PaperlayerView exact revision on arXiv

Source-linked AI summary

Analytical Modeling of Latency and Energy Trade-offs in FiRa-Like BLE-UWB Ranging and Localization Systems

Adelina Giurea, Jorg Wieme, Ben Van Herbruggen, Jeroen Hoebeke, Eli De Poorter

arXiv:2610.33423v1cs.NI

TL;DR

UWB receive-mode power makes continuous monitoring impractical, while BLE discovery and commissioning can make setup costly for short interactions. This paper models end-to-end BLE-assisted UWB sessions and finds discovery highly asymmetric in energy, with long-tailed latency.

  • Problem

    UWB receive-mode power makes continuous monitoring impractical, while BLE discovery and commissioning can impose substantial setup costs during short interactions.

    Source: §I · p. 1

  • Method

    The paper presents an analytical model combining BLE discovery and negotiation with UWB startup and ranging to quantify session latency and energy by phase and device role.

    Source: §VII · p. 13

  • Results

    28×–57× more energy per discovery is consumed by the scanner than the advertiser; 99th-percentile discovery latency is 2×–3.4× the mean, reaching 110 s.

    Source: §VII · p. 13

  • Takeaways & Limitations

    BLE discovery can dominate latency and energy for short interactions, making discovery configuration consequential alongside the ranging scheme.

    Source: §I · p. 1

    The analysis assumes a single advertiser-scanner pair and, for localization, one tag interacting with independent anchors on an otherwise idle medium.

    Source: §VI · p. 13

Abstract

from arXiv · show

Accurate position and distance information provided by Ultra-Wideband (UWB) can enhance a wide range of applications. However, its receive-mode power consumption makes continuous channel monitoring impractical for energy-constrained devices. Consequently, UWB systems often rely on Bluetooth Low Energy (BLE) to discover peers and establish sessions before ranging begins. Although both technologies have been studied in isolation, an framework capturing the latency and energy consumption of the complete workflow is still lacking. To address this gap, we develop a unified analytical model for BLE-assisted UWB systems, from device-to-device ranging to infrastructure-based localization. The model includes closed-form expressions for discovery, parameter exchange, and ranging, and evaluates the resulting latency and energy consumption. Evaluation of three representative BLE configurations shows that the scanner consumes 28-57x more energy per discovery than the advertiser. In addition, the 99th-percentile discovery latency is 2-3.4x higher than the mean and reaches up to 110 s in the lowest-power configuration. The one-time BLE setup overhead is amortized after 24-562 ranging rounds, depending on the discovery configuration. Overall, the analysis demonstrates that the BLE discovery phase can dominate both the latency and the energy budget during short UWB interactions, making discovery configuration as important as the ranging scheme.

I. INTRODUCTION

UWB offers high-accuracy ranging, but its receive-mode energy cost makes continuous discovery impractical, motivating BLE-assisted setup. The paper develops an end-to-end model and positions it against prior work on BLE discovery, UWB ranging, hybrid radios, and standardization.

I. INTRODUCTION

  • The model links BLE discovery and negotiation with UWB ranging or localization, addressing the unmodeled end-to-end costs of setup and multi-anchor sessions.It covers BLE discovery, parameter negotiation, UWB startup, and ranging, and extends to multi-anchor TWR, UL-TDoA, and DL-TDoA.Source: §I · pp. 1–2
  • The analysis quantifies device asymmetry, latency growth with anchors or ranging rounds, and the rounds needed to amortize BLE setup.Source: §I · p. 2

II. RELATED WORK

  • The related-work section is organized around BLE neighbor discovery, UWB ranging and localization, hybrid multi-radio systems, and standardization efforts.Source: §II · p. 2

A. BLE Neighbor Discovery: Latency and Energy

  • Prior BLE discovery models differ in scanning and exchange assumptions, while existing analyses do not carry discovery outcomes into later session establishment and ranging.This model refines the scan-window treatment and maps the successful round and channel into device energy and subsequent phases.Source: A. BLE Neighbor Discovery: Latency and… · p. 2

