v1.5.0

AntHocNet v1.5.0. See CHANGELOG.md.

Install bundle (anthocnet-1.5.0.zip) — the source to
drop onto your own simulator tree:
make install-ns3 NS3DIR=… or make install-ns2 NS2DIR=….
Pre-built simulator images are on GHCR (see docker/README.md),
including release-pinned tags anthocnet-ns{2,3}:<sim-version>-v1.5.0.

Verify your download. The SHA-256 digest is published twice — in
this body and as the anthocnet-1.5.0.zip.sha256 asset:

23aaad4bf58dae1ceb7729b2c378d97a26e42231ceca4b48c3eaff2185727fa1  anthocnet-1.5.0.zip

sha256sum -c anthocnet-1.5.0.zip.sha256   # both files in the same directory

What's Changed

• docs(roadmap): mark v1.4.0 shipped — tag cut 2026-08-09 by @danieljoppi in #397
• feat(bench): measure re-injection identity and fate directly — the #386 counter by @danieljoppi in #398
• feat(bench): promote the #386 soundness identities and correct the reorder gloss by @danieljoppi in #399
• docs(bench): dupRx is app-layer duplicate traffic, not FlowMonitor inflation by @danieljoppi in #400
• feat(protocol): cap per-packet mac re-injections behind MaxReinjectPerPacket by @danieljoppi in #401
• docs(bench): plan the v1.5.0 re-baseline campaign before spending it by @danieljoppi in #403
• ci(bench): rescue the per-merge benchmark-results artifact, and schedule #229 into phase 0 by @danieljoppi in #406
• fix(bench): cap-aware drop-cause attribution — capped-terminal drops of delivered packets are straddle, not loss by @danieljoppi in #407
• ci(bench): clear rescue-artifacts' stale run_ids default — the API substitutes defaults for empty inputs by @danieljoppi in #408
• fix(bench): count the baselines' pending-queue sheds — dsdv drops a fifth of dense traffic in silence by @danieljoppi in #409
• docs(bench): record phase-0 completion in the v1.5.0 campaign plan by @danieljoppi in #410
• feat(routing): flip ReconvHoldCap default 1 s -> 200 ms by @danieljoppi in #411
• feat(bench): vendor gpsr baseline port for the oracle-phase comparison (#296) by @danieljoppi in #412
• docs(bench): re-publish the grid at the 200 ms re-baseline — v1.5.0 phase 1 complete (#371) by @danieljoppi in #413
• feat(bench): vendor aomdv baseline port for the oracle-phase comparison (#296) — builds, does not yet route by @danieljoppi in #414
• docs(bench): publish the v1.5.0 phase-2 re-injection results (#386) by @danieljoppi in #417
• fix(bench): stop the #230 diversity gate crying wolf, and frame the baseline set (#296) by @danieljoppi in #418
• feat(bench): add the oracle shortest-path control arm (#415, #296 item 1) by @danieljoppi in #419
• fix(bench): dsdv cannot run on multi-ISL grids (#420), and two arm allowlists silently dropped the new arms (#421) by @danieljoppi in #422
• docs(bench): publish the v1.5.0 phase-3 oracle control — the campaign is complete by @danieljoppi in #424
• docs(bench): the gpsr arm delivers zero — correct every record that calls it working (#425) by @danieljoppi in #426
• docs(bench): write the gpsr threat-to-validity text the verdict requires (#425, #296) by @danieljoppi in #427
• docs: v1.5.0 exit criteria are met, and record what the campaign left behind by @danieljoppi in #428

Full Changelog: v1.4.0...v1.5.0

v1.4.0

AntHocNet v1.4.0. See CHANGELOG.md.

Install bundle (anthocnet-1.4.0.zip) — the source to
drop onto your own simulator tree:
make install-ns3 NS3DIR=… or make install-ns2 NS2DIR=….
Pre-built simulator images are on GHCR (see docker/README.md),
including release-pinned tags anthocnet-ns{2,3}:<sim-version>-v1.4.0.

