Soft Computing To Sensor Network Reliability, Systems And Their Fpga Implementation

Access Type

Open Access Dissertation

Date of Award


Degree Type


Degree Name



Electrical and Computer Engineering

First Advisor

Harpreet Singh


Soft Computing(SC) has emerged as an effective candidate to deal with complex problems like unmanned Ground Vehicles Reliability(UGVR), Crack Detection and Impact Source Identification (CDISI) where there is lack of precision, certainty and complete truth. This dissertation describes some novel applications of SC in area of sensor networks (SN) and systems with their FPGA implementation. With an increased importance of security issues, it has become critical to determine reliability of SN. As number of sensor nodes is fairly large in SN, it's rather impractical to adopt traditional reliability evaluation methods. The minimum number of communication hops utilizing least number of links rather than all possible hops plays a significant role in network security and reliability. The Minimum Hops Path(MinHP) takes minimum number of links to communicate between the source and the sink node. The Minimum Hops Cutsets(MinHC) is the cutset with minimum number of links in each of the cutset terms. An efficient but approximate Minimum Hop Terminal Reliability scheme that utilizes MinHP and MinHC is proposed. The proposed approximate Min-Max system reliability algorithm for SN outperforms traditional algorithms. These algorithms are shown to provide reasonably accurate results with significant reduction in number of computations. The SN has wide range of military and commercial applications, with UGVR analysis, CDISI as some of the SN and system applications considered in this thesis. The convoy of unmanned vehicles is portrayed as a network, with stations as nodes and links as communication paths between them. The UGVR is evaluated using graph theoretic approaches like spanning tree and BDD, supported by Fuzzy and Neuro-Fuzzy techniques for predicting node and link reliability. The UGVR is simulated with some existing data. Further this network is expanded with each node being network in itself. The FPGA implementation of some standard networks reliability is done. The Fuzzy and NeuroFuzzy CDISI system proposed is implemented on FPGA with objective to fit it in a handheld device. The validation of proposed models was done using Xilinx's Spartan-3 FPGA, ModelSim-XE and SynaptiCAD-VeriLoggerPRO. It is hoped that proposed techniques will go a long way in finding applications in SN areas.