B. UWB Ranging and Localization: Latency and Energy

  • Prior UWB studies often assume devices are already configured, leaving the time and energy of obtaining that configuration outside their scope.The proposed model connects BLE discovery and negotiation to later UWB ranging or localization.Source: B. UWB Ranging and Localization:… · p. 3

C. Hybrid UWB and Out-of-Band (OOB) Systems

  • Hybrid companion-radio approaches include BLE-assisted systems as well as alternatives that target preconfigured nodes and mains-powered anchors.UWB-ND can reduce tag energy by up to 50% for rare requests but increase it by up to 64% for frequent requests, compared with PI discovery.Source: C. Hybrid UWB and Out-of-Band (OOB)… · p. 3

D. Standardization and Ecosystem Trends

  • FiRa and other ecosystems use BLE or another out-of-band transport for discovery and session setup before UWB ranging.The capability request, response, and configuration messages defined by Android provide the basis for the model’s negotiation phase.Source: D. Standardization and Ecosystem Trends · p. 3

III. ANALYTICAL MODEL

The model combines BLE discovery, BLE parameter exchange, and UWB startup and ranging to estimate end-to-end delay and energy. Its discovery analysis links unsynchronized advertising and scanning to packet reception.

III. ANALYTICAL MODEL

  • The end-to-end model combines BLE discovery, BLE parameter exchange, and UWB startup and ranging before the session ends.The workflow then branches into UWB ranging or localization.Source: §III · p. 3, Figure 1 · p. 4

A. BLE Neighbor Discovery

  • Discovery requires the scanner's scan window to overlap an ADV_IND packet on the same channel while the advertiser cycles through channels 37, 38, and 39.Advertising events repeat at Tadv = TAI + µ, while the scanner visits channels cyclically using scan intervals and windows.Source: A. BLE Neighbor Discovery · pp. 3–4, Figure 2 · p. 4

1) Timing:

  • Packet airtime is computed from packet type, payload size, and PHY rate, with a 4 B MIC on encrypted data packets and none otherwise.The packet format used for the timing model is LE 1M PHY.Source: 1) Timing · p. 4, Table II · p. 4

2) Discovery Latency:

  • Expected discovery latency averages over the unknown phase offset; advertiser latency starts at its first ADV_IND transmission, while scanner latency starts at scanning.Reception requires synchronization before the scan window ends, although reception is assumed to finish once synchronization succeeds.Source: 2) Discovery Latency · pp. 4–5

Z 3TSI

The model distinguishes two BLE discovery modes according to the required UWB session type. It accounts for advertiser- and scanner-side energy across unsuccessful and successful discovery events.

Z 3TSI

  • Scannable discovery uses SCAN_REQ/SCAN_RSP without forming a BLE connection; connectable discovery uses CONNECT_IND to establish one for parameter exchange.Scannable discovery can identify infrastructure for TDoA localization without further negotiation or authentication; connectable discovery supports required security-material negotiation.Source: Z 3TSI · p. 5

3) Energy Consumption:

  • The energy model decomposes advertising events and scan intervals into advertiser- and scanner-side costs, distinguishing unsuccessful events from the successful event.On the scanner side, unsuccessful intervals include the full scan window and low-power idle time; in the successful interval, listening ends after ADV_IND reception and the mode-dependent exchange.Source: 3) Energy Consumption · pp. 5–6
  • Expected discovery energy is averaged over the unknown phase offset, and the successful advertiser event terminates differently in scannable and connectable modes.In scannable discovery the advertiser completes all three channels; in connectable discovery it closes the event after CONNECT_IND.Source: 3) Energy Consumption · pp. 5–6

Z 3TSI

The passage points to the expected-energy expression as being derived analogously to the discovery-latency expression.

Z 3TSI

  • The expected-energy expression is formed analogously to the latency expression in (10).The passage identifies this as equation (28).Source: Z 3TSI · p. 6

B. BLE Negotiation and Security Exchange

After connectable BLE discovery, the devices establish a connection and negotiate UWB session parameters before ranging. The modeled exchange uses three BLE messages, each fitting in one packet.