Verify your download. The SHA-256 digest is published twice — in
this body and as the anthocnet-1.4.0.zip.sha256 asset:

540b7bbc828d7ce4c84f00b0512df443e8106507f191a48c94ec8bd7e8f46a21  anthocnet-1.4.0.zip

sha256sum -c anthocnet-1.4.0.zip.sha256   # both files in the same directory

What's Changed

• docs: record the v1.3.0 zenodo doi by @danieljoppi in #325
• feat(bench): paired per-seed verdicts in sweep_summary --vs by @danieljoppi in #326
• feat(core): adopt the thesis's pheromone exponents, betaAnts = betaData = 20 by @danieljoppi in #327
• ci: pin all workflow actions to commit shas and add dependabot by @danieljoppi in #338
• docs: support route, question form, honest bug form, ns-2 notice, codeowners by @danieljoppi in #340
• ci: cancel superseded pr runs, lint the whole skills dir, gate relative links by @danieljoppi in #346
• docs: readme hero — pitch, results figure, architecture diagram by @danieljoppi in #347
• ci: publish the core coverage number by @danieljoppi in #348
• build: devcontainer on the ghcr ns-3 image for one-click codespaces by @danieljoppi in #349
• docs: publish docs/ as a mkdocs-material site on github pages by @danieljoppi in #350
• docs: link the docs site and the codespaces quick start by @danieljoppi in #351
• docs: record the #180 verdict against the proactive rate and emission gate by @danieljoppi in #353
• chore: add spdx and copyright headers across the source tree by @danieljoppi in #354
• fix(bench): pin rng streams per seed so results do not depend on split structure by @danieljoppi in #355
• ci: openssf scorecard, release checksums, and the trust-signal assessment by @danieljoppi in #356
• build(deps): bump actions/checkout from 4.4.0 to 7.0.1 by @dependabot[bot] in #342
• build(deps): bump softprops/action-gh-release from 2.6.2 to 3.0.2 by @dependabot[bot] in #343
• build(deps): bump actions/setup-python from 5.6.0 to 7.0.0 by @dependabot[bot] in #345
• build(deps): bump docker/login-action from 3.7.0 to 4.6.0 by @dependabot[bot] in #344
• chore(deps): bump docker/build-push-action from 6.19.2 to 7.3.0 by @dependabot[bot] in #341
• ci: stop Dependabot bumping the ns-2 image base (#339) by @danieljoppi in #357
• chore(deps-dev): bump ruff from 0.16.0 to 0.16.1 by @dependabot[bot] in #358
• chore(deps): bump amannn/action-semantic-pull-request from 5.5.3 to 6.1.1 by @dependabot[bot] in #359
• chore(deps): bump actions/upload-artifact from 4.6.2 to 7.0.1 by @dependabot[bot] in #360
• fix(bench): pin manet-baselines rng streams and extend the seed-independence gate by @danieljoppi in #361
• feat(bench): emit hopsCommon/hopsSurplus for the #308 phase-2 decomposition by @danieljoppi in #363
• feat(bench): measure channel occupancy per run for the #308 per-hop hypothesis by @danieljoppi in #364
• feat(bench): emit ##HOLD## so pending-queue hold time is reachable, not just measured by @danieljoppi in #367
• docs: mark v1.3.0 shipped and record the provenance pin the corpus already had by @danieljoppi in #368
• feat(bench): emit ##CONFIG## so a run records the configuration it ran with by @danieljoppi in #370
• docs(roadmap): record v1.4.0 progress — the delay-tail verdict is delivered by @danieljoppi in #372
• feat(bench): add a mobility-model knob — steady-state RWP and Gauss-Markov by @danieljoppi in #373
• feat(bench): add a Nakagami fading channel — the other half of the v1.4.0 grid by @danieljoppi in #374
• fix(bench): gauss-markov aborted on NormalPitch — a constant where ns-3 demands a normal by @danieljoppi in #375
• docs(bench): state the provenance of every results page — and the two answers differ by @danieljoppi in #376
• docs(roadmap): the v1.4.0 grid reports — delivery ranking stable, tail ranking is not by @danieljoppi in #378
• docs(bench): publish the v1.4.0 mobility × channel grid, and surface the v1.3.0 DOI by @danieljoppi in #379
• feat(bench): regression floors for the grid arms — and why they are not anchors by @danieljoppi in #380
• feat(bench): add a TCP arm — saturating BulkSend, with ##GOODPUT## as its headline by @danieljoppi in #381
• fix(bench): reorder columns must be absent under TCP, not zero by @danieljoppi in #382
• feat(bench): record the commit every published number was measured at by @danieljoppi in #383
• feat(bench): emit the raw drop-identity counters to test #377's mechanism by @danieljoppi in #384
• fix(bench): the drop-identity overlap must use macDrops, not macTerminal by @danieljoppi in #385
• ci(bench): cap the paper-benchmark step below the job ceiling by @danieljoppi in #387
• fix(bench): carve genuine losses out of the drop residual, not raw MAC drops by @danieljoppi in #388
• docs(roadmap): mark the tcp arm measured and enumerate the v1.4.0 closeout by @danieljoppi in #392
• feat(bench): widen the data-packet predicate to accept tcp on the data port by @danieljoppi in #393
• docs(bench): publish the tcp arm — goodput, paired tests, and the ranking inversion by @danieljoppi in #394
• docs(roadmap): the tcp arm is published and its instrumentation gap is closed by @danieljoppi in #395
• docs: record the v1.4.0 closeout decisions — hold-cap operating point, pins accepted, gate cleared by @danieljoppi in #396

