{"id":"7d1dd652-e318-448e-962b-f3b5cd5ca6f3","arxiv_id":"2507.14234","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A simulated solar-plus-kinetic powered wildlife tag with a 2.5 F supercapacitor keeps GPS fixes every two minutes and hourly NB-IoT transmissions energy neutral in a Belgian winter.","lead":"This paper simulates a wildlife tracker that harvests solar and kinetic energy into a supercapacitor and finds that a 2.5 F capacitor can run GPS fixes every two minutes with hourly NB-IoT transmissions during a winter test window. If the simulation holds on hardware, wildlife tags could run for years without battery replacement and also report animal activity from the harvester itself.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The wolf kinetic-yield input (13.07 J/day, §4.2) is load-bearing but not sourced: the cited trials in the introduction yield 0.69–3.2 J/day, so this energy input must be verified before the 2-minute GPS claim can be accepted.","rationale":"The paper's stated goal is a feasibility study; its strongest quantified result is that a 2.5 F supercapacitor with a 2-minute GPS interval and hourly NB-IoT never crosses Vmin over the 13-day winter scenario. I read the simulation as internally coherent: Equation 1 is a standard RC discretization, the task energies in Tables 1 and 2 are consistently propagated, and Algorithm 1 is well-specified. The weakest step is not the scheduler but the input energy. The abstract claims energy neutrality and higher yield than single-source systems, yet the body contains no single-source comparison; however, that is a framing and reproducibility issue rather than the parameter that can flip the headline number. The wolf kinetic value is the one place where the paper contradicts itself: the introduction reports only dog and pony measurements from [5], while Section 4.2 asserts a wolf-specific 13.07 J/day from the same reference. A simple energy balance shows that a 2.5 F capacitor stores only about 33.8 J between Vmax and Vmin, while a 14 h night with 2-minute fixes consumes about 31 J, leaving a margin comparable to the difference between the asserted 13.07 J/day and the dog/pony values. Thus the unverified kinetic figure is load-bearing. The right response is not rejection but a condition: the authors should either identify the source of the 13.07 J/day figure or run a sensitivity sweep. This is exactly the reader's conditional stance, so I keep the verdict unchanged.","tokens_in":13174,"tokens_out":17988,"duration_ms":215787,"concrete_test":"Obtain Gregersen et al. [5] and locate the derivation of the 13.07 J/day wolf figure, whether it is a measurement, an extrapolation, or a citation. Then, using the paper's own Section 4.1 model (Equation 1) with Table 3 parameters and the same Theuerkauf diurnal shape, rerun the Section 5.2 2.5 F / 2-minute simulation with the kinetic profile scaled to 3.2 J/day and 0.69 J/day integrated energy. Record the minimum capacitor voltage and the number of nights on which V falls below 1.8 V, or below Vthresh_HS = 1.9 V causing missed fixes. If either lower-yield case produces shutdowns, the central 2-minute claim fails for those inputs; if it does not, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The simulation result that matters is the 2-minute-fix, 2.5 F configuration over the 13-day winter scenario. At night, when solar input is zero, the capacitor must supply roughly 2.2 J/h (30 hot starts at 25.15 mJ, about 0.33 ephemeris downloads at 754.5 mJ, one NB-IoT transmission at 541.5 mJ, plus sleep current), i.e. about 31 J over a 14 h night. The usable storage between 5.5 V and Vmin = 1.8 V is about 33.8 J, so the system lives on a thin margin. The kinetic harvester is the only nighttime source, and Section 4.2 assigns it 13.07 J/day for wolves, attributed to Gregersen et al. [5]. However, Section 1 describes [5] as dog trials yielding 2.26–3.2 J/day and an Exmoor pony yielding 0.69 J/day; no wolf figure is mentioned anywhere. If the correct wolf yield is near the dog upper bound or the pony value, the nighttime kinetic contribution drops by roughly 10 J, and within the paper's own linear-capacitor model the 2.5 F configuration falls below Vmin, losing the claimed consistent every-two-minute operation. The manuscript reports no sensitivity analysis around this parameter, and Section 6 concedes the work is simulation-only. This is the load-bearing uncertainty: not the scheduling algorithm, but the unverified energy-input assumption that separates survival from shutdown.