Sweep: pause

Varies: pause time (s) — 0 (constant motion) → 900 s (static).

← Benchmark index · Metrics · Methodology

Provenance — measured at v1.3.0. Every number on this page was produced by the campaign runs named below, all of which predate the v1.3.0 tag (19009be). No commit between v1.2.0 and that tag changes routing behaviour, so checking out v1.3.0 reproduces these cells. The default branch will not: it has since changed the pheromone weighting (#327, betaAnts/ betaData 2.0 → 20) and the per-seed RNG stream assignment (#352), either of which moves measured values. See Provenance and #365.

Restated on CBR sources (#521). The pinned table below was measured at half the documented offered load. A 20-seed re-measure on main is under Restated on CBR sources at the end of this page.

OLSR's PDR column is inflated (#510). These cells predate the fix that counts a send stock OLSR refuses at the source as an offered packet (PR #511, f480d0ae). Before it, OLSR's PDR was computed over the ticks it had a route for. This page stays pinned to v1.3.0 and is not re-measured. The offered-based value below is estimated from this table's own columns, as thrput_olsr / thrput_aodv × PDR_aodv. On these aggregate rows the same estimate reproduces AntHocNet's and DSDV's reported PDR to within ~0.6 pp, which is the throughput rounding.

pause (s)	OLSR PDR as published	offered-based estimate	DSDV PDR
0	65.8	≈ 55.6 (-10.2)	54.1
100	68.4	≈ 57.8 (-10.6)	57.3
300	81.6	≈ 74.7 (-6.9)	75.5
600	92.0	≈ 88.5 (-3.5)	87.8
900	100.0	≈ 99.7 (-0.3)	98.7

The correction reaches 10.6 pp. Where it brings OLSR within about 2 pp of DSDV, do not quote an OLSR-over-DSDV ordering from this page. AntHocNet, AODV and DSDV are unaffected.

What it varies

Reproduces Fig. 2 of the AntHocNet paper (Di Caro/Ducatelle/Gambardella, PPSN VIII 2004, §4): the random-waypoint pause time is swept from 0 to 900 s, taking the field from constant motion to effectively static. It is also the shape checked by the Broch pause-sweep validation anchor — AODV PDR must rise with pause, and DSDV must be worst under high mobility (see methodology.md). Unlike the paper (AODV only), every baseline is run on identical realisations.

Defined as SWEEPS["pause"] in ns3/tools/run-scenarios.py (scenario=paper areaX=2500, pause ∈ {0, 100, 300, 600, 900}).

How it is produced

Sweeps are too heavy for the per-merge benchmarks workflow, which runs only the discrete scenarios. They come from the manual Scenario matrix + charts workflow (scenario-matrix.yml), which renders sweep-pause.png into docs/benchmarks/ and uploads the classified CSV; the table below is filled in by pointing update-benchmarks.py at that CSV.

Raw sweep data rescued from expired artifacts lives in ../campaign/. The published table below comes from the 2026-08-03 20-seed re-run (runs 30850273352 / 30850282055 / 30850291550 / 30850299279 / 30850306777, one point per job, 3.42-opt image, 900 s, range/disk PHY), rescued as ../campaign/308502*-run.csv and summarizable with the benchmark-results skill's sweep_summary.py. The superseded pre-#88/#169 5-seed data remains at 30031904151-pause-disk.csv.

Results

Post-#88/#169, 20 seeds, 95% CIs. These numbers include the T_hop = 3 ms fix (#88, PR

167) and the reactive-hop-cap removal

(#169, PR #170), and meet the methodology runs floor. Versus the superseded 5-seed data, AntHocNet's PDR moved +9.8 to +26.9 pp across the sweep (baselines moved ≤ ~1.5 pp, consistent with the 5→20 run-count change) — the re-run existed precisely because #88/#169 invalidated the old numbers. The delay99 column carries the known #21 tail (remediation tracked in #308). Do not quote path_div_*/entropy from these cells (#230 window caveat).

Sweep pause — mean of 20 run(s) per point, every baseline on identical realisations; ± is the 95% CI half-width (#293). Generated by run-scenarios.py; chart by make-charts.py.