New Contributors

• @dependabot[bot] made their first contribution in #342

Full Changelog: v1.3.0...v1.4.0

v1.3.0

AntHocNet v1.3.0. See CHANGELOG.md.

Install bundle (anthocnet-1.3.0.zip) — the source to
drop onto your own simulator tree:
make install-ns3 NS3DIR=… or make install-ns2 NS2DIR=….
Pre-built simulator images are on GHCR (see docker/README.md),
including release-pinned tags anthocnet-ns{2,3}:<sim-version>-v1.3.0.

What's Changed

• docs: regime comparison tables — networks side by side, mechanism live/inert per regime by @danieljoppi in #299
• docs: software-layer diagrams — stack, ant mechanisms × config switches, per-regime support by @danieljoppi in #303
• docs: ant-type reference, ADR-0019/0020, and orientation updates by @danieljoppi in #304
• docs: twelve diagrams — class, sequence, flow, layout and accounting views of the protocol, metrics and process by @danieljoppi in #305
• fix(docs): unrenderable mermaid block + CI gate, Zenodo DOI, NS-2 deprecation, and the roadmap plan by @danieljoppi in #306
• test(core): exercise GenerationTracker's eviction (#166), and add the matched-delivery delay99 that #308 needs by @danieljoppi in #310
• fix(bench): re-emit ##MATCH## in the compact result block by @danieljoppi in #311
• feat(bench): common-set delay99 — the assumption-free version of #308 by @danieljoppi in #312
• feat(bench): add 95% CIs, paired tests and a column-mapping gate to the result parsers by @danieljoppi in #313
• docs(bench): statistical policy — runs floor, ci methods, warm-up, rng — plus a runs-floor gate by @danieljoppi in #314
• feat(bench): render 95% cis — chart error bars and ± table columns by @danieljoppi in #315
• feat(bench): emit 95% ci columns from --export-sweeps by @danieljoppi in #316
• docs(bench): publish 20-seed pause and area sweeps with 95% cis by @danieljoppi in #317
• ci(bench): per-merge taxonomy at the 10-run publication floor by @danieljoppi in #320
• feat(bench): persist per-run rows in a sibling campaign csv by @danieljoppi in #321
• feat(bench): seed-splitting — firstrun offset threaded end-to-end, exact pooling by @danieljoppi in #322
• docs(bench): publish the 20-seed scale sweep — v1.3.0's sweep set is complete by @danieljoppi in #323
• docs: correct the ns-2 support claim, and consolidate ns-2 into one document by @danieljoppi in #324

Full Changelog: v1.2.0...v1.3.0

v1.2.0

AntHocNet v1.2.0. See CHANGELOG.md.

