Decentralized Data Flows in Algebraic Service Compositions for the Scalability of IoT Systems

Abstract

With the advent of the Internet of Things, scalability becomes a significant concern due to the huge amount of data involved in IoT systems. A centralized data exchange is not desirable as it leads to a single performance bottleneck. Although a distributed exchange removes the central bottleneck, it has network performance issues as data passes among multiple coordinators. A decentralized data flow exchange is the only solution that fully enables the realization of efficient IoT systems as there is no single performance bottleneck and the network overhead is minimized. In this paper, we present an approach that leverages the algebraic semantics of DX-MAN for realizing decentralized data flows in IoT systems. As data flows are not mixed with control flows in algebraic service compositions, we developed an algorithm that smoothly analyzes data dependencies for the generation of a direct relationship between data consumers and data producers. The result prevents passing data alongside control among multiple coordinators because data is only read and written on a data space. We validate our approach using the Blockchain as the data space and conducted experiments to evaluate the scalability of our approach. Our results show that our approach scales well with the size of IoT systems.

An Empirical Study on System Level Aspects of Internet of Things (IoT)

“…Growth in the number of IoT devices and their associated data results in scalability issues. [146]. IoT scalability is characterized as a system's ability to manage the increasing number of devices and their information.…”

Evaluating IoT service composition mechanisms for the scalability of IoT systems

“…The result of composition is not a workflow, but an IoT composite service which is semantically equivalent to a potentially infinite family of (explicit) workflow control flows. As DX-MAN takes the best properties from choreography and orchestration, it enables decentralised data exchanges over the network while decoupling services via (exogenous) workflow control flows [162]. Currently, there is only one platform available to support the definition of algebraic IoT systems [163].…”

“…IoT services must exchange data as efficiently as possible (by minimising network hops), in order to avoid performance bottlenecks, achieve better response time and improve throughput [147]. A decentralised approach provides the most suitable data exchange for service composition since it requires only one network hop to pass data directly from a service producer to a service consumer (see Figure 10(c)) [34,117,152,162,195,197,198,199]. For example, the data generated by Booking.book should be passed to Display.showMap, without passing through any other entity that does not require the produced data (e.g., the CityManagement composite or the Vehicle composite).…”

“…The control flow part defines a workflow for the sequential invocation of . The separation of data flow will additionally enable separate data management for improving performance without considering control flow [35,162]. Similarly, control flow can be used in isolation for defining efficient deployment strategies that do not consider data flow.…”

“…Although a central coordinator facilitates the management of control flow logic, it easily becomes a performance bottleneck because all data passes through it [140,142]. This is because data follows control [71,162]. Furthermore, the resulting composite service is a single flat workflow, for the invocation of named and selected operations, which must be transformed into a service [159,240].…”