Fuzzy-Logic-Based Decision Engine for Offloading IoT Application Using Fog Computing

Introduction

The proliferations of smartphones are increasing dramatically during recent years. Billions of users are using smartphones and various applications. A large number of applications and sensors are coming along with mobile phones such as real time gaming applications, powerful GPS sensors, NFC readers, etc. Hence it is easy to create an IOT application with the sensors which are available in the mobile devices. Majority of these applications are computationally intensive and processing intensive which will drain the battery faster (Buyya et al., 2009). More over currently most of the smart phones are equipped with processors and batteries which are not capable of doing all these works. Hence the critical problem that smart phones are facing is the ability to carry out the above mentioned applications with the existing battery and processing capabilities.

A large number of hardware technologies such as maintaining leakage power, parallel execution, runtime voltage scaling is existing nowadays. Recently cloud computing came into popularity where all the applications are migrating to cloud. The cloud computing with mobile technologies such as ubiquitous computing devices, location based devices, lead to a novel computing technology called Mobile Cloud Computing (MCC). This provides the mobile users with all advantages of cloud including infinite storage space and unlimited computing power. But the main disadvantage of cloud computing is the connection latency. So we need a low latency connection to the server which can save both energy and time. The extension of cloud computing called Fog computing can be utilized for this. Hence for energy and time efficiency and total performance improvement of the application offloading to the Fog servers are the best suitable solution. Therefore, there is strong need for research into smart system based on the concept of offloading to Fog edge servers.

Offloading is the process of transferring the calculations to the cloud in order to achieve energy savings and improved efficiency. The shortcomings of mobile devices can be overcome by the use of offloading. Another question that arises is “Whether offloading is beneficial always?” The answer is “No”. Only if enough energy savings and speed up can be achieved by offloading, applications can be offloaded, otherwise it can be executed in the local mobile itself. So a decision has to be taken for offloading. If the situations are favorable for offloading where it will lead to significant performance improvement a decision can be taken to offload the application. Otherwise the decision should be not to offload but to use a local execution. This decision has to be taken logically by looking at the current situation and take an offloading decision. For such decision making Fuzzy logic decision provides an efficient mechanism. The parameters that are going to be considered in this decision can be formulated as fuzzy parameters.

In this chapter, the offloading is considered as a decision problem. This depends on various features on mobile side and the cloud side. Here a fuzzy logic based approach is used to take a decision on whether to execute locally on the mobile itself or remotely in the cloud. After the decision the application is partitioned into local and remote fragments and the remote part is executed in the cloud server. Therefore, the main contributions of this chapter are:

• 1.

  Develop Mobile Offloading architecture.

• 2.

  A Decision Algorithm Using Fuzzy Logic: The offloading decision is taken with the help of a fuzzy logic based system which will take into consideration of both mobile and cloud side parameters. Based on these parameters a decision is taken and is conveyed to the mobile for further processing.

• 3.

  Performance Analysis: The performance of the system is analyzed based on the error rate in the decision and time taken for the decision. The execution time for local and remote execution is compared for the applications. Even though the execution time in cloud is lesser, it shows much fluctuation compared to the local execution.

This chapter has been divided into five sections: Section 1 introduces the concept and our approach; section 2 provides background; section 3 discusses the proposed software architecture based on Fuzzification and algorithm; section 4 discusses the evaluation of our approach and section 5 concludes the work.