TimeCast: Real-Time Many-to-Many Data-Sharing in Low-Power Wireless Distributed Systems

Many-to-many/all-to-all sharing of data is one of the common requirements in any distributed/decentralized system. In low-power wireless distributed systems composed of highly resource-constrained devices, a many-to-many data-sharing strategy needs to be not only efficient but also sufficiently simple in nature. Advancements in the field of synchronous transmission (ST) have proliferated a number of efficient data-sharing strategies for low-power systems. In this work, we first identify the fundamental and simple possible approaches to accomplish the goal. Through extensive simulation-based comparison among them, we demonstrate the superiority of the approach adopted by a strategy called MiniCast where the desired task is achieved through pipelined execution of the one-to-many floods from the designated source nodes. However, through further in-depth study, we identify certain crucial flaws in MiniCast, which are mostly caused by its substandard realization through radio-event-triggered architecture. To circumvent the problems, we propose a new strategy TimeCast where the radio-event-triggered architecture of MiniCast is replaced by a time-triggered architecture. Through rigorous experimental studies in IoT-testbeds, we demonstrate that TimeCast substantially improves the robustness, and thus, due to lesser failures, it achieves all-to-all data-sharing up to 57.1% faster while consuming up to 63.6% lesser radio- on time compared to MiniCast.