IEEE Transactions on Signal Processing Submission Guide

Official pages tell you where to submit, how to format, and which EDICS categories exist. They do not tell you whether your manuscript is a TSP paper or an application-specific paper that belongs in TIP, TASLP, TCI, TMLCN, SPL, or another venue.

This guide tells you what IEEE TSP editors look for before review. The review tells you whether your paper passes the TSP-specific readiness checks that official IEEE instructions cannot evaluate from a generic ScholarOne checklist. Paid Manusights reviews include a 60-day money-back guarantee; submitted manuscripts are not used for model training.

If you want the quick pre-upload call, run an IEEE Transactions on Signal Processing manuscript fit check before submitting through ScholarOne.

For a broader check before choosing an IEEE Signal Processing Society route, use the Manusights AI manuscript review and compare the feedback against TSP, TIP, TASLP, TCI, TMLCN, OJ-SP, and Signal Processing Letters.

TSP uses EDICS classifications to route submissions. This is not a minor metadata step. EDICS tell the editorial system what technical community should evaluate the paper. If the EDICS selection is broad, fashionable, or mismatched, the manuscript can be routed to reviewers who see the contribution as off-center.

The practical rule: choose EDICS that describe the core signal-processing contribution, not the application surface. For example, if the real contribution is a new distributed estimator, do not make the EDICS look like a biomedical or wireless application paper just because those are the data examples.

Regular manuscripts submitted to TSP should not exceed 13 double-column pages, including references and author biographies. Correspondence submissions should not exceed 2 double-column pages, including references.

The final-publication cost rule is separate. Final manuscripts longer than 10 printed pages incur mandatory overlength charges. Authors should not treat the 13-page submission limit as permission to write a loose paper. A strong TSP paper usually uses the page budget to make the theorem, algorithm, proof sketch, assumptions, experiments, and limitations easy to audit.

TSP is the SPS theory-and-methods journal. It is strongest for papers where the contribution generalizes beyond one application.

That can include:

• detection and estimation theory

• statistical signal processing

• sampling, compressed sensing, sparse and low-rank recovery

• optimization methods for signal processing

• graph signal processing

• distributed and federated signal processing

• adaptive filtering and array processing

• information-theoretic or learning-based signal-processing methods

• signal-processing methods for data science when the SP contribution is clear

The application can matter, but it should not be the only reason the paper exists.

The wrong submission is often not low quality. It is simply pointed at a reviewer community that is not the natural audience for the claim.

TSP maintains a reproducible research page and encourages papers whose results can be verified through code, data, or clear simulation procedures. The current page lists many reproducible TSP papers, which is a signal about editorial culture even when reproducibility is not the only acceptance criterion.

For authors, this means the manuscript should make enough detail visible for reviewers to test the claim: assumptions, parameter settings, baselines, simulation protocol, datasets, code availability where feasible, and limitations. A theory paper still needs numerical experiments that are not selected only to favor the proposed method.

This is what editors check before review: whether the submission is new and in scope, whether the 13-page initial file includes enough assumptions, proof logic, and reproducibility detail to audit the claim, and whether the selected EDICS describe the method rather than only the application dataset.

The editorial policy states the venue fit in unusually concrete terms: novelty and appropriateness both matter, EDICS routing is part of the review architecture, and reproducibility detail helps editors decide whether a theory or method claim can be audited.

General SP contribution. A strong application result is not enough if the method does not teach the signal-processing community something general.

Assumptions and guarantees. TSP readers look hard at model assumptions, identifiability, convergence, optimality, sample complexity, robustness, and what the theory does not cover.

Experimental discipline. Baselines should be current and fair. Synthetic experiments should expose failure modes, not only success cases. Real-data examples should test whether the method remains useful outside the authors' ideal model.

Venue fit. If the paper is mainly image, audio, speech, computational imaging, biomedical, wireless, radar, or controls work, the manuscript must show why TSP is still the right home.

TSP's editorial flow follows IEEE Signal Processing Society policies and what TSP authors report through community channels. Treat as planning ranges, not promises.

• Day 0: ScholarOne upload. The ScholarOne submission portal portal accepts the package, runs page-limit and EDICS checks, and routes to a Senior Area Editor.

