PaperlayerView exact revision on arXiv

Source-linked AI summary

Mobile Backscatter Communication for the Battery-less Internet of Things

Weining Song, Thiemo Voigt, Stefanos Kaxiras, Yuan Yao, Luca Mottola

arXiv:2609.01465v1cs.NI

TL;DR

Battery-less mobile IoT backscatter communication must cope with rapidly changing channels and unpredictable harvested energy. The paper introduces lightweight RSSI-trend transmission control with NVM persistence, achieving higher throughput and lower transmission energy than a rate-adaptive baseline.

  • Problem

    Existing backscatter systems assume static environments, whereas mobile energy-harvesting devices experience rapidly varying channel conditions and energy availability.

  • Method

    The system uses short-term RSSI trends to regulate transmissions and NVM to retain packets and control state across poor channels and energy failures.

  • Results

    Up to 5.16 × higher throughput and up to 47.3% lower transmission energy are achieved versus a rate-adaptive baseline, with 0.23% – 7.3% additional energy overhead.

  • Takeaways & Limitations

    Transmission control can exploit improving channel conditions while conserving energy and preserving packets for later opportunities in mobile, battery-less IoT.

Abstract

from arXiv · show

We enable backscatter communication in the battery-less mobile Internet of Things (IoT). Backscatter communication is extensively studied in static settings. Existing designs are, however, fundamentally mismatched with mobility and time-varying energy patterns. Channel conditions rapidly fluctuate, impacting the achievable data rates and thus transmission costs. Energy availability varies unpredictably, possibly forcing devices to remain quiescent to recharge energy buffers. The two issues compound each other: while recharging, a battery-less mobile IoT device may miss more favorable channel conditions. We design a lightweight decision system that dynamically determines when to transmit by checking short-term trends in signal strength, while using Non-volatile Memory (NVM) to retain packets in unfavorable channel conditions and across energy failures. Using a prototype we built and real-world mobility and power traces, we compare our design against a rate-adaptive baseline that only considers the instantaneous channel conditions. Experimental results show that our system improves throughput by up to 5.16x while reducing transmission energy consumption by up to 47.3%, with only 0.23% - 7.3% additional energy overhead.

I. INTRODUCTION

Battery-less mobile IoT backscatter systems must adapt to rapidly changing channel conditions and erratic harvested energy. This paper addresses that mismatch with lightweight trend-aware transmission control and NVM packet persistence.

  • Motivation: 131 kbps at 4 meters costs 1.344 µJ per packet, while longer distances reduce data rate and increase transmission energy.The relationship makes poor channel conditions especially costly for battery-less devices.
  • Motivation: Mobile energy-harvesting devices face rapidly varying channel conditions and scarce, erratic energy, creating coupled transmission-timing challenges.Energy failures can force devices to recharge while mobile, causing them to miss favorable channel conditions.
  • Approach: The system monitors RSSI trends rather than instantaneous signal strength, transmitting more aggressively as conditions improve and throttling transmissions as they worsen.This runtime adaptation trades packet latency for potentially higher data rates and lower transmission energy while requiring minimal compute overhead.
  • Approach: NVM stores unsent packets and control state across energy failures and poor channel conditions, allowing later transmission when energy and channel conditions improve.This extends NVM beyond checkpointing program state to proactively regulate communication decisions.
  • Evaluation: The prototype uses an MSP430FR5969 MCU and evaluates agricultural measurement, activity recognition, and cold-chain monitoring with real-world signal and power traces.The system is compared with a rate-adaptive baseline that adjusts data rate using instantaneous signal strength.
  • Related Work: The paper positions its contribution as system-level transmission control under mobility and energy harvesting, distinct from physical-layer mobility analysis and intermittent-carrier scheduling.Related designs include Xu et al.'s intermittent-WiFi system and Muratkar et al.'s physical-layer characterization.

III. SYSTEM DESIGN

The system design section presents the overall architecture and focuses on transmission control and energy management.

  • III. SYSTEM DESIGN: The system design includes transmission control and energy management as its primary features.

A. Intuition

Mobile backscatter must coordinate transmission with changing channel quality and harvested energy, rather than relying only on instantaneous signal strength. The design uses RSSI trends, energy-aware transmission aggressiveness, NVM buffering, and persistent state to exploit favorable opportunities despite intermittent operation.

