pith. sign in

arxiv: 2604.12620 · v1 · submitted 2026-04-14 · 📡 eess.SP

Joint Activity Detection and Channel Estimation for Massive Random Access Using SBL and SCA

Esa Ollila , Majdoddin Esfandiari , Daniel P. Palomar This is my paper

Pith reviewed 2026-05-10 14:54 UTC · model grok-4.3

classification 📡 eess.SP
keywords joint activity detectionchannel estimationsparse Bayesian learningsuccessive convex approximationmassive random accesscovariance learninggrant-free accessmMTC
0
0 comments X

The pith

A covariance-learning SBL method with SCA estimates sparse device powers to jointly detect activity and recover channels in massive random access.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper addresses the challenge of joint activity detection and channel estimation when a massive number of devices with sporadic traffic access a base station without grants. It develops a covariance-learning sparse Bayesian learning approach that applies successive convex approximation to partially linearize and minimize the scaled negative log-likelihood of the observed data. This produces estimates of the sparse vector of device signal powers, from which active devices are identified. Empirical Bayesian estimation then yields the channel estimates for those devices. The resulting CL-SCA procedure is shown through simulations to be efficient and to outperform prior methods.

Core claim

By embedding successive convex approximation inside a covariance-learning sparse Bayesian learning model, the scaled negative log-likelihood can be minimized to recover the sparse signal-power vector; active devices are then read off from the support of this vector and their channels are obtained via empirical Bayes.

What carries the argument

The CL-SCA procedure, which uses successive convex approximation to linearize the non-convex negative log-likelihood of the sample covariance and thereby estimate the sparse power vector.

If this is right

  • Active devices can be identified directly from the support of the estimated power vector without separate detection steps.
  • Channel estimates follow from standard empirical Bayesian formulas once the active set is known.
  • The method scales to large numbers of devices because only a sparse power vector is optimized.
  • Simulation comparisons establish lower error rates and faster run times than existing JADCE algorithms.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the sparse-power model holds under realistic traffic, the approach could lower pilot overhead in grant-free mMTC deployments.
  • The same SCA linearization of the log-likelihood could be reused for other covariance-based sparse estimation tasks.
  • Replacing the empirical-Bayes channel step with a joint estimator might further tighten performance bounds.

Load-bearing premise

Successive convex approximation finds a sufficiently accurate minimum of the non-convex negative log-likelihood and the assumed sparse activity model matches real device traffic.

What would settle it

Simulations in which device activity is dense rather than sparse, or in which the SCA iterates fail to produce power estimates whose support matches the true active set, would show the claimed superiority disappearing.

Figures

Figures reproduced from arXiv: 2604.12620 by Daniel P. Palomar, Esa Ollila, Majdoddin Esfandiari.

Figure 1
Figure 1. Figure 1: Probability of missed detection vs. M for different numbers of active devices (K) when L = 20, 30, and L = 50 from bottom to top, respectively; replacement for each Monte-Carlo (MC) trial. The number of MC trials is 10000. The number of MTDs is N = 300. The LSFCs (βn-s in (1)) are uniformly distributed between [−15, 0] in dB scale. B. Discussion of results We investigate the effects of L, M, and K on the p… view at source ↗
Figure 2
Figure 2. Figure 2: Average running time vs. the number of active devices [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Channel estimation NMSE vs. the number of antennas [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
read the original abstract

In massive machine-type communication (mMTC) applications, a key challenge is joint device activity detection and channel estimation (JADCE) under grant-free random access, as a massive number of devices with sporadic traffic seek to connect to the base station. We address JADCE for massive random access using a covariance learning-based sparse Bayesian learning (SBL) approach. Specifically, we first use the successive convex approximation (SCA) framework to partially linearize the scaled negative log-likelihood function (LLF) of the data, then minimize it to estimate the sparse vector of devices' signal powers. After identifying active devices from these power estimates, empirical Bayesian estimation is used to obtain channel estimates. Simulation results demonstrate the efficiency and performance superiority of the proposed CL-SCA method compared to other existing methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript proposes a covariance learning-based sparse Bayesian learning (SBL) approach using successive convex approximation (SCA) for joint activity detection and channel estimation (JADCE) in grant-free massive random access. It first applies SCA to partially linearize the scaled negative log-likelihood of covariance observations to recover a sparse vector of device signal powers, thresholds these estimates to detect active devices, and then applies empirical Bayes estimation for the channels. The central claim is that simulations demonstrate the efficiency and performance superiority of the proposed CL-SCA method over existing approaches.

Significance. If the SCA procedure produces power estimates close to the global minimum of the original non-convex likelihood, the method could offer a practical, covariance-based alternative for JADCE in mMTC with sporadic traffic. The integration of SBL with SCA for power estimation is a reasonable adaptation to the high-dimensional sparse setting. However, the asserted simulation superiority cannot be fully assessed without details on baselines or statistical validation, and the lack of convergence analysis for SCA limits the strength of the contribution.

major comments (1)
  1. [SCA-based optimization description] The SCA framework is used to partially linearize the scaled negative log-likelihood function of the covariance observations before minimization to obtain device powers. No verification is supplied that the surrogate is convex, matches the original function and gradient at each iterate, or satisfies a sufficient-decrease condition. This is load-bearing for the central claim because biased or suboptimal power estimates would propagate directly into activity detection thresholds and channel estimates, potentially rendering the reported gains artifacts of the particular simulation regime rather than a general property of the estimator.
minor comments (2)
  1. [Abstract] The abstract asserts simulation superiority but supplies no information on the specific baselines, SNR ranges, number of devices, Monte Carlo trials, error bars, or statistical tests used to support the claim.
  2. [Method overview] The workflow description would benefit from an explicit statement of how the empirical Bayes channel estimation step depends on the accuracy of the preceding power estimates and activity decisions.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the thorough review and constructive feedback. We address the major comment below and have revised the manuscript to incorporate the suggested clarifications on the SCA procedure.

read point-by-point responses

  1. Referee: The SCA framework is used to partially linearize the scaled negative log-likelihood function of the covariance observations before minimization to obtain device powers. No verification is supplied that the surrogate is convex, matches the original function and gradient at each iterate, or satisfies a sufficient-decrease condition. This is load-bearing for the central claim because biased or suboptimal power estimates would propagate directly into activity detection thresholds and channel estimates, potentially rendering the reported gains artifacts of the particular simulation regime rather than a general property of the estimator.

