To better illustrate some of the complexities involved with
openness, consider feedback control. Many sensor and actuator
systems heavily utilize feedback control theory to provide robust
performance. The classical methodology includes creating a
model of the system and then deriving a controller using wellknown
techniques to meet stability, overshoot, settling time, and
accuracy requirements.Asensitivity analysis is also possible and
strongly encouraged. However, openness and scale create many
difficulties for this methodology. The openness means that the
model of the system is constantly changing. Human interaction is
an integral aspect of openness (see Section III-H) and this makes
modeling extremely difficult, and the scaling and interactions
across systems also dynamically change the models and creates a
need for decentralized control. While some work has been
performed in topics such as stochastic control, robust control,
distributed control, and adaptive control, these areas are not
developed well enough to support the degree of openness and
dynamics expected in some IoT systems. A new and richer set of
techniques and theory is required. It is especially important to
understand how large numbers of control loops might interact
with each other. To date, there have already been examples where
control loops have competed with each other, one indicating an
increase in a control variable, while the other loop indicating a
decrease in the same variable at the same time. Such dependencies
(see Section III-B.) must be addressed in real-time and in an
adaptive manner to support the expected openness of IoT.
Openness is also playing a major role in industrial things on the
Internet. Remote access across factories or to individual products
is often very beneficial to Industry. However, security concerns
arise, especially if there is any safety issue involved.
To better illustrate some of the complexities involved with
openness, consider feedback control. Many sensor and actuator
systems heavily utilize feedback control theory to provide robust
performance. The classical methodology includes creating a
model of the system and then deriving a controller using wellknown
techniques to meet stability, overshoot, settling time, and
accuracy requirements.Asensitivity analysis is also possible and
strongly encouraged. However, openness and scale create many
difficulties for this methodology. The openness means that the
model of the system is constantly changing. Human interaction is
an integral aspect of openness (see Section III-H) and this makes
modeling extremely difficult, and the scaling and interactions
across systems also dynamically change the models and creates a
need for decentralized control. While some work has been
performed in topics such as stochastic control, robust control,
distributed control, and adaptive control, these areas are not
developed well enough to support the degree of openness and
dynamics expected in some IoT systems. A new and richer set of
techniques and theory is required. It is especially important to
understand how large numbers of control loops might interact
with each other. To date, there have already been examples where
control loops have competed with each other, one indicating an
increase in a control variable, while the other loop indicating a
decrease in the same variable at the same time. Such dependencies
(see Section III-B.) must be addressed in real-time and in an
adaptive manner to support the expected openness of IoT.
Openness is also playing a major role in industrial things on the
Internet. Remote access across factories or to individual products
is often very beneficial to Industry. However, security concerns
arise, especially if there is any safety issue involved.
การแปล กรุณารอสักครู่..