Reliable routing of neural network based characterized data in wireless body sensor network

Abstract

An enormous quantity of data like Electrocardiogram (ECG) as well as newlineElectroencephalogram (EEG) signals are gathered as well as chronicled from human newlinebody for appraising and monitoring the enactment of human physiological newlinecondition. The appropriate examination of these signals is a momentous issue in newlinemedical diagnosis as well as for the treatment. Even though the EEG and ECG signals newlinefurnish valuable information concerned with the brain and heart, no technique is newlineformulated for the automatic classification of these signals. Conventionally the newlineexamination of these signals is performed manually by expert medical practitioners newlinewho are less in number. Hence, there is the necessity for developing automatic newlineclassification technique that will do appraisal of these signals. This thesis represents newlinea method for the classification of ECG and EEG signals and the reliable routing newlinealgorithm. newlineElectrocardiogram (ECG) plays a vital part in the medical field. By means of newlineelectrocardiograph device heart s electrical signals are chronicled in cardiology. newlineUtilization of ECG can be done for determining various heart condition. The newlinetechnique used in this work integrates the study of the ECG signals and their newlineclassification. Here, the ECG signal is categorized among five distinct categories i.e. newlineNSR, APB, AFIB, LBBB, RBBB as well as PVC. The other kind of important newlineinformation obtained from patient is Electroencephalogram (EEG). The person s newlinemental functionality can be measured with the help of EEG. It can be employed to newlinedetermine the various disorders related with brain functionality. The proposed newlinework concentrates on two types of disorders i.e. Seizures and Schizophrenia. newlineThe information about the patients is obtained through the sensors attached to the newlinehuman body which is a form of WSN known as WBSN usually used in the medical newlineapplications. For example WBSN can be utilized for uninterrupted monitoring the newlinephysiological condition of the patient remotely. The main purpose of the system is newlinecharacterize the ECG as well EEG signal efficiently and then transfer the data to the newlinedestination reliably. To achieve this aim the various steps used are, preprocessing of newlineI newlinethe raw data with the purpose of removal of noise form raw data and make it ready newlinefor the process of feature extraction. It is the second phase in which various features newlinehave been extracted. From the extracted features few features will be selected in newlinefeatures selection phase. For selecting features I-BAT algorithms is utilized. The newlineobtained results show that the performance of I-BAT selection technique is better as newlinecompared to BAT as well GA. In the next step classification is performed using newlineArtificial Neural Network (ANN). The performance of ANN is compared with other newlinethree distinct techniques: k-nearest neighbor algorithm (k-NN), Naïve Bayes, newlineAdaptive neuro-fuzzy inference system (ANFIS) which have been employed to newlineperform the classification. The obtained results indicate that the performance of newlineANN is batter compared to other three techniques. The proposed systems overall newlineclassification accuracy is 96.68%. newlineIn the next phase of the work reliable transmission of health info is done. To newlineperform routing of the info various steps are considered such as encoding of the newlinedata, encryption and next step is choosing the appropriated path to the desired newlinedestination. Modified Huffman Encoding is used for the encoding the information. newlineNext procedure implemented is encryption to provide the security the HECC newlinetechnique is used for this purpose. Here three distinct techniques: HECC, ECC and newlineRSA have been evaluated to determine the performance of encryption and the newlineobtained results show that the HECC performance better compared to other two newlinetechniques. Once the data is secured, transfer of the info is done by employing newlineParameter Based Reliable Routing algorithm. In this the nodes on route are selected newlinedepending on the distance as well as residual energy of a node. The obtained results newlineshows that the performance of PBRR increases the effectiveness of the system as it newlinetransfers the information reliably to the destination. newlineKeywords: Electroencephalogram (EEG), Electrocardiogram (ECG), Artificial Neural newlineNetwork (ANN), Parameter Based Reliable Routing Protocol (PBRRP) newline