  • Motivation: 1232 kbps at -14 dBm falls to 0.33 kbps at -46 dBm, while energy consumption rises from 0.14 µJ to 533.72 µJ for a 256-bit packet.The latter cost is over 3000 × higher, making poor-channel transmission especially costly.
  • Motivation: A transmission at 2.5 s can drop capacitor voltage from 4.0 V to 2.2 V and force recharging until 12 s, missing a strong signal window at 7 s–9 s.This illustrates how energy depletion and mobility jointly affect transmission timing.
  • RSSI Trend Recognition: An EMA smooths recent RSSI samples, adapts its smoothing factor to irregular sampling intervals, and uses T_n = R_n − S_n to identify channel trends.Positive T_n indicates increasing channel quality, while negative T_n indicates decreasing quality; τ controls filter memory.
  • Transmission Budget: The aggressiveness counter combines RSSI trends with recent charging duration to determine how many packets to schedule during each active period.Longer charging implies fewer future active phases, so the system increases concentration of transmissions; decreasing trends reduce the counter toward NVM buffering.
  • Energy Management: NVM stores packets and control state across low-energy periods, allowing execution to resume without losing data after energy failures.The persisted state includes the aggressiveness counter and RSSI trend-recognition algorithm state.

IV. EVALUATION

The evaluation uses a prototype, benchmarks, baselines, and real-world signal and power traces to assess throughput, energy consumption, and control overhead. Compared with a rate-adaptive baseline, the design substantially improves throughput while reducing transmission energy with limited added overhead.

  • Evaluation Scope: The evaluation measures throughput under real-world signal and power traces and analyzes energy consumption and transmission-control overhead.Two microbenchmarks additionally examine how system factors and parameter settings affect throughput.
  • Results: 5.16 × throughput improvement over a rate-adaptive baseline is achieved by the proposed design.This is the reported maximum throughput gain.
  • Results: 47.3% lower transmission energy consumption is reported with only 0.23%–7.3% additional energy overhead.These are the reported maximum reduction and overhead range.

A. Prototype

The prototype combines low-power control, RSSI sensing, backscatter hardware, timekeeping, and energy management on off-the-shelf components. Its controller runs on an MSP430FR5969 MCU and uses a separate CC1352P7 for RSSI measurements.

  • Hardware: The controller runs application logic and transmission control on an MSP430FR5969 MCU with 64 KB of on-chip FRAM.The prototype uses off-the-shelf hardware.
  • Hardware: A TI CC1352P7 supplies RSSI measurements to the MSP430FR5969 over SPI, while the backscatter tag uses the LoRea architecture.The signal-strength sensing and backscatter components are separate prototype elements.
  • Timing: An EM3028-C7 RTC supports time-adaptive trend recognition and aggressiveness control while timestamping application-layer data.The RTC provides timing support for the prototype’s transmission-control logic.

B. Benchmarks

The evaluation benchmarks the mobile backscatter system across representative applications, real-world mobility and energy traces, and baselines with different packet-preservation capabilities.

  • Applications: The study evaluates AEM, AR, and CEM applications with differing computation, buffering, and transmission demands.The applications are agricultural environment measurement, activity recognition, and cold-chain equipment monitoring.
  • Throughput benchmark: 1.48 × average throughput is achieved by the mobile backscatter system in the normalized average-throughput benchmark.The benchmark reports average throughput relative to the rate-adaptive baseline.
  • Hardware setup: The prototype uses a capacitor sized to meet the maximum energy demand of each application.This sizing allows even the most energy-intensive application to complete an active period.
  • Evaluation framework: A measurement-driven simulation framework models long-term behavior using prototype measurements of energy consumption and execution time.The framework is used because extended prototype runs under reproducible signal and energy conditions are impractical.
  • Traces: The simulator replays two one-hour jogging power traces and 20 indoor signal traces collected across movement paths and speed profiles.Signal traces last around two minutes and are extended to one hour by repetition.
  • Baselines: The rate-adaptive baseline changes data rate using instantaneous signal strength, while the NVM-based extension stores packets blocked by weak signals or energy failures.The NVM-based baseline preserves packets but does not use channel trends or energy status for transmission control.

E. Throughput

Throughput improves when transmission control uses signal trends and buffering to defer packets until stronger conditions, but gains diminish when traces lack clear trends or transmission opportunities.

