Ieee Transactions On Signal Processing

IEEE Transactions on Signal Processing (ieee-transactions-on-signal-processing)

Journal positioning

IEEE Transactions on Signal Processing is the archival venue for the theory and methods of signal processing: estimation, detection, sampling and reconstruction, filtering, spectral analysis, and array, graph, and statistical signal processing, together with the optimization machinery that underpins them. The defining expectation is a new signal-processing method with analysis — an algorithm whose behavior is characterized (consistency, convergence, performance bounds, identifiability) or that demonstrably outperforms correct, current baselines under a clear model. A generic machine-learning paper, or an application study that merely runs an existing pipeline on new data, is a poor fit. This skill is a fit / venue-selection / re-framing tool. It does not replace the journal's current official author guidelines. Before submitting, re-check the live IEEE Transactions on Signal Processing author information and submission system.

When to trigger

• The author names this journal for an estimation, detection, sampling, or array/graph signal-processing manuscript and wants a fit/framing check.
• A method must be re-framed from "our algorithm gives good results" into a result with a signal model, analysis, and the right baselines.
• The author is choosing between this Transactions and an applications-oriented SP venue, a machine-learning venue, or ieee-transactions-on-communications.
• The author needs the SP-theoretic framing bar and desk-reject heuristics specific to signal-processing methods.

Scope & topic fit

• Statistical signal processing: parameter estimation, detection and hypothesis testing, Cramér–Rao-type bounds, Bayesian and robust estimation.
• Sampling, reconstruction, and sparse/compressive methods; sampling theory beyond Nyquist; dictionary and subspace methods with recovery guarantees.
• Array and sensor-array processing: beamforming, direction-of-arrival, source localization, distributed and networked sensing.
• Graph signal processing and signal processing over networks: spectral methods, filtering, and sampling on graphs.
• Optimization for signal processing: convex/nonconvex methods, ADMM and proximal algorithms, with convergence or optimality analysis tied to an SP problem.

Method & evidence bar

• The contribution is a method with analysis: an estimator/detector/algorithm whose properties (bias/variance, consistency, convergence rate, identifiability, recovery conditions) are characterized, or a clearly superior empirical result.
• Baselines must be the correct, current competitors under the same signal model and conditions; beating a strawman or an outdated method is not evidence.
• When the claim is empirical, experiments must use realistic models, report variance across trials, and isolate the source of improvement; when the claim is theoretical, proofs must be complete.
• The signal/observation model and assumptions must be explicit; performance claims must hold under those assumptions, with sensitivity to model mismatch discussed.
• Position precisely against prior SP results: tighter bound, weaker assumptions, lower complexity, or a genuinely new processing principle.

Structure & house style

• IEEE format; the journal publishes full Regular Papers and shorter Correspondence items — match the article type to the contribution and re-check current definitions on the live guide.
• The signal model and problem statement come early and precisely; the proposed method and its analysis are the core, with derivations in-text or in appendices.
• The introduction motivates the SP gap, not an application novelty; relate the method to classical SP theory.
• Figures are analytical and comparative: MSE-vs-SNR curves, ROC curves, convergence plots, and benchmark comparisons under matched conditions.