B. BLE Negotiation and Security Exchange

  • After CONNECT_IND, the devices enter the BLE Connection state, and the first received packet establishes the connection.The model assumes LE 1M PHY in both directions and excludes a PHY update procedure.Source: B. BLE Negotiation and Security Exchange · p. 6

1) Per-Message Payload Sizes:

  • The model assumes that devices exchange a profile identifier and session-specific parameters rather than a complete set of PHY- and MAC-layer configuration values.Typical UWB initialization includes an address, session ID, STS session key, ranging role, and configuration profile.Source: 1) Per-Message Payload Sizes · p. 6

2) Message Flow:

  • The parameter flow requests supported ranging capabilities, returns the supported parameter sets, then sends the selected values with the session ID and STS.The messages are a capability request, capability response, and ranging configuration.Source: 2) Message Flow · pp. 6–7
  • The three-message exchange uses exactly three BLE packets; its largest message is 43 B, below the 250 B payload maximum, and ranging starts after the last packet.The third message is the dominant message and varies from 19 to 43 B with the selected STS mode.Source: 2) Message Flow · p. 7

3) Packet Airtime:

This section models the airtime and energy of BLE commissioning and UWB session setup. It highlights that negotiation latency can be role-symmetric even when commissioning energy is not.

3) Packet Airtime:

  • The encrypted parameter exchange occupies 0.89–1.08 ms over the considered payload range, well below the minimum connection interval.Source: 3) Packet Airtime · p. 7

4) Negotiation Latency:

  • From CONNECT_IND completion to M3, negotiation latency is identical for both roles, assuming the central transmits uniformly within its window.Source: 4) Negotiation Latency · p. 7

5) Energy Consumption:

  • The peripheral pays a receive-window penalty and receives the largest packet while transmitting only the middle one, making commissioning energy unequal despite equal latency.Source: 5) Energy Consumption · p. 7

C. UWB Ranging

  • After BLE negotiation, both devices activate and configure their UWB subsystems using the exchanged session parameters.Source: C. UWB Ranging · p. 7

1) Startup:

The section derives UWB startup, packet-airtime, ranging-latency, and energy quantities from transceiver preparation and packet structure. It distinguishes ranging modes and device roles in those calculations.

1) Startup:

  • Before transmission or reception, UWB wake-up includes oscillator start-up and an idle interval for RF synthesizer lock.Source: 1) Startup · p. 7

2) Packet Airtime:

  • UWB packet airtime is based on the SHR and data portion; SP3 carries only SHR and STS for timestamp-only frames such as POLL and RESPONSE.Source: 2) Packet Airtime · p. 8

3) Ranging Latency and Energy Consumption:

  • SS-TWR uses one request-reply exchange, while DS-TWR adds a reverse-direction exchange to cancel most clock-drift error; ranging energy also depends on frame airtime, mode, and device role.Source: 3) Ranging Latency and Energy… · p. 8

T UWB

D. End-to-End Latency and Energy

End-to-end session metrics combine BLE discovery, optional commissioning, and UWB ranging. Latency starts when scanning begins and ends after the final ranging round.

D. End-to-End Latency and Energy

  • Session metrics compose BLE discovery, BLE negotiation/commissioning, and UWB ranging, with commissioning included for connectable discovery but absent for scannable discovery.Source: D. End-to-End Latency and Energy · p. 9

1) End-to-end Latency:

  • Session latency runs from the initiator starting to scan until the last ranging round completes; discovery uses scanner latency, while later phases are common to both devices and independent of phase offset.Source: 1) End-to-end Latency · p. 9

2) End-to-end Energy:

  • Energy is accounted per device because the roles transmit and receive different numbers of frames in each phase.Source: 2) End-to-end Energy · p. 9

IV. DISCUSSION AND MODEL SCOPE

Across the evaluated BLE configurations, discovery materially affects session energy and latency, while its setup cost is amortized at configuration-dependent ranging counts. Lower duty cycles reduce waiting power but increase discovery latency and energy per event.

