Page 83 - ITU Journal Future and evolving technologies Volume 2 (2021), Issue 5 – Internet of Everything
P. 83
ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 5
representation of the physical entity that implements the Friends information is hosted. The SIoT hosts only one
functionality required for managing the communications Friend Table for each CLO. The FT content represents the
and for supporting the common control plane [17]. A CLO CLO friends and contains information about the friend‑
carries metadata associated with a given object, such as ships. The Relationship Manager (RM) module imple‑
information on object’s nature, status and list of friends. ments the logic through which a relationship can be cre‑
Among the information characterizing the CLO, seman‑ ated, deleted or updated. It is responsible for providing
tic descriptions facilitate the interactions between digi‑ rules for implementing the social relationships among ID‑
tal twins, supporting the detection and management op‑ labelled CLOs. When a new relationship is created, the RM
erations despite devices heterogeneity. A CLO, as a soft‑ writes directly in the Friend Table distributed database
ware entity, can be implemented in SIoT repositories and of the involved CLOs. The Relationship Browser (RB)
hosted in the cloud or on the edge of the network infras‑ module deals with navigating the social graph while im‑
tructure. SIoT is responsible for managing the life cycle plementing the search algorithms. In particular, the RB
of each digital twin, store and update their status infor‑ deals with the analysis of the FTs to ind a target destina‑
mation in real time and disseminating data from the dig‑ tion/set of destinations to be reached among the friends
ital twin to the physical world through a new data com‑ of a given entity. The Identity Service (IS) module is re‑
munication delivery method and scheme called Sociocast sponsible for managing the Identi iers lifecycle. IS man‑
[18]. In particular, Sociocast leverages the SIoT concept ages the IDs of the created or removed CLOs. Subscrip‑
to support group communications among nodes in an ef‑ tion Service (SS) allows to the peer or cluster RM to sub‑
icient and effective way on the control plane. With data scribe to information about CLO changes (for example po‑
driven logistics, dynamic optimization of basic logistics sition) and provides support for pushing the relevant up‑
processes is at the forefront of the next generation of lo‑ dates. SS is exploited whenever the friendship tables of
gistics services. Finding optimal routes for vehicles is a the CLOR managed by two different RMs need to be up‑
problem that has been studied for many decades from dated. The Virtual Instance (VI) mdule represents the
a theoretical and practical point of view. What is typi‑ virtual instance of CLO: when a CLO has to interact, make
cally associated with the Vehicle Routing Problem (VRP) decisions or perform computation, this component is in‑
is a generalization of the Travelling Salesperson Problem, stantiated. It is released when the CLO gets idle. It repre‑
where multiple vehicles are available. This class of rout‑ sents the digital twin of the CLO, extends its capabilities
ing problems is notoriously hard; it not only falls into and carries out social behaviours.
the class of NP‑complete problems, but also it cannot be
solved optimally in practice, even for moderate instance
sizes. More importantly, processing VRP optimization
on large graphs in real‑time demands employing addi‑
tional techniques, such as heuristics and/or graph prun‑
ing. Different clustering approaches have been used in
pruning the input graph for VRP. For example, Ruhan et
al. [19] uses k‑means clustering in combination with a re‑
balancing algorithm to obtain areas with balanced num‑
bers of customers. Bent et al. also study the bene its
and limitations of vehicle and customer‑based decompo‑
sition schemes [20], demonstrating better performance
with the latter. In COG‑LO, linear optimization was ap‑
plied as an exact optimization approach for solving VRP.
The combination of an enriched social‑like behavior and
instant‑by‑instant knowledge of an object’s state, allows Fig. 1 – SIoT peer
SIoT to support requests from optimization systems by
effectively pruning CLO graphs on the basis of social re‑
lationships. Therefore, it proves to be a critical part for 3. COG‑LO FRAMEWORK
real‑time optimization.
The COG‑LO framework aims to provide a holistic solution
2.3 SIoT platform for handling the high operational dynamicity of the logis‑
tics environment. It follows a layered architecture and
Conceptually, the Social Internet of Things (SIoT) plat‑ provides interoperable solutions regarding data models,
form consists of various clusters of SIoT peers. Each SIoT information exchange and security mechanisms. This
peer is made up of different functional blocks and exposes way it enables transparent coordination and exchange of
its data and services via REST calls (Fig. 1). information between different objects and systems based
The Cognitive Logistic Objects Repository (CLOR) is a on heterogeneous access protocols in a secure manner.
data structure that contains all CLO information. Friend
Tables (FTs) are the data structure where the CLO
© International Telecommunication Union, 2021 71