Abstract
Concurrent transmissions (CTX) are a state-of-the-art technique for low-power wireless networking, enabling performance unmatched by conventional routing-based approaches. However, their use is notoriously complex, as it hinges on many low-level system aspects distracting developers from the protocol logic. This is exacerbated in ultra-wideband (UWB) radios, recently shown to enable higher CTX performance than IEEE 802.15.4 and BLE, but significantly more energy-hungry, posing new problems to developers. A case in point is the deep sleep mode in popular UWB radios: it yields negligible consumption (50 nA) in periods of inactivity—crucial in real-world deployments—during which, however, clock drifts may disrupt the tight time synchronization required by CTX. Existing systems avoid the complexity of reconciling these conflicting needs via solutions with significantly higher consumption, hampering real-world applicability. We address these challenges with SAMU, a novel framework allowing developers to fully exploit the advantages of CTX over UWB via simple, expressive, and flexible abstractions removing the underlying complexity, embodied in an optimized and efficient runtime support. A distinctive trait of SAMU is the fine-grained control it provides over individual CTX, unlocking design opportunities unavailable in existing frameworks offering only high-level building blocks (e.g., CTX-based floods). We illustrate the power of SAMU by showing how existing CTX protocols can be easily built atop it, and evaluate the resulting performance in a large-scale testbed. Our results confirm that SAMU successfully combines high-level programming with a low-latency, energy-efficient, and reliable runtime support, providing a foundation for the protocol design and real-world deployment of CTX-based systems.
Funding Information
  • European Union?Next Generation EU under the Italian National Recovery and Resilience Plan (CUP B53C22003970001)
  • Agenzia Spaziale Italiana (2024-5-E.0CUP Master n. I53D24000060005)

This publication has 25 references indexed in Scilit: