SyncCast: Real-Time Self-Adaptive and Fault-Tolerant All-to-All Data-Sharing in IoT
Jagnyashini Debadarshini, Sudipta Saha · 2023
Advancements in the domain of Synchronous- Transmission (ST) have brought a drastic improvement in the all-to-all data-sharing strategies in low-power wireless distributed systems. However, in the current design of the ST-based protocols, the inability of overhearing from the neighbors, and the lack of simultaneous system-wide commencement of the target process act as major bottlenecks in the faster percolation of data. The physical layer phenomena Capture-Effect (CE) is considered to be one of the fundamental strengths of ST. However, since the scope of CE largely depends on network topology, an ST-based protocol also mostly fail to exploit the full benefit of CE. Existing strategies try to exploit the topology information implicitly through on-the-fly exploration which, however, results in only partial advantage. In this work, we first introduce a few fundamental design ingredients to accelerate the percolation of data. Next, we propose a TDMA-based self-adaptive all-to-all data-sharing strategy SyncCast which explicitly learns the topology of the underlying network in real-time, and then on-the-fly optimizes the transmission plans to maximize the advantage from CE. Through extensive experiments in IoT-testbeds we demonstrate that despite being simple, SyncCast is upto 65 % faster and consumes 74% less Radio-on time compared to the existing state-of-the-art solution for all-to-all data-sharing.