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a simulation framework for a wildlife tracking tag that harvests energy from solar and kinetic sources, stores it in a supercapacitor, and uses an energy-aware scheduler to perform GPS fixes and NB-IoT transmissions. The system is evaluated over a 13-day winter irradiance dataset from Antwerp, Belgium, with three GPS fix intervals (1, 2, 5 minutes) and three capacitor sizes (1 F, 2.5 F, 5 F). The central result is that a 2.5 F capacitor supports consistent every-two-minute GPS fixes and hourly NB-IoT transmissions without dropping below the operating voltage Vmin across the entire dataset. The kinetic harvester is also proposed as a motion proxy by sampling its harvested current.","tokens_in":13434,"tokens_out":5223,"duration_ms":54448,"significance":"If the results are correct, the paper offers a useful component-level simulation model for energy-harvesting wildlife tags and identifies a plausible capacitor size and operating configuration for solar-plus-kinetic energy neutrality. The dual use of the kinetic harvester as an activity sensor is an interesting design idea that could reduce tag complexity. The simulation is internally consistent and the threshold-based scheduler is clearly described. However, the paper is simulation-only, relies on an unverified kinetic energy input for wolves, and claims a comparison against single-source systems that is not actually performed. These issues currently temper the strength of the feasibility claim.","major_comments":[{"comment":"The abstract states that the approach 'significantly increas[es] data yield and reliability compared to single-source systems,' but Section 5 reports results only for the proposed multi-source system with different fix intervals and capacitor sizes. No single-source baseline (solar-only or kinetic-only) is simulated or tabulated. Either add such a comparison or revise the abstract and Section 5.2 to clearly limit the claim to the multi-source configuration.","section":"Abstract and Section 5"},{"comment":"The wolf-specific average daily kinetic harvest of 13.07 J is attributed to Gregersen et al. [5], yet Section 1 describes [5] as reporting 2.26–3.2 J/day for dogs and 0.69 J/day for an Exmoor pony, with no wolf figure. This value is load-bearing because nighttime operation depends almost entirely on the kinetic harvester (as seen in Figure 5), and the 2.5 F configuration has a thin voltage margin. The authors should provide a direct source for the 13.07 J value and, more importantly, add a sensitivity analysis over a plausible range (e.g., 0.69–3.2 J/day) to show under which conditions the every-two-minute fix claim would fail.","section":"Section 4.2"},{"comment":"The solar irradiance dataset is described as 'private' and is cited to [11], but the title of [11] ('Harvesting energy from soil-air temperature differences for batteryless iot devices: A case study') does not match the described content of solar irradiance measurements in Antwerp. This citation mismatch, together with the private status of the dataset, makes the daytime energy input difficult to verify. The authors should either make the dataset available (or deposit it in a repository), provide summary statistics such as daily insolation values, and correct the citation or clarify the origin of the solar data.","section":"Section 4.2 and Reference [11]"}],"minor_comments":[{"comment":"The average number of fixes per day for the 2-minute interval is reported as 700.00, which is 20 fixes per day short of the theoretical 720 (24 h × 60 min / 2 min). The paper says the daily fix count has 'negligible standard deviation' and reflects reliable scheduling, but the shortfall is not explained. A brief note on why some fixes are skipped (e.g., start-up transients, capacitor recharge delays) would be helpful.","section":"Section 5.2.3, Table 4"},{"comment":"The notation '𝑉t+1' and '𝑉t' is ambiguous in print; using explicit subscripts (e.g., V_{t+1}, V_t) and a single symbol list near Eq. (1) would improve readability.","section":"Section 4.1, Eq. (1)"},{"comment":"The column headers 'MCUActiveBase' and 'GPSBackupCurrent' are somewhat cryptic; expanding them to 'MCU active base current' and 'GPS hardware backup current' in the table caption would make the table self-explanatory.","section":"Section 3.5, Table 2"},{"comment":"The three curves for