A Sensor Data Stream Recovery Scheme for Event-Driven IoT Applications
Yuuichi Teranishi, Ei Khaing Win, Tomoki Yoshihisa, Shinji Shimojo · 2017
In this paper, we propose a novel efficient sensor data stream recovery scheme called "synchronized recovery stream merging (SRSM)," for reliable event-driven IoT applications, assuming multicast delivery of data streams. SRSM synchronizes and merges multiple data streams for recovery by waiting for other data streams keeping the delivery data order, as long as the required restriction for delivery latency is satisfied. SRSM reduces the number of streams required to generate and send on the data stream sender dramatically than applying an existing stream merging scheme for recovery, thus enables to save the network bandwidth of the sender. Through extensive simulations, we evaluated the performance of SRSM and confirmed that about 48% - 60% network bandwidth was saved on the sender in the frequent failure situation compared to the existing schemes, keeping the average satisfy rate of requested delivery latency.