Install bundle (anthocnet-1.2.0.zip) — the source to
drop onto your own simulator tree:
make install-ns3 NS3DIR=… or make install-ns2 NS2DIR=….
Pre-built simulator images are on GHCR (see docker/README.md),
including release-pinned tags anthocnet-ns{2,3}:<sim-version>-v1.2.0.

What's Changed

• docs: record the per-version Zenodo DOIs, which nothing tracked by @danieljoppi in #191
• feat(bench): measure radio energy for all four protocols by @danieljoppi in #219
• docs(context): stop asserting issue state in prose — every "open" issue it named was closed by @danieljoppi in #220
• feat(bench): measure packet reordering, the documented cost of multipath by @danieljoppi in #221
• feat(bench): break down packet drops by cause, and fix a silent queue eviction found doing it by @danieljoppi in #222
• feat(bench): measure path diversity, path length and flow fairness by @danieljoppi in #223
• feat(ns3): route next hops per interface and add the isl-grid scenario by @danieljoppi in #224
• docs(bench): stop quoting generated numbers in the caveat — it has now rotted four times by @danieljoppi in #225
• docs: survey aco routing prior art on leo constellations by @danieljoppi in #227
• test(ns3): assert isl-grid builds the topology it claims by @danieljoppi in #228
• docs: check the leo survey's section 2 against the source and withdraw a claim by @danieljoppi in #231
• docs: record what the ducatelle thesis answered, and one estimator mismatch it exposed by @danieljoppi in #232
• docs: adr-0015 — one anthocnet build, satellite substrate in the image by @danieljoppi in #236
• test(bench): satellite validation anchors on the isl topology by @danieljoppi in #238
• docs: what protocols actually run on satellites, and what that does to our baselines by @danieljoppi in #239
• docs: why satellite is not manet — diagrams and the differences that bite by @danieljoppi in #240
• feat(core): per-ant-type gates and directed reactive discovery by @danieljoppi in #243
• fix(benchmarks): catch the metric defects the campaigns exposed, and the gate starvation that hid them by @danieljoppi in #242
• docs(adr): record the directed-reactive decision (ADR-0016) and its open hazard by @danieljoppi in #246
• feat(bench): add the thesis's eq 5.1 delay jitter alongside the FlowMonitor column by @danieljoppi in #247
• fix(bench): carry jitter_eq51_ms through the campaign CSV by @danieljoppi in #249
• feat(core): land the proactive emission gate, default OFF, with its experiment by @danieljoppi in #252
• fix(skills): validate ##BENCH## cells' diagnostic lines, not just the table by @danieljoppi in #253
• fix(core): restore proactiveInterval to 10 s — #252 leaked the 2 s rate it claimed not to ship by @danieljoppi in #255
• feat(bench): split the benchmark surface into MANET and satellite views by @danieljoppi in #257
• fix(ns2): add missing ns-default.tcl entries for four tcl-bound parameters by @danieljoppi in #261
• test(core): pin every Config default; add cross-tree parity check by @danieljoppi in #264
• feat(skills): satellite results pipeline — validate, parse and commit isl-grid output by @danieljoppi in #265
• feat(ns3): fast ISL failure detection via interface-down + failcell detect/reconverge split by @danieljoppi in #266
• fix(bench): keep OOM forensics — print and upload time -v + stderr on a failed benchmark run by @danieljoppi in #267
• fix(bench): stream ##RSS## samples from the paper run — forensics that survive runner death by @danieljoppi in #268
• fix(bench): cap sim address space at 13 GiB so an OOM dies gracefully, not with the runner by @danieljoppi in #269
• fix(bench): let --energyJ=0 disable the energy model — the #256 OOM is its cancelled switchToOff events by @danieljoppi in #270
• fix(bench): leak-free energy accounting — integrate the PHY State trace, drop BasicEnergySource + WifiRadioEnergyModel by @danieljoppi in #271
• feat(bench): adopt the dedicated diversity cell — floor-relative gate at the churn-free window by @danieljoppi in #273
• docs: proactiveInterval provenance moves from unknown to measured — the #248 sweep verdict by @danieljoppi in #274
• feat(core): virtual pheromone ages on the regular table's time-proportional clock (#262) by @danieljoppi in #275
• feat(core): the ILinkState congestion queries name the next hop — the stamp moves to where the interface is known (#206) by @danieljoppi in #276
• feat(ns3): per-interface transmit-queue congestion reader for the p2p/ISL regime + ADR-0017 (#206) by @danieljoppi in #277
• test(core): uniformity probe experiment — v/r on the #262 clock, the #180 gate's empirical basis by @danieljoppi in #278
• feat(bench): the asymmetric-congestion corridor cell on the ISL torus — #216 cell 1 by @danieljoppi in #280
• fix(bench): the corridor background transmits at constant duty — the OnOff default off-phase zeroed the congestion signal every other second by @danieljoppi in #282
• fix(core): receptions refresh the 1-hop pheromone at the metric's unloaded 1-hop value — not the legacy constant 1.0 (#279) by @danieljoppi in #283
• feat(core): the proactive emission gate compares per-link — the #180 re-derivation, ADR-0018 by @danieljoppi in #284
• docs: open-source front door + accuracy audit — fix count/default drift, add docs map, ADR index, CI reproducibility by @danieljoppi in #285
• docs: add the PR template — verification and record sections mirror the repo's merge discipline by @danieljoppi in #286
• docs: metrics.md energy section catches up with the #271 model — trace integration, not BasicEnergySource by @danieljoppi in #287
• feat(bench): corridor-cell pheromone sampler — the #216 mechanism trace for arms that fail to shift by @danieljoppi in #288