    Authors: We agree that explicit verification of the SCA surrogate properties is necessary to support the claims. In the revised manuscript, we have added a new subsection (Section III-B) that constructs the surrogate function explicitly, proves its convexity as a quadratic approximation, demonstrates that it matches the original scaled negative log-likelihood and its gradient at the current iterate, and verifies the sufficient-decrease condition via the standard Lipschitz-gradient assumption on the non-convex LLF. These properties follow the general SCA framework of Razaviyayn et al. (2013) and ensure that the iterates converge to a stationary point of the original problem. We have also expanded the simulation section with additional baseline details (including exact parameter settings for all compared methods) and 500 Monte Carlo trials with reported standard deviations to allow statistical assessment of the performance gains. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation or claims

full rationale

The paper derives a covariance-learning SBL estimator by applying SCA to partially linearize and minimize the scaled negative log-likelihood of covariance observations, yielding power estimates that are then thresholded for activity detection before empirical-Bayes channel estimation. None of these steps reduce by construction to the inputs: the SCA surrogate is an approximation whose convergence properties are external to the final performance metric, the activity threshold is a post-processing rule, and the superiority claim rests on independent simulation comparisons against other methods rather than on any fitted parameter being relabeled as a prediction. No self-citations, uniqueness theorems, or ansatzes imported from prior author work are load-bearing in the provided chain. The workflow is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review yields no explicit free parameters, axioms, or invented entities; the approach implicitly rests on standard sparse activity and Gaussian noise assumptions common to SBL literature.

pith-pipeline@v0.9.0 · 5441 in / 1068 out tokens · 60742 ms · 2026-05-10T14:54:41.527148+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

10 extracted references · 10 canonical work pages

  1. [1]

    Improved scaling law for ac- tivity detection in massive MIMO systems,

    S. Haghighatshoar, P. Jung, and G. Caire, “Improved scaling law for ac- tivity detection in massive MIMO systems,” in2018 IEEE International Symposium on Information Theory (ISIT). IEEE, 2018, pp. 381–385

    work page 2018

  2. [2]

    Covariance based joint activity and data detection for massive random access with massive mimo,

    Z. Chen, F. Sohrabi, Y .-F. Liu, and W. Yu, “Covariance based joint activity and data detection for massive random access with massive mimo,” inICC 2019, 2019, pp. 1–6

    work page 2019

  3. [3]

    Non-Bayesian ac- tivity detection, large-scale fading coefficient estimation, and unsourced random access with a massive MIMO receiver,

    A. Fengler, S. Haghighatshoar, P. Jung, and G. Caire, “Non-Bayesian ac- tivity detection, large-scale fading coefficient estimation, and unsourced random access with a massive MIMO receiver,”IEEE Trans. Inf. Theory, vol. 67, no. 5, pp. 2925–2951, 2021

    work page 2021

  4. [4]

    Phase transition analysis for covariance-based massive random access with massive mimo,

    Z. Chen, F. Sohrabi, Y .-F. Liu, and W. Yu, “Phase transition analysis for covariance-based massive random access with massive mimo,”IEEE Transactions on Information Theory, vol. 68, no. 3, pp. 1696–1715, 2021

    work page 2021

  5. [5]

    Activity detection for massive random access using covariance-based matching pursuit,

    L. Marata, E. Ollila, and H. Alves, “Activity detection for massive random access using covariance-based matching pursuit,”IEEE Trans- actions on V ehicular Technology, vol. 74, no. 11, pp. 17 292–17 303, 2025

    work page 2025

  6. [6]

    Robust activity detection for massive random access,

    X. Wang, E. Ollila, and S. A. V orobyov, “Robust activity detection for massive random access,”IEEE Transactions on Signal Processing, vol. 73, pp. 3513 – 3527, 2025

    work page 2025

  7. [7]

    Structured compressed sensing: From theory to applications,

    M. F. Duarte and Y . C. Eldar, “Structured compressed sensing: From theory to applications,”IEEE Trans. Signal Process., vol. 59, no. 9, pp. 4053–4085, 2011

    work page 2011

  8. [8]

    Decomposition by partial linearization: Parallel optimization of multi- agent systems,

    G. Scutari, F. Facchinei, P. Song, D. P. Palomar, and J.-S. Pang, “Decomposition by partial linearization: Parallel optimization of multi- agent systems,”IEEE Transactions on Signal Processing, vol. 62, no. 3, pp. 641–656, 2013

    work page 2013

  9. [9]

    An empirical Bayesian strategy for solving the simultaneous sparse approximation problem,

    D. P. Wipf and B. D. Rao, “An empirical Bayesian strategy for solving the simultaneous sparse approximation problem,”IEEE Transactions on Signal Processing, vol. 55, no. 7, pp. 3704–3716, 2007

    work page 2007

  10. [10]

    Parallel and distributed successive convex approximation methods for big-data optimization,

    G. Scutari and Y . Sun, “Parallel and distributed successive convex approximation methods for big-data optimization,” inMulti-Agent Op- timization: Cetraro, Italy 2014. Springer, 2018, pp. 141–308

    work page 2014