Data Alignment for Multiple Temporal Data Streams without Synchronized Clocks on IoT Fusion Gateway

Abstract

Communication among devices and collecting concurrent stream data in M2M, CPS, and IoT are the essential features for these systems. For the sake of simplicity and reliability, many middle-ware solutions support homogeneous/single network deployment. However, when the number of device vendors continue to grow, it is inevitable that the system consist of devices having heterogeneous network interfaces. WuKong is an intelligent middle-ware for machine-to-machine, cyberphysical, and Internet-of-Things systems. To support dynamic connectivity and heterogeneous network protocols in the system, Multiple Protocol Transport Network (MPTN) gateway is designed to be a distributed messaging gateway to enable messaging among multiple networks. To leverage the routing capability in existing network protocols, MNTP gateway converts an end-toend message request to a multiple segment messaging based on network topology available to MNTP gateway. To collect concurrent stream data, MPTN gateway supports data alignment mechanism to label sampling data onto a common time domain. The mechanism adapt a coordinated model to collect concurrent data streams, and convert the sample counter/time on sensor device to time instances on time domain of MPTN gateway. The evaluation results show that the data alignment errors are much less than sampling errors and can be ignored.