Full Changelog: v1.1.0...v1.2.0

v1.1.0

AntHocNet v1.1.0. See CHANGELOG.md.

Install bundle (anthocnet-1.1.0.zip) — the source to
drop onto your own simulator tree:
make install-ns3 NS3DIR=… or make install-ns2 NS2DIR=….
Pre-built simulator images are on GHCR (see docker/README.md),
including release-pinned tags anthocnet-ns{2,3}:<sim-version>-v1.1.0.

What's Changed

• docs: record the two-step release / post-release Zenodo DOI procedure by @danieljoppi in #106
• chore(release): exclude .github and .claude from the source archives by @danieljoppi in #108
• docs: add the JOSS software-paper draft (paper/) by @danieljoppi in #113
• feat(bench): add a --propagation override to run-scenarios.py by @danieljoppi in #114
• docs: add author ORCID by @danieljoppi in #115
• fix: correct the Zenodo DOI (concept DOI, README badge + CITATION.cff) by @danieljoppi in #116
• docs(paper): note JOSS submission is deferred, and why by @danieljoppi in #117
• ci(scenario-matrix): raise job timeout, upload partial results on failure by @danieljoppi in #118
• fix(bench): flush the CSV after every point, not just at script exit by @danieljoppi in #119
• feat(bench): add --point for single-point sweep dispatch by @danieljoppi in #120
• ci: add manual artifact-rescue workflow for campaign CSVs by @danieljoppi in #127
• chore(lint): fix ruff 0.16.0 findings in ns3/tools and pin ruff by @danieljoppi in #136
• ci(bench): determinism gate — identical seeds must produce identical results (#129) by @danieljoppi in #135
• feat(core): pheromone-table size gauge (regular/virtual) in counters + --diag (#133) by @danieljoppi in #141
• feat(bench): emit per-run ##RUN## rows and paired A/B verdicts (#128) by @danieljoppi in #137
• feat(bench): byte-normalized routing load nrl_bytes (#132) by @danieljoppi in #140
• feat(bench): emit ##PERF## wall-clock + peak-RSS lines per sim run (#131) by @danieljoppi in #139
• ci(adapter): add ASan/LSan ns-3 job + valgrind NS-2 smoke, non-blocking (#130) by @danieljoppi in #138
• ci(bench): correct the 6h-ceiling comment and add a step-level timeout to scenario-matrix (#125) by @danieljoppi in #156
• feat(infra): publish an optimized-profile ns-3 image (ns3:<ver>-opt) (#123) by @danieljoppi in #157
• fix(infra): split the Docker Hub mirror into non-fatal steps so a bad token can't skip GHCR publishes (#158) by @danieljoppi in #159
• chore(infra): park the Docker Hub mirror — GHCR is the only published registry (#158) by @danieljoppi in #160
• feat(skills): campaign-loop tooling, scenario validation (#134), and ADR-0014 by @danieljoppi in #161
• feat(core): add an ILinkMetric registry mapping metric names to instances (#143) by @danieljoppi in #163
• test(core): add a discrete-event testbench harness for multi-node core tests (#154) by @danieljoppi in #164
• ci(core): measure and report core test coverage on every PR (#162) by @danieljoppi in #165
• fix(core): adopt T_hop = 3 ms from the 2007 thesis, replacing the provisional 50 ms (#88) by @danieljoppi in #167
• fix(core): stop bounding reactive floods — a broadcast budget is a hop limit (#169) by @danieljoppi in #170
• docs(bench): split the benchmark page into per-scenario and per-sweep pages by @danieljoppi in #168
• docs: add a protocol configuration and calibration guide by @danieljoppi in #171
• feat(skill): compare a sweep across two code generations with a baseline control by @danieljoppi in #172
• docs(bench): correct the results caveats — they claimed the numbers were older than they are by @danieljoppi in #175
• docs(context): point the "are the defaults right?" question at the new provenance table by @danieljoppi in #176
• fix(core): bound the reactive flood per (node, generation), not per ant path by @danieljoppi in #174
• docs(config): record that the thesis supersedes [1] on antAcceptanceFactor by @danieljoppi in #178
• chore: ignore .claude/worktrees/, the agent isolation checkouts by @danieljoppi in #183
• docs(config): record the thesis audit outcome — §3.2 drops from 11 unsourced defaults to 4 by @danieljoppi in #184
• fix(bench): correct the thesis scenario preset from the source, and let scenario-matrix pass extraArgs by @danieljoppi in #187
• docs(bench): the results caveat contradicted the table directly above it by @danieljoppi in #189
• feat(core): adopt the thesis's two-factor acceptance band as the reactive-setup default by @danieljoppi in #190

