Connectionless Bluetooth LE Channel Sounding via PAwR for Scalable and Energy-Efficient Ranging

TL;DR

This paper introduces a connectionless Bluetooth LE channel sounding architecture using PAwR, improving scalability and energy efficiency by reducing overhead, collisions, and active device charge in dense deployments.

Contribution

It presents a novel connectionless CS architecture combining LE CS Test with PAwR, enabling scalable, collision-free, and energy-efficient BLE ranging in dense environments.

Findings

Reduces active device charge by up to 48% in steady state.

Cuts initiation overhead by approximately 98%.

Projects capacity of 16,384 active devices per PAwR train.

Abstract

Bluetooth Core Specification v6.0 introduces Channel Sounding (CS) as a standardized high-accuracy ranging primitive for Bluetooth Low Energy (BLE). However, standard CS usage remains tied to per-pair LE asynchronous connection logical transport (LE ACL) connections, which adds initiation overhead, limits concurrent partners, and transfers results over the connection itself. We present a connectionless CS architecture that combines the LE CS Test command with Periodic Advertising with Responses (PAwR). A Central Orchestrator, a Gateway, and synchronized Tag/Anchor devices coordinate measurement configurations and aggregate results at the application layer. Each device derives its role, channel sequence, and response slot assignment from its device index and a Peer-to-Peer Assignment Matrix distributed via PAwR. The deterministic channel sequence prevents same-step collisions across parallel CS procedures, while matrix updates reconfigure arbitrary device-to-device pairings within a PAwR subevent group. A compact data plane omits fields recoverable from the shared measurement configuration and reduces the serialized ranging-data payload by approximately 69%, enabling result reporting through PAwR response slots. A proof-of-concept evaluation on the Nordic nRF54L15 platform shows that deterministic channel management eliminates the collision-induced outliers observed under simulated dense-deployment channel overlaps. At a 1 s update cycle, the architecture reduces steady-state active charge by 40-48% relative to a fair connected baseline and cuts per-switch initiation overhead by approximately 98%. Under per-cycle partner switching, these effects combine to up to 88% lower total charge over a 24 h horizon. An empirical timing model projects a capacity upper bound of 16,384 active devices per PAwR train at four CS procedures per device per cycle, 37 channels, and a single antenna path.