Delay Mitigation in Offloaded Cloud Controllers in Industrial IoT
Abstract:
This paper investigates the interplay of cloud computing, fog computing and Internet of Things (IoT) in control applications targeting the automation industry. In this context, a prototype is developed to explore the use of IoT devices that communicate with a cloud-based controller, i.e., the controller is offloaded to cloud or fog. Several experiments are performed to investigate the consequences of having a cloud server between the end device and the controller. The experiments are performed while considering arbitrary jitter and delays, i.e., they can be smaller, equal or greater than the sampling period. The paper also applies mitigation mechanisms to deal with the delays and jitter that are caused by the networks when the controller is offloaded to the fog or cloud.
Existing system:
A typical closed control loop is shown in Fig. 2. It consists of a controller that controls a physical process through sensors and actuators. The controller is usually located close to the actual process. Fig. 3 depicts the closed loop control when the controller is offloaded to the network, e.g., to a fog node or a cloud. In this case there are network delays in the control loop that should be accounted in the response times of the systems. These delays can have serious effects on the systems with real-time requirements such as making them unpredictable and unstable. A detailed investigation is needed to show the feasibility of offloaded controllers to the fog versus cloud in the delay-sensitive industrial automation systems. The first tier of the fog architecture, shown in Fig. 1, can be used for closed loop control between the industrial machines and the fog nodes. The second and third tiers can be used to process data to monitor the machines and prevent future hardware failures. While the latter has been investigated in the existing works, this paper focuses on the closed loops with offloaded controllers to the fog as well as to the cloud. The devices in the context of IoT are not very resourceful to support computation-intensive tasks. Using cloud computing, control can be provided as a service allowing the execution of computation-intensive algorithms in the cloud. However, according to [10], cloud computing cannot be used to control the automation industrial machines, mainly, due to unpredictable network delays. In order to overcome the limitation of cloud computing, this paper advocates the effectiveness of local cloud computing infrastructure, especially fog computing for the delay-sensitive industrial automation systems.
Proposed System:
In order to address the challenges discussed in the previous subsection, we develop a control system application prototype by exploiting the principles of IoT, cloud computing and fog computing. Using the prototype, we perform a number of experiments to investigate the impact of local and wide area networked controller on the closed loop control. In order to do the performance evaluation, we consider arbitrary jitter and delays, i.e., they can be smaller, equal or greater than the sampling periods. Additionally, we apply two delay mitigation mechanisms for the end device. These mechanisms do not use any internal information from the controller, in fact these mechanisms rely only on the received data. It should be noted that the goal of this paper is not to invent new techniques for interplay of IoT, cloud computing and fog computing in the industrial automation, but to investigate and show the feasibility of existing techniques in this area. The contributions in this paper include a comparative evaluation of various scenarios in the above-mentioned context including (1) local controller, (2) controller offloaded to the cloud and (3) controller offloaded to the fog. The contributions also include the application of delay mitigation techniques, i.e., prediction methods have been implemented in the end devices, whereas the adaptive PI algorithm is implemented in the cloud-based controller. The techniques, prototype and experiments that are presented in this paper are applied in the industrial settings within the domain of the automation control systems, provided by one of our industrial partners. The proof of concept provided in this paper serves as the foundation for the advanced cloud-based controllers which will be eventually used in the automation industry, e.g., robotic arm and collaborative machines.
Conclusion:
In this paper we have investigated the effects of offloading the controller to the fog and cloud. We have built a prototype to create a realistic scenario. Moving the controller to a remote server degrades system’s performance. Even small delays affect the control loop when they exceed the sampling period. The forecasting methods that we have investigated, work better with variable delays because the predicted values that are used are more scattered, as opposed to constant values where the predictions are concentrated at the beginning. However, these methods are challenged in the case when consecutive predicted values are needed, especially in the start up of the control system. Whereas, the adaptive PI controller uses an intuitive idea for mitigating delays inside the network. A formal proof for the robustness of such approaches is needed. Since not much research has been conducted in this area, there are some possibilities left to try out in the future. The polling and non-polling approaches can be expanded into a more sophisticated system. Moreover, loads can be introduced on both ends to simulate a more complex approach. These methods can be tested with a more complicated system. Another interesting future work is to develop a smart resource manager that is capable of making offloading decisions at runtime. Such a manager can be realized at various levels in the IoT infrastructure such as public cloud, private cloud and fog.
References:
[1] B. Sosinsky, Cloud Computing Bible, ISBN 978-0-470- 90356-8, Wiley Publishing, Inc., 2011.
[2] K. Ashton, “That ’Internet of Things’ Thing, in the real world things matter more than ideas,” http://www.rfidjournal.com/articles/view?4986, June 2009, [Online; Accessed 06-February-2015].
[3] IEEE Internet of Things, http://iot.ieee.org/about.html, accessed: January 2017.
[4] F. Bonomi, “Keynote talk: Connected vehicles, the internet of things, and fog computing,” in 8th ACM International Workshop on VehiculAr Inter-NETworking (VANET), 2011.
[5] F. Bonomi, R. Milito, J. Zhu, and S. Addepalli, “Fog computing and its role in the internet of things,” pp. 13–16, 2012.
[6] H. Madsen, G. Albeanu, B. Burtschy, and F. L. Popentiu-Vladicescu, “Reliability in the utility computing era: Towards reliable fog computing,” in 20th International Conference on Systems, Signals and Image Processing (IWSSIP), 2013, pp. 43–46.
[7] I. Stojmenovic, “Fog computing: A cloud to the ground support for smart things and machine-to-machine networks,” in Australasian Telecommunication Networks and Applications Conference (ATNAC), 2014, pp. 117–122.
[8] R. Langmann and L. Meyer, “Automation services from the cloud,” in 11th IEEE International Conference on Remote Engineering and Virtual Instrumentation (REV), 2014, pp. 256–261.
[9] O. Givehchi and J. Jasperneite, “Industrial automation services as part of the Cloud: First experiences,” Proceedings of the Jahreskolloquium Kommunikation in der Automation–KommA, Magdeburg, 2013.
[10] H. P. Breivold and K. Sandstr¨om, “Internet of Things for Industrial Automation Challenges and Technical Solutions,” in 8th IEEE International Conference on Internet of Things (iThings 2015), Dec. 2015.