Full Changelog: v1.0.0...v1.1.0

v1.0.0

AntHocNet v1.0.0. See CHANGELOG.md.

Install bundle (anthocnet-1.0.0.zip) — the source to
drop onto your own simulator tree:
make install-ns3 NS3DIR=… or make install-ns2 NS2DIR=….
Pre-built simulator images are on GHCR (see docker/README.md),
including release-pinned tags anthocnet-ns{2,3}:<sim-version>-v1.0.0.

What's Changed

• docs: Update DOI badge in README.md by @danieljoppi in #102
• feat(ns3)!: per-reason pending-queue hold cap, default ReconvHoldCap=1.0s (#21, #103) by @danieljoppi in #104
• chore(release): prepare v1.0.0 — enable major bumps + paper-fidelity table by @danieljoppi in #105

Full Changelog: v0.5.0...v1.0.0

v0.5.0

AntHocNet v0.5.0. See CHANGELOG.md.

Install bundle (anthocnet-0.5.0.zip) — the source to
drop onto your own simulator tree:
make install-ns3 NS3DIR=… or make install-ns2 NS2DIR=….
Pre-built simulator images are on GHCR (see docker/README.md),
including release-pinned tags anthocnet-ns{2,3}:<sim-version>-v0.5.0.

What's Changed

• feat(ns3): expose T_hop as the HopTime attribute (#88) by @danieljoppi in #92
• feat(ns3): attribute pending-queue hold time by reason (#21) by @danieljoppi in #93
• fix(ns3): drive re-discovery for held data with a reactive-retry timer (#21) by @danieljoppi in #94
• feat(core): enforce the maxPathLength ant hop cap with a drop (#26, item 6.2) by @danieljoppi in #95
• feat(core)!: multipath reactive setup (1.5× acceptance filter) with linkfail churn bound, default on (#96) by @danieljoppi in #97
• docs: publications digests — PPSN 2004 paper [1] digested, thesis placeholder by @danieljoppi in #98
• feat(core): fill hello advert slots active-first, then by best pheromone (#26 item 6.5) by @danieljoppi in #99
• feat(core)!: adopt the paper's routing exponents betaData=2 / betaAnts=1 (#70) by @danieljoppi in #100
• fix(ns3): release held packets on backward-ant traversal and retry-timer route hits (#21) by @danieljoppi in #101

Full Changelog: v0.4.0...v0.5.0

v0.4.0

AntHocNet v0.4.0. See CHANGELOG.md.

