Formal Verification of a Hybrid Machine Learning-Based Fault Prediction Model in Internet of Things Applications

Alireza Souri,Mehdi Hosseinzadeh,Amin Salih Mohammed,Moayad Yousif Potrus,Mazhar Hussain Malik,Fatemeh Safara

doi:10.1109/access.2020.2967629

Abstract

By increasing the complexity of the Internet of Things (IoT) applications, fault prediction become an important challenge in interactions between human, and smart devices. Fault prediction is one of the key factors to achieve better arranging the IoT applications. Most of the current research studies evaluated the fault prediction methods using simulation environments. However, formal verification of the correctness of a fault prediction method has not been reported yet. This paper presents a behavioral modeling and formal verification of a hybrid machine learning-based fault prediction model with Multi-Layer Perceptron (MLP) and Particle Swarm Optimization (PSO) algorithms. In particular, the PSO is used for feature selection. Then, the fault prediction is considered as a behavior to be verified formally. The fault prediction behavior is divided into two types of behaviors: dimension reduction behavior and prediction behavior. For each of the behaviors, one formal model is designed. The behavioral models designed are mapped into the Labeled Transition System (LTS). The Process Analysis Toolkit (PAT) model checker is employed to evaluate the behavioral models. The accuracy of the fault prediction method is done by some existing specifications such as deadlock-free and reachability properties in terms of linear temporal logic formulas. Also, the verification of the fault prediction behaviors is used to detect the defect metrics of information-centric IoT applications. Experimental results showed that our proposed verification method has minimum verification time and memory usage for evaluating critical specification rules than other research studies.

Highlights

The quality of Internet of Things (IoT) applications [1] has grown rapidly throughout recent years, and issues related to that have gained more importance for software developers [2], [3]
According to fault prediction behavioral models, a set of states and events are as follow: RD_States=(Begin, Input_metrics, Normalization, Data_refi, PSO_refi, Pop_init, Initializing, Check_x, Comput_v, Calc_Pbest, Pbest_B_max, Calc_Mbest, Update_M, Mbest_L_max, Mbest_selec, Redu_Pop, Min_metrics); PR_States =; A path on the Reduced Dimensionality (RD) and PR behavioral models of the fault prediction mechanism is defined as follows: Definition 2: A directed path DP is a set of the restricted states and actions informs of transition relations with initial state si and final state sj that is shown an example as follows: DP
EXPERIMENTAL RESULTS This section shows the experimental results on the formal verification of the fault prediction behaviors that are performed by an Intel Core i5, 2.6 GHz, 8GB RAM, Windows 10 system and by Process Analysis Toolkit (PAT) model checker 3.4.1

Summary

INTRODUCTION

The quality of Internet of Things (IoT) applications [1] has grown rapidly throughout recent years, and issues related to that have gained more importance for software developers [2], [3]. A. Souri et al.: Formal Verification of a Hybrid Machine Learning-Based Fault Prediction Model in IoT Applications. Most of the papers on fault prediction examine their proposed method through simulation and experiments. Another way to verifying an information-centric IoT application is model checking [19]–[21]. To verify the proposed fault prediction method formally, logical problems are analyzed and behavioral specifications are checked. To propose a machine learning-based faults prediction model using the MLP and PSO algorithms in IoT applications.

RELATED WORK

DIMENSION REDUCTION AND FAULT PREDICTION

BEHAVIORAL MODELING OF THE FAULT PREDICTION APPROACH

MODEL CHECKING APPROACH FOR THE FAULT PREDICTION BEHAVIORS

EXPERIMENTAL RESULTS

CONCLUSION AND FUTURE WORKS