different GPS fix intervals may be difficult to distinguish in grayscale; using different line styles or adding markers in the legend would aid readability.","section":"Section 5.2.1, Figure 4"}],"recommendation":"major_revision","confidential_remarks":"The paper's central feasibility claim is plausible but currently rests on two unaddressed supports: the unsourced 13.07 J/day wolf kinetic input and the unsupported comparison against single-source systems claimed in the abstract. The private solar dataset with a mismatched citation is also a reproducibility concern. All of these are fixable within the scope of a revision, so I do not think rejection is warranted, but the authors must either supply the missing evidence or soften the claims accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a solid feasibility simulation, not a demonstration. The new combination is solar plus kinetic harvesting with NB-IoT backhaul, and the dual use of the kinetic harvester's Coulomb counter as an activity proxy without a dedicated accelerometer. That dual-use idea is genuinely neat and not in the cited single-source trackers. The energy accounting is transparent: component-level currents from datasheets, NB-IoT power measured on real hardware with a precision analyzer, capacitor voltage model taken from prior work, and an energy-aware scheduler with voltage thresholds. On those assumptions, the 2-minute GPS fix / hourly NB-IoT configuration on a 2.5 F capacitor stays above Vmin across the 13-day winter dataset. I checked the arithmetic in the stress-test note and it is right: the night budget is tight, and the system lives on a thin margin.\n\nWhere it is soft, in proportion: the abstract says 'significantly increasing data yield and reliability compared to single-source systems,' but the body never performs that comparison; it only varies fix interval and capacitor size. That is an overclaim and should be fixed. Bigger issue: the 13.07 J/day wolf kinetic yield in Section 4.2 is load-bearing but not sourced. Section 1 describes Gregersen et al. as dog trials at 2.26–3.2 J/day and an Exmoor pony at 0.69 J/day; no wolf figure appears in the cited paper as described. If the real wolf yield is closer to 3 J/day, the nighttime kinetic contribution drops by about 10 J and, within the paper's own capacitor model, the 2.5 F configuration goes below Vmin. The paper does no sensitivity analysis around this input, and Section 6 concedes the work is simulation-only. Also, the private solar dataset is cited to a paper about soil-air temperature differences, which looks like a reference mismatch. There is no shipped code or data.\n\nThese are not fatal for a feasibility study, but they are exactly the places where a reader should push. The central scheduling/energy-neutrality logic holds up on its own terms; the fragility is the energy-input assumption, not the algorithm.\n\nAudience: designers of batteryless wildlife tags and the energy-harvesting IoT community. A serious referee should get this, with the expectation of major revision: clarify the wolf kinetic source or add a sensitivity sweep, and make the single-source comparison real. If those land, this becomes a useful reference point. I'd bring it to reading group.","headline":"A transparent simulation-based feasibility study whose headline 2-minute GPS claim rests on an unverified wolf kinetic-yield input that needs checking before the numbers are quoted.","tokens_in":14052,"tokens_out":2349,"would_cite":true,"duration_ms":25616,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that a wildlife tracking tag combining solar and kinetic energy harvesting, supercapacitor storage, and energy-aware scheduling can stay energy-neutral while taking GPS fixes every two minutes and transmitting over…","keywords":["energy harvesting","energy-neutral IoT","wildlife tracking","supercapacitor","energy-aware scheduling","NB-IoT","kinetic energy harvesting","solar energy harvesting"],"falsifier":"Instrument a free-ranging or captive wolf with a tag using the modeled solar panel, kinetic harvester, 2.5 F supercapacitor, and scheduler, and record the capacitor voltage across several consecutive winter nights; if the voltage crosses below 1.8 V on any night with the assumed irradiance, the central energy-neutrality claim is false. A quicker proxy is to measure the kinetic harvester's average daily energy yield in joules for the actual target species and check whether it approaches 