  • Average throughput: 1.39×, 1.40×, 1.22×, 1.48×, and 1.42× throughput gains are reported across applications and power traces relative to the rate-adaptive baseline.AEM achieves 1.39× under both power traces; AR achieves 1.40× and 1.22×; CEM achieves 1.48× and 1.42×.
  • Signal-dependent gains: 63.4% of the C2 4 V trace remains below the transmission threshold, creating weak-channel periods where NVM buffering and RSSI trends improve throughput.The rate-adaptive baseline discards packets when transmission is impossible, while the NVM baseline stores packets without exploiting signal trends or energy status.
  • Exceptions: The AR application performs worse than the NVM-based rate-adaptive baseline under jogging node 1 and C2 4 V.Insufficient transmission opportunities cause packet retention in NVM, while the baseline transmits whenever minimum signal and energy requirements are met.
  • Signal-dependent limits: C1 3 C and C1 3 V produce the lowest throughput because RSSI fluctuates between -20 dBm and -30 dBm without a clear trend.Without weak-channel periods requiring deferral, transmission control has limited opportunity to improve performance.
  • Limitations: The mobile backscatter system offers comparable or slightly reduced performance in scenarios lacking clear signal trends.The AR application also shows slightly lower throughput when its lightweight workload makes overhead more noticeable.
  • Throughput per trace: 619 packets are transmitted by the mobile backscatter system, versus 454 for rate-adaptive and 555 for NVM-based rate-adaptive.The system shifts transmissions toward stronger signal regions between -37.5 dBm and -28 dBm.

F. Energy

The mobile backscatter system reduces transmission energy by deferring transmissions to favorable channel conditions, improving throughput across applications while adding limited control overhead.

  • AEM: 12% total system energy savings accompany a reduction in AEM transmission energy from 16% to 4%.The system achieves 1.36 × the rate-adaptive baseline's throughput in AEM.
  • AR: 16% lower total system energy consumption accompanies a reduction in AR transmission energy to 27%.Throughput reaches 1.36 × the rate-adaptive baseline.
  • CEM: 1.50× throughput is achieved in CEM by reducing transmission energy to 8%, compared with 24% and 26% for the baselines.CEM transmits less frequently because it compresses data until a block is full, increasing the relative importance of RSSI monitoring.
  • Overall energy: 47.3% lower transmission energy consumption is achieved compared with the rate-adaptive baseline.Transmission-control overhead ranges from 0.23% to 7.3% of total energy across evaluated configurations.

G. Packet Age

The system increases packet age because it stores packets in NVM and proactively defers transmission, with especially long delays when transmission opportunities are limited.

  • Definition and mechanism: Packet age is measured from packet generation to transmission across the three applications and all signal and power traces.The mobile system stores packets in NVM both when transmission is impossible and when it proactively defers transmission.
  • Packet-age comparison: 13.25 seconds is the average packet age for the NVM-based rate-adaptive baseline.The standard rate-adaptive baseline is omitted because it transmits immediately based on instantaneous signal strength.
  • AR behavior: 338 and 505 seconds are the average packet ages for AR under the H1 3 C and H1 3 V signal traces, respectively.These long ages arise from limited transmission opportunities under the jogging node 0 power trace.

H. Micro-benchmarks

Micro-benchmarks show that throughput gains increase with weaker signal periods and persist across rate–RSSI profiles, although an AR exception occurs at a −14 dBm offset.

  • Signal traces: Higher throughput emerges as downward RSSI offsets increase, because longer weak-signal periods create more opportunities for trend-based deferral.The original trace falls below the transmission threshold only 7% of the time.
  • Exception: At a −14 dBm offset, AR throughput is lower than the NVM-based rate-adaptive baseline.Many packets are generated during weak signal periods, while favorable conditions become less frequent.
  • Rate–RSSI profiles: Lower maximum data rates produce higher throughput for the mobile backscatter system across different rate–RSSI profiles.The measured profile was replicated from prior work, alongside idealized profiles with 500 kbps and 2000 kbps maximum rates.

I. Adaptability Under Dynamic Conditions

The system targets battery-less mobile IoT devices by adapting transmission decisions online to changing RSSI and energy availability rather than relying on environment-specific offline tuning.

  • Dynamic conditions: Offline optimization can exploit known emitter placement, mobility patterns, and energy intakes, but these factors are fundamentally unpredictable in practice.This unpredictability makes pre-tuned transmission strategies impractical.
  • Adaptive control: Online transmission adaptation uses observed RSSI trends and energy availability to exploit opportunities under changing conditions.The approach avoids environment-specific tuning.
  • Evaluation outcome: Up to 5.16 × higher throughput and up to 47.3% lower transmission energy are reported with 0.23%–7.3% additional energy overhead.The evaluation uses a real hardware prototype, three IoT applications, and real-world power and signal traces.
Loading 2609.01465v1…