• Days 1 to 14: Administrative review and editor assignment. Editorial staff verify EDICS classification, 13-page initial limit, IEEE format compliance, and originality declarations; the Associate Editor evaluates scope fit.

• Days 14 to 30: Reviewer invitations. TSP typically invites three reviewers with EDICS-matched expertise. Finding theory-heavy reviewers in fast-moving subfields can extend the timeline.

• Days 30 to 120: Peer review. Reviewer reports return on a 6 to 12 week cadence; proof-heavy papers extend the timeline because reviewers verify derivations and convergence claims line by line.

• Days 90 to 150: First editorial decision. Major revision is the most common outcome for papers that pass desk review. Outright acceptance is rare at first decision.

• Days 150 to 360: Revision rounds and acceptance. Single-revision acceptances run roughly 8 to 10 months; multi-round revisions push closer to 12 months.

Before submitting to TSP, prepare the full IEEE submission package:

• Manuscript in IEEE Transactions format, within the 13-page double-column initial limit

• Cover letter explaining the signal-processing contribution and EDICS rationale

• ORCID identifiers for all authors (required for IEEE submissions)

• Author contributions statement following IEEE author-role guidance

• Funding statement disclosing grants, sponsor support, or institutional funding

• Conflicts of interest disclosure for all authors

• Ethics statement where human-subjects data, biometric data, or sensitive datasets are involved

• Data and code availability statement; reproducibility is a TSP editorial signal

• Supplementary information for proofs, additional simulation results, or convergence analyses

• Suggested reviewers with EDICS-matched expertise (typically three names)

• Conference predecessor PDF plus a difference-explanation document when the submission extends prior conference work

Across signal-processing manuscripts targeting IEEE Transactions on Signal Processing, three recurring decision risks matter most across submissions that TSP editors filter out at the desk-screen stage.

(Per IEEE SPS published guidelines, TSP enforces a 13-double-column-page submission limit (10-point font, including all elements:

• requires fundamental signal-processing theory or methodology with novelty AND appropriateness as the two acceptance criteria.

• explicitly designates papers as Immediate Rejection candidates when: out-of-scope (signaled by few citations to recent SP publications)

• limited novelty (no new theory / algorithm / experimental data)

• incomplete (insufficient simulation / experiment / bibliography / context)

• or containing primarily mathematical derivations whose results reveal little useful insight or rest on unrealistic assumptions.

EDICS routing determines reviewer assignment and quality screen.) Use the three checks below before you open ScholarOne submission portal upload slot.

Signal dataset leads the abstract

Across TSP-targeted manuscripts, we consistently see authors submit papers where the abstract leads with the application context (biomedical EEG / ECG / fMRI / MEG signals, image / video / hyperspectral data, speech / music / environmental audio, 5G / 6G / Wi-Fi wireless channels, radar / sonar / lidar returns, seismic / geophysical / underwater signals, financial / network / IoT time series) and the signal-processing methodological contribution appears only by section III or IV.

TSP senior area editors specifically check whether the abstract names the general signal-processing problem the paper solves (parameter estimation, detection, classification, source separation, sparse recovery, low-rank approximation, manifold learning, graph signal processing, distributed estimation / consensus, online learning, change-point detection, hypothesis testing, decision fusion, sampling theory, compression / quantization, system identification, adaptive filtering, beamforming, MIMO signal processing) rather than the application domain, and whether the contribution is the signal model / estimator / algorithm / bound / convergence rate / sample-complexity result rather than the application performance.

Manuscripts with application-first framing get redirected within 2-3 weeks: biomedical-signal papers to IEEE Transactions on Biomedical Engineering / TMI / JBHI, image / video papers to IEEE TIP / TPAMI / TCSVT, audio / speech papers to IEEE/ACM TASLP, wireless / communications papers to IEEE TWC / T-COM / TCCN / TVT, radar / sonar papers to IEEE TAES / OE, sensor-network papers to IEEE TSIPN / IoT-J, ML-application papers to IEEE TNNLS / TPAMI.

The fix is to rewrite the first abstract sentence to name the general signal-processing problem, demote the application domain to a single demonstration paragraph in the experiments section, ensure the introduction cites 15+ recent TSP / TSP-aligned papers showing the SP-community context, and ensure the contribution statement is methodological (estimator, algorithm, identifiability result, performance bound) rather than empirical.