13.07 J rather than the 0.69–3.2 J range reported for ponies and dogs.","tokens_in":12931,"feed_emoji":"🐺","tokens_out":6531,"duration_ms":64606,"temperature":0.7,"pith_summary":"Long-term animal tracking is limited by tag batteries, which are hard to replace in the wild. This paper argues that a tag harvesting both solar energy and kinetic motion energy can run indefinitely on a supercapacitor without battery swaps, and that an energy-aware scheduler can keep it alive through a worst-case winter. In simulation over 13 days of Antwerp winter weather, a 2.5 F supercapacitor never falls below the minimum operating voltage while the tag takes a GPS fix and a motion reading every two minutes and transmits the batch over NB-IoT every hour. A sympathetic reading is that the paper establishes a feasible component-level design point for maintenance-free, multi-source wildlife tracking, not a field-tested device.","feed_headline":"Solar-plus-kinetic tag keeps wolf GPS fixes flowing every 2 minutes","feed_subtitle":"A 2.5 F supercapacitor and an energy-aware scheduler hold voltage above cutoff through a 13-day winter run.","key_machinery":"The argument rests on a discrete-time capacitor model (Equation 1) that updates the supercapacitor voltage from the harvested current, the equivalent load resistance, the capacitance, and the time step; on an energy-aware scheduler (Algorithm 1) that only permits GPS or NB-IoT tasks when the measured voltage clears task-specific thresholds, with cold start reserved for full restarts; and on component-level energy characterizations derived from datasheets and direct measurements. A kinetic pendulum harvester feeds a PMIC whose lossless Coulomb counter samples the harvested current, and because that current tracks movement intensity, the same component doubles as an activity sensor. The solar path uses a 40x40 mm panel with a stated efficiency, an assumed cosine-loss factor, and a PMIC efficiency figure, while the kinetic path is built from a wolf daily energy estimate distributed across a diurnal activity curve.","core_discovery":"On the paper's own terms, the central discovery is that combining solar and kinetic energy harvesting with a supercapacitor and an energy-aware task scheduler keeps a wildlife tracker energy-neutral at a practically useful sampling rate. For a 2.5 F capacitor and a 2-minute GPS interval, the simulated capacitor voltage stays above the 1.8 V cutoff for the full 13-day winter dataset, while the system records GPS fixes and Coulomb counter readings every two minutes and transmits the accumulated data hourly over NB-IoT. The comparison runs show a clear trade-off: 2-minute fixes deliver 2.6 times more GPS measurements per day than 5-minute fixes with no loss of reliability, whereas 1-minute fixes cause warm-start fallbacks, a cold start, and a shutdown gap that creates multi-hour data holes. Both the 2.5 F and 5 F capacitors are workable, while the 1 F capacitor depletes too easily, and the 5 F capacitor fails to fully recharge on one low-sun day.","pith_inferences":["In the editor's reading, the wolf kinetic yield of 13.07 J/day is the least-supported input; a sensitivity sweep from 0.69 to 13.07 J/day would show at what yield the 2-minute schedule breaks, and that sweep is feasible without hardware.","The design assumes NB-IoT coverage and a specific winter location, so the energy budget would need revalidation for other latitudes, seasons, or communication backhauls; the scheduling framework itself is portable.","The activity proxy could support adaptive duty cycling, skipping GPS fixes during detected rest periods, which the paper lists as future work but does not simulate.","The modular architecture could accept a third harvesting source such as thermal energy to close the overnight gap for less active species than wolves."],"forward_implications":["With the modeled components, a 2-minute GPS interval plus hourly NB-IoT transmission runs energy-neutral through the 13-day winter scenario on a 2.5 F or 5 F supercapacitor.","Dropping the GPS interval from 5 to 2 minutes raises daily fix counts by a factor of 2.6 with no warm starts, cold starts, or shutdown gaps.","Pushing the interval to 1 minute destabilizes the system: warm-start ephemeris downloads occur, a cold start is triggered, and a shutdown produces data gaps of multiple hours.","A 1 F supercapacitor is too small for reliable 2-minute fixes, while a 5 F capacitor is viable but stays short of