IEEE Systems, Man, and Cybernetics Magazine - January 2018 - 12

of vast human resources outside of the organizations. In
such cases, open product refers to a class of products that
are developed by the crowd in an open way. The trend is
also promoted by the boom of social networks, such as
Facebook and WeChat, which can provide good communication and sharing spaces. With the emergence of virtual
communities of like-minded people, playing different roles
(customers, manufacturers, professionals, and more)
offers excellent chances to solicit contributions/collaborations from different individuals/organizations with various competences and thoughts. The challenge is to choose
the scope and scale of dynamic cooperation and control
the quality of contributions. If too many people are involved, not only may high-quality contributions be overwhelmed by massive trivial and unimportant ones, but
also the cost may increase significantly or not enough contributions can be acquired. Thus, new metrics, methods,
and online supporting tools (possibly domain specific) to
address such challenges are necessary.
A typical case is customized or personalized products, which can best meet individual customers' needs.
More companies are heading toward providing customized or personalized products to survive in the fragmented, diversified, and competitive marketplace. The challenges
lie in defining the functions of a simple graphical-aided
software tool for consumers and third parties and the
just-in-time production of modules by different parties
and the efficient assembling of modules. Quality control
(including safety, reliability, and performance, among
others) and warranties are also important issues for
customized or personalized products [38]. Other challenges come from aspects of production variability and
financial viability [39]. We recently proposed a framework to support design and production of customized/
personalized products under IoT-enabled manufacturing
clouds [40].
Open product also means that a product can work and
collaborate with other (new) devices or software on an
unanticipatedly wide scale. This requires commonly
accepted standards and platforms to enable interoperability. In the ecosystem of IoT-enabled workshops, renovated or new facilities need to efficiently cooperate with
current machines to automate the production. However,
unexpected collaborations can pose great challenges. Mostly, a product is designed to work in specific contexts (i.e.,
has its own assumptions and control strategy without much
knowledge of other products/systems [7]); thus, it usually
cannot deal with such openness when renovated or new
products are involved. Some of those products can even
mutually interfere when functioning. This demands, in part,
that the products are intelligent and autonomous (i.e., an
intelligent product that contains sensing, memory, data processing, reasoning, and communication at various intelligence levels) [28], such as intelligent agents [41], [52]. Open
will also cause grand challenges in the dimensions of security and privacy, as we will briefly discuss later.
12

IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE Janu ar y 20 18

Services Provision and Composition
XaaS is now prevalent in the cloud. It is important to integrate various manufacturing resources and capabilities as
cloud services and improve interoperability between services and efficiency of service collaboration during a stage
or across multiple stages of the whole product life cycle,
especially for users who need multiple services to fulfill an
individual complex task [42].
Another perspective is to leverage abundant services
from multiple industrial clouds and address the uncertainty issue under today's highly dynamic business environments. We have proposed a hybrid framework for
integrating multiple manufacturing clouds [43] in which
clouds can form federations to use their aggregated
resources and users can have a wider selection of services. Due to the relatively long execution time of manufacturing services, various disruptions can occur and cause
a deviation from the target. Thus, the dynamic adjustment
of service execution plans is needed to guarantee optimal
performance. In this process, the IoT can capture and
report the critical events in a real-time manner and thus
make the control of service execution a closed loop. We
recently developed a framework [44] that uses the IoT's
real-time sensing ability on service execution, big data's
knowledge extraction ability on services, and event-driven dynamic service-selection optimization to deal with
disturbances and continuously adjust the service selection to be more effective and efficient. Proper formulation
of the dynamic service selection for varying uncertainties
should be built [45].
User-Centric Pervasive Environment
The IoT has been developed to respond in an intelligent way to
the presence of users, thereby providing better support to
them in carrying out specific tasks. In manufacturing, this
means IoT objects/systems/environments should have the
ability to automatically perceive user needs through context
awareness so that users can quickly acquire the needed services and focus on their tasks. Compared to closed environments in ambient intelligence, the IoT needs to deal with open
scenarios, whereby new functions/capabilities should be
accommodated at runtime and may not be considered at the
design time [4]. The IoT systems that involve humans also
exacerbate this challenge, as human behaviors are driven by a
huge range of factors and tend to be much more complex and
volatile. This further requires IoT systems to be truly autonomous and intelligent and equipped with a self-learning ability to handle new scenarios properly.
After the perception of user needs, it is necessary to
present available services and big data analytics in an easily
understood and user-friendly method. Visualization can be
of great help; however, it is not easy to visualize unstructured data in a flexible way. Furthermore, the visualization
system should be interactive so that users can choose what
they want to see and use. To support this, intelligent
machines need to autonomously interact with each other

Table of Contents for the Digital Edition of IEEE Systems, Man, and Cybernetics Magazine - January 2018