Type Communications (mMTC). When millions of smart meters, environmental sensors, and wearables sporadically wake up to send tiny packets of data, the control signaling overhead required to coordinate these grants completely overwhelms the network. As argued by Gerhard Wunder et al. in Sparse Signal Processing Concepts for Efficient 5G System Design, transitioning to next-generation wireless architectures requires a fundamental shift away from rigid coordination toward "one-shot" transmission protocols.

The solution lies in a powerful paradigm known as grant-free random access. Instead of waiting for an assignment, devices simply transmit their data immediately. To make this work without catastrophic collisions, networks exploit a hidden structural property of machine traffic: activity sparsity.

The Core Insight: Activity Sparsity

While the total pool of potential devices in a cell might be massive ($N = 10^5$ or $10^6$), only a tiny fraction ($K \ll N$) are actually active at any given instant. This sporadic behavior means that if we stack the operational states of all users into a vector, that vector is highly sparse.

Instead of trying to prevent overlapping transmissions, a grant-free network allows them to happen. The base station is then tasked with a mathematical unmixing problem: sorting through a compressed aggregate measurement to determine which devices are active and what channels they are transmitting through. As synthesized by Liang Liu et al. in Sparse Signal Processing for Grant-Free Massive Connectivity: A Future Paradigm for Random Access Protocols in the Internet of Things, this framing transforms massive access from a resource scheduling bottleneck into a compressed sensing problem.

Traditional (Orthogonal): [User 1] -> [Slot 1], [User 2] -> [Slot 2] ... (Scales poorly)
Grant-Free (Sparse):      [Active Users] -> [Simultaneous Transmission] -> [Compressed Sensing Recovery]

Formulating Joint Activity and Channel Estimation

To separate the overlapping signals, every device in the network is assigned a unique, pseudo-random signature sequence (or pilot preamble). When active devices transmit their preambles simultaneously, the received signal matrix $\mathbf{Y}$ at a base station with $M$ antennas over a preamble length of $L$ intervals can be modeled as:

$$\mathbf{Y} = \mathbf{A}\mathbf{X} + \mathbf{W}$$

Where:

  • $\mathbf{A} \in \mathbb{C}^{L \times N}$ is the known sensing matrix containing the unique pilot sequences of all $N$ potential users.
  • $\mathbf{X} \in \mathbb{C}^{N \times M}$ is the channel matrix, where the $i$-th row represents the channel vector of user $i$.
  • $\mathbf{W}$ represents additive noise.

Because only $K$ users are active, the matrix $\mathbf{X}$ is row-sparse: it contains exactly $N - K$ rows that are entirely zero.

The goal of the base station is to recover $\mathbf{X}$ from $\mathbf{Y}$. In traditional systems, to guarantee orthogonal separation, the pilot length $L$ would have to be at least as large as the total number of users ($L \ge N$). Compressed sensing theory flips this constraint entirely. Because $\mathbf{X}$ is sparse, the base station can accurately recover the active rows even when the pilot length is significantly shorter than the user pool ($L \ll N$).

As highlighted by Yongpeng Wu et al. in Massive Access for Future Wireless Communication Systems, this formulation enables Joint Activity Detection and Channel Estimation (JADCE). By solving a single sparse recovery problem, the receiver simultaneously identifies the active transmitters and learns their channel states, eliminating the handshake phase entirely.

The Algorithmic Engine: Approximate Message Passing

While framing massive access as a sparse recovery problem is elegant, solving it in real-time for millions of devices requires algorithms with exceptionally low computational complexity. Standard optimization techniques, such as interior-point methods for $\ell_1$-minimization, scale poorly with the problem dimensions.

To bridge this gap, researchers rely on Approximate Message Passing (AMP) algorithms. Originally derived from loopy belief propagation on factor graphs, AMP is an iterative framework that updates estimates of the sparse matrix $\mathbf{X}$ by passing probabilistic "messages" between nodes representing the received signals and nodes representing the user states.