Install bundle (anthocnet-0.4.0.zip) — the source to
drop onto your own simulator tree:
make install-ns3 NS3DIR=… or make install-ns2 NS2DIR=….
Pre-built simulator images are on GHCR (see docker/README.md),
including release-pinned tags anthocnet-ns{2,3}:<sim-version>-v0.4.0.

What's Changed

• chore(release): stamp CITATION.cff date-released automatically by @danieljoppi in #78
• feat(release): release-pinned + :latest image tags, mirrored to Docker Hub (#77) by @danieljoppi in #79
• docs: remove docs/improvements specs (backlog now in GitHub issues) by @danieljoppi in #80
• ci: cap benchmark artifact retention at 7 days by @danieljoppi in #81
• chore(backlog): add priority-label convention, refresh stale backlog pointers (#82) by @danieljoppi in #83
• fix(bench): pin the paper's 2 Mbit/s radio — ns-3 default rate control halves single-hop PDR (#51) by @danieljoppi in #84
• fix(core): bound the LinkFail origin storm + honest convergence metric (#20, #23) by @danieljoppi in #85
• fix(ns3): cut the delay tail via a measured 3 s pending-queue hold (#21) by @danieljoppi in #86
• feat: measured A2 MAC metric (ns-3 + ns-2) + blocking benchmark anchors by @danieljoppi in #87
• feat: paper-parity QoS metrics, across-run dispersion, and the provisional thesis preset by @danieljoppi in #90

Full Changelog: v0.3.0...v0.4.0

v0.3.0

AntHocNet v0.3.0. See CHANGELOG.md.

Install bundle (anthocnet-0.3.0.zip) — the source to
drop onto your own simulator tree:
make install-ns3 NS3DIR=… or make install-ns2 NS2DIR=….
Pre-built simulator images are on GHCR (see docker/README.md).

What's Changed

• feat(bench): #24 propagation/range/area knobs, stock-baseline control, and PDR diagnostics by @danieljoppi in #50
• docs(bench): point validation anchors at #51 single-hop loss root cause by @danieljoppi in #52
• feat(core): bound proactive broadcasts and add repair wait/discard timer by @danieljoppi in #53
• feat(ns3): re-inject MAC-failed data packets and expose detector-D tuning by @danieljoppi in #54
• feat(core): congestion-aware per-hop metric from MAC queue occupancy by @danieljoppi in #67
• feat(bench): gateway-hotspot scenario (--sink) to give the A2 metric a verdict by @danieljoppi in #72
• feat(bench): instrumentation for A2 analysis — skills + --qdiag queue diagnostic by @danieljoppi in #74
• fix(ns3): read AC_BE_NQOS so A2 has a queue signal — re-land --qdiag + CLAUDE.md by @danieljoppi in #75
• chore: fix bench_parse verdict + delay99, correct A2 docs (validated), install caveman by @danieljoppi in #76

Full Changelog: v0.2.0...v0.3.0

v0.2.0

AntHocNet v0.2.0. See CHANGELOG.md.

Install bundle (anthocnet-0.2.0.zip) — the source to
drop onto your own simulator tree:
make install-ns3 NS3DIR=… or make install-ns2 NS2DIR=….
Pre-built simulator images are on GHCR (see docker/README.md).

What's Changed

• docs: add Zenodo DOI to CITATION.cff and README by @danieljoppi in #42
• feat(ns3): wire MAC transmit-failure repair hook (detector D) by @danieljoppi in #43
• fix(core): debounce detector D to stop the link-failure storm (#19) by @danieljoppi in #44
• docs: add ADR-0013 issue-tracking policy + detector-D handoff (#19) by @danieljoppi in #48

Full Changelog: v0.1.1...v0.2.0