Check whether your TSP abstract foregrounds the signal-processing contribution →

Theory-strength language without the assumptions, proof rigor, or numerical evidence to support it

We frequently see TSP manuscripts use theory-strength language in the abstract and introduction (optimal, minimax-optimal, asymptotically optimal, robust, stable, convergent, globally convergent, consistent, asymptotically unbiased, BLUE, ML-optimal, MAP-optimal, MMSE-optimal, sample-complexity-optimal, near-optimal, oracle-optimal, sharp, tight, general) without providing the formal support TSP reviewers expect for those words.

TSP's editorial culture treats these words as load-bearing claims requiring formal analysis: optimality requires a proof against a named benchmark (CRLB / BCRB / Hammersley-Chapman-Robbins / Ziv-Zakai / Bayesian risk / minimax risk over a specified parameter class); robustness requires identification of the contamination model (Huber's epsilon-contamination, heavy-tailed noise with named distribution class, adversarial perturbation with named budget); convergence requires a formal theorem with explicit assumptions, rate, and conditions;

consistency requires a proof under a stated probabilistic model; sample-complexity claims require an upper bound matched to a lower bound (information-theoretic / Fano / Le Cam / Assouad).

Manuscripts that use these words descriptively without formal support face revision-or-reject decisions or get classified as Immediate Rejection candidates under the "primarily mathematical derivations whose results reveal little useful insights" criterion when the math is present but unconnected to a meaningful SP insight.

The fix is to either remove the unsupported claim and rephrase descriptively (e.g., "the method performed reliably across the simulated scenarios" rather than "the method is robust"), or invest in the supporting analysis with theorems, proofs (in appendices), and explicit assumption statements; pair every theorem with a one-paragraph "discussion of assumptions" subsection that explains which assumptions are tight and which can be relaxed.

Check whether your TSP theory claims are supported by assumptions, proofs, and experiments →

EDICS routes to the wrong reviewers

The third recurring pattern in TSP-targeted manuscripts is incorrect EDICS (Editor's Information Classification Scheme) selection at submission.

TSP requires authors to select 1-3 EDICS codes from the published list (statistical signal processing, adaptive systems, sensor array and multichannel signal processing, signal processing for communications and networks, audio and acoustic signal processing, image and multidimensional signal processing, machine learning for signal processing, signal processing theory and methods, biomedical signal processing, signal processing for sensors, signal processing over networks, signal processing for big data, computational imaging, quantum signal processing, and others).

The EDICS selection drives reviewer assignment: a paper coded under "biomedical signal processing" goes to biomedical SP reviewers who apply biomedical-validation standards; a paper coded under "signal processing theory and methods" goes to theory reviewers who apply theorem-and-proof standards; a paper coded under "machine learning for signal processing" goes to ML-SP reviewers who apply ML-benchmark standards.

Authors who code by application (biomedical, communications, audio, image) when the contribution is methodological get reviewers who flag the lack of application-validation rigor; authors who code by method when the contribution is application-specific get reviewers who flag the lack of theoretical depth.

Manuscripts with mismatched EDICS face longer review cycles (4-6 months vs the typical 3-month target) and revision requests that are unfair to the actual contribution.

The fix is to read the published EDICS list at Signalprocessingsociety source page before submission, select the primary EDICS that matches the contribution type (not the application domain), include a secondary EDICS only if the paper genuinely spans two methodological areas, and use the optional cover-letter section to flag the contribution type explicitly so the senior area editor can route correctly even if the EDICS selection is ambiguous.

Check whether your TSP EDICS choices match the manuscript's real contribution →

Check whether your TSP manuscript is submission-ready →

Before submitting to IEEE Transactions on Signal Processing, an IEEE Transactions on Signal Processing submission readiness check identifies whether the package meets the editorial bar before you commit to the submission.

• IEEE Transactions on Signal Processing journal profile

• IEEE Transactions on Wireless Communications submission guide

• IEEE Transactions on Pattern Analysis and Machine Intelligence submission guide

Last verified: May 2026 against IEEE Signal Processing Society materials.

While the manuscript is in peer review, use the companion IEEE Transactions on Signal Processing Under Review status guide to interpret portal movement, follow-up timing, and reviewer-risk preparation without confusing the status page with the submission guide.