The intuition behind AMP's success rests on two pillars:

  1. Onsager Reaction Terms: In dense matrices, standard message passing suffers from correlation buildup between iterations. AMP introduces a correction term (the Onsager term) that effectively whitens the residual error, ensuring that the effective noise at each iteration behaves like independent Gaussian noise.
  2. Bayesian Denoising: Because the effective noise is Gaussian, the algorithm can apply a non-linear thresholding function (a denoiser) at each step to suppress the inactive users (forcing near-zero values to absolute zero) while refining the channel estimates of active users.

In Sparse Activity Detection for Massive Connectivity, Zhilin Chen et al. demonstrate that this iterative framework solves the JADCE problem with remarkably low computational overhead, making real-time base station processing viable.

Furthermore, the physical scaling of the hardware interacts beautifully with the algorithm. Liang Liu et al. proved in Massive Connectivity With Massive MIMO - Part I: Device Activity Detection and Channel Estimation that as the number of base station antennas ($M$) grows large, the law of large numbers ensures that the activity detection error drops asymptotically to zero. The spatial dimensions provided by Massive MIMO act as independent looks at the sparse activity pattern, decoupling user identification from the noise floor.

Expanding the Paradigm

The core mechanism of activity-sparsity exploitation has proven highly adaptable, evolving to address diverse architectural challenges across 5G and 6G design:

  • Data-Assisted Recovery: Traditional compressed sensing architectures separate the pilot phase from the data phase. However, Xinyu Bian et al. in Supporting More Active Users for Massive Access via Data-assisted Activity Detection noted that a user's active status is implicitly embedded within their data payload as well. By designing an AMP variant that feeds back data decoding confidence into the activity detection loop, the authors demonstrated a ~20% increase in the number of supported active users without increasing pilot overhead.
  • Satellite IoT Constellations: The paradigm is not limited to terrestrial cells. Keke Ying et al. in Quasi-Synchronous Random Access for Massive MIMO-Based LEO Satellite Constellations adapted sparse recovery to Low Earth Orbit (LEO) satellite networks. Despite the severe Doppler shifts and propagation delays inherent to space-based receivers, treating user access as a quasi-synchronous sparse recovery problem allows satellites to handle massive sensor arrays spanning remote geographic areas.
  • Integrated Sensing and Communication (ISAC): Beyond decoding data, the same sparse reflections can be repurposed. In Integrated Massive Communication and Target Localization in 6G Cell-Free Networks, Junyuan Gao et al. deployed Generalized AMP within a cell-free network to simultaneously identify active communication devices and passively localize physical targets in the environment, demonstrating that sparsity-driven frameworks can unify communication and radar sensing.

By shifting the network's perspective—treating simultaneous uncoordinated transmissions not as destructive collisions, but as a structured, compressed mathematical puzzle—sparse signal recovery provides the theoretical foundation for scaling wireless networks to handle billions of bursty, interconnected devices.

Key Foundational Work

Sparse Signal Processing Concepts for Efficient 5G System Design (2015) — contributed 1 concept elements to this direction.
Efficient and reliable low-power backscatter networks (2012) — contributed 1 concept elements to this direction.
History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 16 papers
  2. Version 1 · 2026-10-04 · 32 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_sparse_signal_recovery_activity_sparsity_bf03efcc,
  title = {sparse signal recovery (activity sparsity exploitation)},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/sparse-signal-recovery-activity-sparsity-exploitation/rdr_bf03efccbfbc4b7782ea19be2361ef15},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 2, frozen text):

@misc{lacuna_sparse_signal_recovery_activity_sparsity_bf03efcc_v2,
  title = {sparse signal recovery (activity sparsity exploitation)},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/sparse-signal-recovery-activity-sparsity-exploitation/rdr_bf03efccbfbc4b7782ea19be2361ef15/v/2},
  note = {Version 2, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake