Page 86 - ITU Journal Future and evolving technologies Volume 2 (2021), Issue 5 – Internet of Everything

ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 5

3.2.1 Data storage, integration and exchanges The Message and Service Bus (MSB) plays a key role in
this direction. It acts as the mediation middleware be‑
The main sources of information of the COG‑LO frame‑ tween the various components comprising the COG‑LO
work are the databases of the pilot partners. They pro‑ ecosystem, and is assigned with the interaction, coordi‑
vide the core data set, based on which all processes of the nation and orchestration of COG‑LO components and op‑
system are executed. Each data model design, as well as erations. In that respect, the MSB supports message ex‑
the storage scheme of each source, is different and out of change between system entities, circulation of events, in‑
the scope of the project. The information integration is teraction between CLOs. To this end, it facilitates cooper‑
achieved through data connectors that are implemented ation within communities of CLOs and creation of ad hoc
as part of the Message and Service Bus of COG‑LO. They channels between CLOs by enabling the dynamic estab‑
are attached to the data sources and they assume the re‑ lishment of message topics.
sponsibility of transforming source data into the COG‑LO Another fundamental challenge that the MSB tackles
data model. is the transformation of platform independent speci i‑
The core system where the physical data model of COG‑LO cations of the underlying components operational be‑
is deployed is the SIoT infrastructure. The social graph haviour to the COG‑LO platform‑speci ic model. The MSB
instantiates all virtual instances of the logistics objects, provides for the integration of infrastructure entities and
which are retrieved from the pilots’ data sources, manip‑ data sources, by means of data connectors.
ulates them and produces new data, i.e., their social rela‑ In addition, the MSB is the main COG‑LO system entity en‑
tionships. The properties of the virtual objects as well as forcing mechanisms for data security, privacy and trust,
their interconnections in the social graph are available in and enabling secureorchestration of COG‑LOcomponents
real time to the Cognitive Advisor and the CLOs of the net‑ as regards the execution of the necessary data lows, so
work. Various messages are exchanged (i) among CLOs, that the reference operational scenarios are eventually
(ii) between CLOs and the Cognitive Advisor and (iii) be‑ ful illed.
tween external event sources (e.g., ic Management The Message and Service Bus provides a set of interfaces
System) and the Message and Service Bus. The structure for the integration with COG‑LO services, components,
of the messages follows the COG‑LO data model. applications, as well as with external data sources and in‑
The implementation of the SIoT data storage relies on the frastructure entities:
Apache Ignite platform [24]. It is a memory‑centric dis‑
• Entity management: this interface is used for adding,
tributed database, caching and processing platform for
updating or deleting entities participating in the
transactional, analytical and streaming workloads deliv‑
COG‑LO system. It also provides a lookup method for
ering in‑memory speeds at petabyte scale. The inher‑
ent architectural design of Apache Ignite, which employs getting details of COG‑LO entities.
a distributed approach for both data storage and data • Messaging: this interface is used by COG‑LO services
caching, made it a natural solution for the implementa‑ and applications for data communication. It also al‑
tion of the SIoT platform, where several social Cognitive lows accessing information provided by various data
Logistics Objects are required to be stored and updated sources, either internal or external to the COG‑LO
in distributed fashion and simultaneously. system.
The SIoT data infrastructure is organized in a cluster of
SIoT peers. One SIoT peer manages the data related to • Data connector: this interface provides a uni ied so‑
the VIs of the CLOs it is responsible for. Cluster nodes lution for accessing information stored in heteroge‑
discover each other automatically enabling cluster scal‑ neous systems, under a common transactional inter‑
ing when necessary. The nodes are divided into two main face. It enables data interactions with the rest of the
categories: server and client. Server nodes are storage platform based on the COG‑LO common semantic in‑
and computational units of the cluster that hold both data formation model.
and indexes and process incoming requests along with Data sources connected to the COG‑LO platform (e.g., a
computations. The platform is based on a durable mem‑ traf ic management system or public train timetables ser‑
ory architecture that allows storing and processing data vice) are initially registered to the MSB and data ex‑
and indexes both in‑memory and on‑disk, ensuring per‑ changes are handled by the corresponing data connec‑
formance as well as durability. tors. Thelatterconsistofasetofdata lows, whichpublish
data to or ingest data from the platform, effectively hiding
3.3 Interoperability the implementation details of each data source.

The vastness of virtualized devices but above all the het‑
3.3.1 Data low management and orchestration
erogeneity of their physical counterparts, requires so‑
phisticated techniques to guarantee a high degree of in‑ AspartoftheMessageandServiceBus, thedata lowman‑
teroperability in terms of communication and interaction. agement and orchestration solution enables end users to
con igure the way COG‑LO components interoperate, in
order to react to logistics events. It offers a user‑friendly

74 © International Telecommunication Union, 2021