LyndHub
Jadwal Sholat

Memuat jadwal sholatโ€ฆ

โ† All journalsSearch index

Ilmu Komputer & AI editorial

Open AccessOA2026

Rethinking Battery-free Sensing Communication via Wake-up Radios

MagPie: A microampere wake-up radio and backed LP-RTC architecture for All-to-One collection without a powered control anchor
Gaosheng Liuยท 2026ยท DOI 10.48550/arXiv.2608.25292

The core problem

Battery-free sensors operate under severe energy constraints: they expose short, stochastic communication windows and may lose timing state whenever their main energy domain browns out. This creates a fundamental coordination problem for static, single-hop All-to-One collection, where many intermittently powered devices must deliver data to a collector without a continuously powered control anchor. The paper identifies two coupled failure modes: (1) first-contact failure, because a node's radio-on window is too brief and unpredictable for a collector to discover it, and (2) phase loss, because any timing state acquired during an earlier exchange is erased by a brownout. MagPie addresses both by combining a microampere wake-up radio (WuR) with a separately backed low-power real-time clock (LP-RTC). The WuR widens the first-contact window, while the LP-RTC preserves the acquired phase for later exchanges. Energy gates, epoch-versioned schedules, and idempotent slot allocation extend this mechanism to static, single-hop All-to-One collection without a powered control anchor. The analysis accounts for role selection and duty-cycled listening, and bounds scheduled-retry tails only under

Innovation

The evaluation separates three scopes and reports them independently. In independent, administratively censored simulation trials, MagPie completes 100/100 first rendezvous events in each of five trace-parameterized harvesting scenarios, whereas the Find baseline completes between 36 and 100 out of 100 depending on the scenario. This indicates that the WuR-widened first-contact window is the dominant factor in reliable discovery under stochastic harvesting. The single-collision-domain slotted-Aloha study shows that adaptive K is necessary at high contender density; the paper reports mean, P95, and confidence intervals through 120 components, demonstrating that fixed-K configurations degrade as the number of contenders grows. The controlled STM32WL33 experiments validate three properties: alignment between the WuR and LP-RTC, clock persistence across main-domain brownouts, and six-device slot execution. The longest reported functional run is 11.05 hours. These results are bounded by the study's stated scope: the authors do not claim measured end-to-end energy, ambient-harvesting performance, or multi-hop scalability. The paper is explicit that highly variable harvesting still limits
Battery-free sensors operate under severe energy constraints: they expose short, stochastic communication windows and may lose timing state whenever their main energy domain browns out. This creates a fundamental coordination problem for static, single-hop All-to-One collection, where many intermittently powered devices must deliver data to a collector without a continuously powered control anchor. The paper identifies two coupled failure modes: (1) first-contact failure, because a node's radio-on window is too brief and unpredictable for a collector to discover it, and (2) phase loss, because any timing state acquired during an earlier exchange is erased by a brownout. MagPie addresses both by combining a microampere wake-up radio (WuR) with a separately backed low-power real-time clock (LP-RTC). The WuR widens the first-contact window, while the LP-RTC preserves the acquired phase for later exchanges. Energy gates, epoch-versioned schedules, and idempotent slot allocation extend this mechanism to static, single-hop All-to-One collection without a powered control anchor. The analysis accounts for role selection and duty-cycled listening, and bounds scheduled-retry tails only under explicit conditional quantile coverage. The study is deliberately scoped: it does not claim measured end-to-end energy, ambient-harvesting performance, or multi-hop scalability, and it notes that highly variable harvesting still limits collection because coordination cannot create missing energy.

MagPie's architecture separates the wake-up path from the timing path. A microampere WuR provides an always-on, ultra-low-power listening channel that can be triggered by a collector, effectively widening the window during which a battery-free node can be discovered. A separately backed LP-RTC maintains wall-clock phase across main-domain brownouts, so a node that has already completed a first rendezvous can re-enter the schedule at the correct epoch rather than restarting discovery. Three mechanisms extend this core to All-to-One collection: energy gates that admit a node into a slot only when sufficient energy is available; epoch-versioned schedules that invalidate stale slot assignments after a brownout; and idempotent slot allocation that tolerates duplicate requests without corrupting the schedule. The analytical model incorporates role selection and duty-cycled listening, and it bounds scheduled-retry tails only under explicit conditional quantile coverage. Evaluation is organized into three scopes. First, independent, administratively censored simulation trials compare MagPie against a Find baseline across five trace-parameterized harvesting scenarios. Second, a single-collision-domain slotted-Aloha study examines adaptive K under high contender density, reporting mean, P95, and confidence intervals through 120 components. Third, controlled STM32WL33 experiments validate alignment, clock persistence, and six-device slot execution, including an 11.05-hour functional run. The retry-tail bound can be expressed as a conditional quantile statement: for a scheduled retry count and a target quantile , the tail is bounded only when the conditional coverage assumption holds, i.e.,

, where
denotes the conditioning event on energy availability and role state. The slotted-Aloha contention process can be summarized by the adaptive-K rule
, where
is the estimated contender count at frame and is a scaling factor. The overall flow is shown below.

Why it matters

The central insight of MagPie is that first-contact and phase persistence are distinct problems that require distinct hardware support. A wake-up radio alone widens the discovery window but does not preserve timing state; an LP-RTC alone preserves phase but cannot help a node that has never been discovered. By separating these concerns, MagPie avoids the need for a powered control anchor, which is often unavailable in battery-free deployments. The energy-gate, epoch-versioned schedule, and idempotent slot allocation mechanisms are what make the approach robust to brownouts: stale schedules are invalidated, duplicate requests are harmless, and nodes only participate when energy permits. The analytical treatment is appropriately conservative. Bounding scheduled-retry tails only under explicit conditional quantile coverage means the bound is not a blanket guarantee; it holds when the conditioning assumptions about energy availability and role state are satisfied. This is an honest limitation, and it aligns with the paper's refusal to claim measured end-to-end energy or multi-hop scalability. The slotted-Aloha result reinforces a broader lesson for battery-free networking: contention control must adapt to contender density, because fixed-K policies collapse at high density. The 11.05-hour STM32WL33 run provides a concrete, if narrow, validation of clock persistence and six-device slot execution. The taxonomy candidates for this work span Architecture, Cybersecurity, Network, and Cryptography; the security relevance arises because epoch-versioned schedules and idempotent allocation are effectively lightweight anti-replay and anti-desynchronization primitives, though the paper does not develop a formal threat model. Overall, MagPie is best understood as a coordination-layer contribution: it improves rendezvous reliability and schedule integrity under intermittent energy, while leaving energy harvesting, end-to-end energy accounting, and multi-hop routing as open problems.

Who should read this

CS practitioners and researchers

Opening member contentโ€ฆ