pause time (s)	protocol	PDR % ±95	mean delay (ms) ±95	99th delay (ms) ±95	throughput (kbps)	NRL ±95	jitter (ms)	dOff90 (ms)
0	anthocnet	89.9 ± 0.6	74.6 ± 3.3	961.5 ± 10.7	5.90	43.404 ± 0.87	115.93	inf
0	aodv	77.4 ± 1.3	36.4 ± 2.1	725.3 ± 61.5	5.11	45.026 ± 1.19	50.01	inf
0	olsr	65.8 ± 1.2	8.1 ± 0.7	29.5 ± 1.3	3.67	7.395 ± 0.21	10.67	inf
0	dsdv	54.1 ± 1.6	14.8 ± 1.6	283.0 ± 146.6	3.60	32.211 ± 0.99	22.00	inf
100	anthocnet	86.9 ± 1.1	76.7 ± 3.9	976.2 ± 14.7	5.70	45.321 ± 1.36	121.00	inf
100	aodv	79.3 ± 1.4	40.4 ± 3.3	897.1 ± 72.2	5.23	38.289 ± 1.23	54.38	inf
100	olsr	68.4 ± 1.4	7.6 ± 0.6	30.9 ± 1.2	3.81	7.030 ± 0.20	9.13	inf
100	dsdv	57.3 ± 1.7	18.0 ± 2.4	158.3 ± 66.4	3.81	28.637 ± 0.87	21.59	inf
300	anthocnet	85.7 ± 1.8	63.7 ± 3.3	939.0 ± 17.1	5.63	43.775 ± 2.58	103.92	inf
300	aodv	85.5 ± 2.9	29.7 ± 3.5	556.2 ± 122.0	5.64	29.717 ± 1.36	37.33	inf
300	olsr	81.6 ± 1.0	6.4 ± 0.5	25.6 ± 0.7	4.93	5.374 ± 0.14	5.78	inf
300	dsdv	75.5 ± 2.0	17.7 ± 3.1	101.8 ± 18.8	5.02	20.484 ± 0.57	18.57	inf
600	anthocnet	86.0 ± 3.4	54.0 ± 4.5	896.4 ± 32.6	5.65	43.398 ± 5.24	89.93	inf
600	aodv	87.7 ± 6.4	23.3 ± 3.6	268.2 ± 127.8	5.79	26.710 ± 2.66	26.21	inf
600	olsr	92.0 ± 0.6	5.2 ± 0.3	20.7 ± 0.5	5.84	4.478 ± 0.09	2.87	66.2
600	dsdv	87.8 ± 1.8	15.7 ± 2.8	70.7 ± 13.5	5.83	16.954 ± 0.39	14.31	inf
900	anthocnet	86.8 ± 5.1	43.2 ± 6.7	788.0 ± 76.9	5.70	43.541 ± 8.47	74.00	inf
900	aodv	90.3 ± 9.6	13.7 ± 1.4	90.3 ± 9.7	5.96	24.010 ± 4.84	10.65	inf
900	olsr	100.0 ± 0.0	4.0 ± 0.2	10.3 ± 0.9	6.58	3.931 ± 0.07	0.35	7.7
900	dsdv	98.7 ± 1.3	13.6 ± 2.3	52.9 ± 10.1	6.56	14.524 ± 0.34	10.18	inf

Restated on CBR sources (#521)

Re-measured on constant-bit-rate sources. Until #521 every source ran ns-3's default 1 s on / 1 s off, so the pinned table above was taken at half the documented offered load. That table stays as the dated v1.3.0 record.

Provenance. main @ 3edbab6a, 3.42-opt, 900 s, range/disk PHY, 20 seeds per point, one point per scenario-matrix job: runs 37233291212 (0), 37233292762 (100), 37233294241 (300), 37233296147 (600) and 37233298023 (900). Rescued as ../campaign/<run>-run.csv (aggregate) and <run>-run-runs.csv (per seed). scenario_check.py results: 0 FAIL; the WARNs are the #230 path-diversity window caveat and route-flap reorder notes. OLSR's PDR is offered-based (#510), so the estimate table at the top of this page does not apply to this block.

Attribution. This is not a CBR-only A/B. 3edbab6a also carries every change after v1.3.0 that the provenance note names (#327 pheromone weighting, #352 RNG streams), plus #510 (OLSR accounting) and

522 (TTL compensation). sweep_summary.py --vs the pinned CSVs reports

its baseline control as FAIL, as expected: every arm's offered load doubled. Quote this block's within-point orderings and paired deltas, not its difference from the pinned table.

Sweep pause — mean of 20 run(s) per point, every baseline on identical realisations; ± is the 95% CI half-width (#293). Rendered with update-benchmarks.py's sweep builder from the rescued CSVs.