IV. DISCUSSION AND MODEL SCOPE

  • The evaluation uses connectable BLE discovery with parameter negotiation followed by DS-TWR, across low-latency, balanced, and low-power BLE configurations.Source: §IV · p. 9

A. Energy Costs of the Scanner and the Advertiser Roles

B. Discovery Latency Analysis

C. Discovery Energy versus Ranging Energy

V. FROM RANGING TO LOCALIZATION

The localization extension models discovery of an anchor group and compares tag-side UWB latency and energy across TWR, UL-TDoA, and DL-TDoA. It assumes coordinated infrastructure and distinguishes schemes by how many frames the tag exchanges or receives.

A. Discovery of an Anchor Group

  • Group discovery ends when any in-range anchor is found; its expected latency decreases as the number of visible anchors increases.The anchor offsets are independent, and three visible anchors halve latency versus one in the reported discussion.Source: A. Discovery of an Anchor Group · p. 11

B. UWB Phase

  • Localization assumes BLE-visible anchors are also in UWB range, synchronized TDoA anchors, and infrastructure-provided schedules; scheduling and synchronization are outside the model.At least three anchors are needed in 2D and four in 3D.Source: B. UWB Phase · p. 11
  • TWR repeats a full exchange per used anchor, UL-TDoA has the tag transmit one frame received by visible anchors, and DL-TDoA has anchors transmit frames for the tag to receive.TWR uses at least three anchors in 2D and four in 3D; UL-TDoA adds no tag exchange per additional anchor.Source: B. UWB Phase · pp. 11–12

T UWB

  • The total UWB-phase latency adds infrastructure reporting and computation only for UL-TDoA, where the infrastructure computes the position.TWR and DL-TDoA have the tag compute its own position.Source: T UWB · p. 12
  • UL-TDoA tag energy is independent of anchor count because the tag transmits one frame, whereas DL-TDoA requires receiving frames from visible anchors.The model counts tag energy because anchors are typically mains-powered.Source: §V · p. 11, T UWB · p. 12

C. Discussion

The discussion finds that more visible anchors reduce expected group-discovery latency, while localization schemes scale differently in tag-side latency. The conclusion emphasizes discovery’s importance and identifies modeling assumptions that constrain the analysis.

C. Discussion

  • DS-TWR latency rises by 1.26 ms per anchor, from 4.8 ms with three anchors to 26.1 ms with twenty, because each adds a three-frame exchange and idle gap.The discussion notes that secure localization also requires BLE discovery and negotiation with each participating anchor.Source: C. Discussion · p. 12
  • DL-TDoA latency rises by 0.56 ms per anchor as the tag receives each anchor’s frame and remains awake between receptions.The tag’s energy also grows with visible-anchor count because deep sleep is unavailable during the block.Source: C. Discussion · p. 12
  • UL-TDoA tag latency stays at 1.2 ms regardless of anchor count because the tag transmits one frame, while position computation remains in the infrastructure.The discussion presents this as attractive for low-power, large-scale industrial RTLS applications.Source: C. Discussion · pp. 12–13
  • With twenty visible anchors, expected group-discovery latency falls by a factor of 10 to 13 across configurations, reaching 16 ms, 88 ms, and 2.8 s.Three visible anchors already halve latency versus one: 154 ms to 77 ms for low-latency and 35.8 s to 11.3 s for low-power.Source: C. Discussion · p. 12, Figure 8 · p. 13

VI. LIMITATIONS AND FUTURE WORK

  • The analysis models a single advertiser-scanner pair and a single tag interacting with independent anchors on an otherwise idle medium; contention, collisions, and anchor scheduling remain unmodeled.Its energy model also assumes sleep between scan and advertising windows, and uses a mean-delay approximation for non-resonant configurations.Source: §VI · p. 13

VII. CONCLUSION

  • The conclusion reports that scanners use 28×–57× more energy per discovery than advertisers, and discovery’s 99th-percentile latency is 2×–3.4× the mean, reaching 110 s.BLE setup overhead is amortized only after 24–562 ranging rounds; discovery can dominate short-session latency and energy.Source: §VII · p. 13

Related papers

Loading 2610.33423v1…