Hierarchical Fault Detection and Recovery Framework For Self-Healing WSN

Abstract

Wireless Sensor Network (WSN) contains several sensor nodules that are linked to each other wirelessly. Errors in WSN may perhaps be because of several causes which bring about hardware damage, power thwarts, incorrect sensor impression, faulty communication, sensor deficiencies, etc. This damages the network process. In this paper, we propose to develop a Hierarchical Fault Detection and Recovery Framework (HDFR) for Self-Healing WSN. This framework consists of three modules: Fault detection, fault confirmation and fault recovery. In fault detection module, Particle Swarm Optimization (PSO) algorithm is applied for estimating the discrete Round Trip Paths (RTPs). Along the established RTPs, round trip delay (RTD) time values are estimated. Then based on the RTD, the suspected nodes are identified. In fault confirmation module, the nodes are confirmed to be either in FAULTY or ACTIVE state. In fault recovery module, the primary controller (PC) will establish an alternate route via the secondary controllers (SC) by excluding the faulty nodes. Then, it will resend the stored packets to the sink via the newly established route. By experimental results, it is shown that the HDFR framework achieves better detection accuracy and packet delivery ratio.

Prediction of Springback Behavior of Vee Bending Process of AA5052 Aluminum Alloy Sheets Using Machine Learning

“…The proposed kinematic hardening model was obtained by coupling the Chaboche kinematic hardening model (Chaboche 1986) with the condition function, which replaces the material constants of the kinematic hardening model using data of AA5182-O, 719B, and 780R AHSS materials.The coupling enabled the accounting of mechanical properties in different rolling directions [12].Chen et al [13]demonstrated that FEA simulation models using ABAQUS with 6016-T4 aluminum alloy in Vee bending according to a material constitutive model with varying elastic modules results in better SB prediction than models with a constant elastic modulus. However, these models showed limitations in reliable prediction of radius and included bend angle compared with the other models [13][14][15].Several computing tools,such as genetic algorithms, fuzzy logic, and artificial neural networks (ANN), are used in metal forming in the fields of optimization, design, and prediction.However,these computing tools must be aided by sufficient reliable data about the forming process, proper tool selection, and appropriate utilization of computational models [16,17].…”