pause time (s)	protocol	PDR % ±95	mean delay (ms) ±95	99th delay (ms) ±95	throughput (kbps)	NRL ±95	jitter (ms)	dOff90 (ms)
0	anthocnet	91.0 ± 0.5	30.5 ± 1.1	490.6 ± 35.6	11.99	13.986 ± 0.38	45.83	inf
0	aodv	79.8 ± 1.0	27.5 ± 2.6	426.8 ± 93.6	10.51	23.235 ± 0.25	34.34	inf
0	olsr	55.4 ± 1.1	6.2 ± 0.6	26.9 ± 1.7	7.30	3.709 ± 0.08	7.61	inf
0	dsdv	53.0 ± 1.0	14.3 ± 1.1	238.9 ± 58.1	6.99	16.528 ± 0.32	20.64	inf
100	anthocnet	88.3 ± 0.9	25.7 ± 1.3	388.6 ± 32.3	11.63	13.461 ± 0.61	36.37	inf
100	aodv	81.2 ± 1.4	25.8 ± 2.2	488.2 ± 104.7	10.69	18.523 ± 0.48	31.67	inf
100	olsr	57.4 ± 1.7	6.0 ± 0.4	27.0 ± 1.4	7.56	3.522 ± 0.10	6.55	inf
100	dsdv	57.8 ± 1.5	16.0 ± 2.2	200.2 ± 83.1	7.61	14.230 ± 0.28	19.32	inf
300	anthocnet	88.5 ± 2.4	16.5 ± 1.5	214.2 ± 23.0	11.65	11.587 ± 1.49	19.80	inf
300	aodv	85.4 ± 2.8	19.6 ± 2.2	261.4 ± 79.5	11.25	12.867 ± 0.79	22.47	inf
300	olsr	72.0 ± 3.2	5.4 ± 0.5	23.4 ± 1.2	9.49	2.768 ± 0.12	4.28	inf
300	dsdv	73.2 ± 2.9	13.7 ± 2.0	100.2 ± 19.0	9.64	10.524 ± 0.33	14.44	inf
600	anthocnet	89.7 ± 4.3	12.3 ± 1.5	165.2 ± 20.5	11.83	11.028 ± 3.01	12.07	inf
600	aodv	89.9 ± 4.3	14.6 ± 2.4	146.5 ± 69.5	11.85	9.592 ± 1.25	14.72	inf
600	olsr	84.2 ± 4.6	4.7 ± 0.4	18.0 ± 1.0	11.09	2.334 ± 0.14	2.29	inf
600	dsdv	83.8 ± 4.4	13.7 ± 2.2	68.5 ± 9.6	11.05	8.743 ± 0.26	11.50	inf
900	anthocnet	90.5 ± 6.6	6.2 ± 0.7	17.1 ± 4.9	11.93	11.217 ± 5.35	2.07	inf
900	aodv	93.4 ± 6.7	6.9 ± 1.1	45.6 ± 13.4	12.31	6.503 ± 1.54	4.58	inf
900	olsr	94.1 ± 6.7	3.8 ± 0.4	9.7 ± 1.5	12.40	2.073 ± 0.18	0.24	inf
900	dsdv	93.0 ± 6.5	11.0 ± 2.2	46.5 ± 9.2	12.26	7.444 ± 0.22	7.54	inf

Paired per-seed deltas, AntHocNet minus each baseline (mean ± 95 % CI half-width, 20 seeds; ns = the CI spans zero):

pause (s)	ΔPDR − aodv (pp)	ΔPDR − olsr (pp)	ΔPDR − dsdv (pp)	Δdelay99 − aodv (ms)	Δdelay99 − olsr (ms)	Δdelay99 − dsdv (ms)
0	+11.2 ± 1.1	+35.6 ± 1.0	+38.0 ± 0.8	+64 ± 101 (ns)	+464 ± 36	+252 ± 63
100	+7.1 ± 1.0	+30.9 ± 1.3	+30.5 ± 1.0	−100 ± 101 (ns)	+362 ± 32	+188 ± 69
300	+3.1 ± 1.1	+16.4 ± 1.6	+15.3 ± 1.3	−47 ± 79 (ns)	+191 ± 22	+114 ± 27
600	−0.1 ± 1.5 (ns)	+5.6 ± 1.4	+5.9 ± 1.2	+19 ± 53 (ns)	+147 ± 20	+97 ± 17
900	−2.9 ± 1.9	−3.6 ± 1.9	−2.5 ± 1.7	−29 ± 14	+7.5 ± 5.0	−29 ± 11

What holds and what changed versus the pinned table: - The mobility crossover holds and moves later. AntHocNet's delivery lead shrinks as motion stops. Pinned, it was already below all three baselines at 600 s. On CBR it ties AODV at 600 s and still leads OLSR and DSDV by ~6 pp. - The static point (900 s) is still a loss. AntHocNet is 2.5 to 3.6 pp below every baseline, each significant. Pinned, the gaps were 3.5 (AODV), 13.2 (OLSR, pre-#510) and 11.9 (DSDV) pp. The ±6.6 pp aggregate CIs are placement-driven: each static seed is one fixed topology, which the paired deltas cancel. - AntHocNet's static tail is now short. Its delay99 at 900 s is 17.1 ms, against 788 ms pinned. That is below AODV's and DSDV's (−29 ms each) and 7.5 ± 5.0 ms above OLSR's. - NRL: AntHocNet's is below AODV's from 0 to 300 s (−9.25 ± 0.28 / −5.06 ± 0.49 / −1.28 ± 1.19), level at 600 s, and above at 900 s (+4.71 ± 4.14). - OLSR versus DSDV: OLSR leads at 0 s (+2.38 ± 0.55 pp) and 900 s (+1.07 ± 0.62 pp), DSDV leads at 300 s (+1.14 ± 0.65 pp), and they are level at 100 and 600 s.