full recharge on low-sun days, so the middle size is the practical choice.","The kinetic harvester's Coulomb counter readings provide a movement-intensity proxy at negligible extra energy cost, adding a behavior channel without dedicated sensors."],"supporting_citations":[{"why":"Supplies the kinetic energy harvester design and the dog and pony yield figures; the paper's wolf estimate of 13.07 J/day is attributed to this source.","marker":"[5]"},{"why":"Provides the diurnal wolf activity pattern used to distribute the daily kinetic energy across dawn, day, dusk, and night.","marker":"[18]"},{"why":"Supplies the winter solar irradiance dataset from Antwerp used as the worst-case solar input.","marker":"[11]"},{"why":"Defines the multi-source energy combiner architecture and the 0.88 efficiency factor applied to combined harvesting.","marker":"[10]"},{"why":"Basis for the RC capacitor voltage model (Equation 1) that drives the simulation.","marker":"[13]"},{"why":"The nRF9160 NB-IoT module whose measured power profile sets the transmission energy cost.","marker":"[9]"},{"why":"The SAM-M10Q GPS datasheet defining hot start, warm start, and ephemeris download currents.","marker":"[20]"},{"why":"Provides the cold-start time-to-fix estimate used in the energy consumption tables.","marker":"[4]"}],"fun_headline_variants":["Solar and kinetic harvest power wildlife tags without battery swaps","Energy-neutral wildlife tag samples GPS every 2 minutes","Hybrid energy harvest boosts wildlife tracker data yield","13-day winter run: solar-kinetic tag stays above cutoff","Wildlife tracking goes energy-neutral with solar and kinetic"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The system's overnight survival rests on the assumed kinetic energy yield for wolves, an average of 13.07 J per day whose provenance is not fully documented in the paper; if real wolves yield closer to the 0.69–3.2 J/day measured in dog and pony trials, the capacitor would drop below the 1.8 V cutoff at night and the two-minute fix schedule would fail.","fun_headline_variants_meta":{"raw":{"variants":["Solar and kinetic harvest power wildlife tags without battery swaps","Energy-neutral wildlife tag samples GPS every 2 minutes","Hybrid energy harvest boosts wildlife tracker data yield","13-day winter run: solar-kinetic tag stays above cutoff","Wildlife tracking goes energy-neutral with solar and kinetic"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000443,"raw_usage":{"total_tokens":2273,"prompt_tokens":1002,"completion_tokens":1271,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":618,"completion_tokens_details":{"reasoning_tokens":1193}},"tokens_in":618,"tokens_out":1271,"duration_ms":14224,"temperature":1.0,"reasoning_tokens":1193,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T16:34:11.258656+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Instrument a free-ranging or captive wolf with a tag using the modeled solar panel, kinetic harvester, 2.5 F supercapacitor, and scheduler, and record the capacitor voltage across several consecutive winter nights; if the voltage crosses below 1.8 V on any night with the assumed irradiance, the central energy-neutrality claim is false. A quicker proxy is to measure the kinetic harvester's average daily energy yield in joules for the actual target species and check whether it approaches 13.07 J rather than the 0.69–3.2 J range reported for ponies and dogs.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the diurnal wolf activity pattern used to distribute the daily kinetic energy across dawn, day, dusk, and night."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the winter solar irradiance dataset from Antwerp used as the worst-case solar input."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the multi-source energy combiner architecture and the 0.88 efficiency factor applied to combined harvesting."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Basis for the RC capacitor voltage model (Equation 1) that drives the simulation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The nRF9160 NB-IoT module whose measured power profile sets the transmission energy cost."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The SAM-M10Q GPS datasheet defining hot start, warm start, and ephemeris download currents."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the cold-start time-to-fix estimate used in the energy consumption tables."}],"review_version":1}