Theses and Dissertations
Permanent URI for this collectionhttp://ir.daiict.ac.in/handle/123456789/1
Browse
4 results
Search Results
Item Open Access Malicious Node Detection for various Heterogenous IoT Communication Protocols(2021) Mekala, Priyanka; Goel, SupriyaIn recent years security is an increased concern for IoT devices. Due to limited capabilities compared to traditional computer systems, these tiny devices cannot run the heavy encryption algorithms required for preventing attacks. Nowadays, IoT comprises several communication protocols like Bluetooth Low energy, WiFi, and Zigbee for different applications including home automation, smart city etc. With such a heterogeneous system, it becomes complex to provide security as with every different protocol comes more vulnerabilities in the network. Anomaly-based detection methods have received increasing interest from the scientific community in the last few years. It acts as a second layer to the system’s security. With deep packet inspection, it evaluates the network traffic and forms a set of informative features formalizing the normal and anomalous behaviour of the system. We classify among a normal or abnormal activity using machine learning algorithms and present the results of our detection system implemented on a heterogeneous IoT testbed. This system is applicable for companies, offices, government organization or secret agencies who want to increase their network security to protect their systems.Item Open Access Investigation into a low cost low energy IoT enabled wireless sensor node for particulate matter prediction for environmental applications(Dhirubhai Ambani Institute of Information and Communication Technology, 2019) Shah, Jalpa Bharatkumar; Mishra, BiswajitIn recent years, increased transportation, removal of trees for making buildings, establishment of new industries, are the main sources of increased pollution. Increased pollution is one of the major challenge faced by all countries as it a ects environment and human health. On of the way to deal with this challenge is monitoring the environment quality and taking corrective steps for the same. The conventional instruments used for environment monitoring are accurate but costly, time consuming, requires human intervention and lacking in terms of portability. Internet of Things (IoT) enabled wireless sensor node is one of the ideal solutions for real time monitoring of environment in today's urban ecosystems. We have developed a low power IoT enabled wireless sensing and monitoring platform for simultaneous monitoring, real time data of ten di erent environmental parameters such as Temperature, Relative Humidity, Light, Barometric Pressure, Altitude, Carbon dioxide (CO2), Volatile Organic Compounds (VOCs), Carbon Monoxide (CO), Nitrogen Dioxide (NO2) and Ammonia (NH3). We have tried to achieve low power through modi cation in sensor node hardware architecture and developing prediction model which eliminates the need of power hungry sensor. The proposed hardware architecture for wireless sensor node helps in reducing power and number of interfacing pins required from the microcontroller. The proposed wireless sensor node architecture is also adaptable for any other applications after replacement or removal of sensors and/or modi cation of supply. The developed system consists of the transmitter node and the receiver node. The data received at the receiver node is monitored and recorded in an excel sheet in a personal computer (PC) through a Graphical User Interface (GUI), made in LabVIEW. An Android application has also been developed through which data is transferred from LabVIEW to a smartphone and enables IoT. The system is validated through experiments and deployment for real time monitoring. For the proposed system reliability of transmission achieved is 97.4%. Power consumption of the sensor node is quanti ed which is equal to 25.67mW and can be varied by varying the sleep time or sampling time of the node. Battery life of approximately 31 months can be achieved for the measurement cycle of 60 secs. PM2.5 is one of the important pollutants for measuring air quality. Existing methods and instruments used for the measurement of PM2.5 are more laborious, not applicable for both online and o ine, having response time from a few minutes to hours and lacking in terms of portability. In this work we present the correlation study of PM2.5 with other pollutants based on the data received by Central Pollution Control Board (CPCB) online station at N 23 0' 6.6287, E 72 35' 48.7816. Based on the correlation results, CO, NO2, SO2 and VOC parameters (Benzene, Toluene, Ethyl Benzene, M+P Xylene, O-Xylene) are selected as predictors for developing PM2.5 prediction model. PM2.5 prediction model is developed using Arti cial Neural Network (ANN), resulting in a simple analytical equation. Since the proposed model is expressed in simple mathematical equation, it can be deployed on a wireless sensor node enabling online monitoring of PM2.5. Closeness of predicted and actual values of PM2.5 are veri ed through processing derived model equations using low cost processing tool (e.g. excel sheet), thereby eliminating the need for proprietary tools. The RMSE and regression coe cient of the derived model is 1.7973µg/m3 and 0.9986 respectively. Predicted and actual values of PM2.5 are found very close to each other and variation is in the acceptable range. Derived model is recalibrated in terms of predictors and coe cients to test it, in a di erent city, using data of developed low power wireless sensor node. Based on the availability of the sensors on wireless sensor node, recalibration is done for the reduction of predictors to three; CO, NO2 and VOC. For recalibrated model, results show RMSE of 7.5372 µg/m3 and R2 0.9708. The obtained results show the feasibility and e ectiveness of the proposed approach. Improvement in these results is possible by recalibrating prediction model based on data from multiple stations at the place of deployment. Predicted model can be used for online or o ine measurement. Time involved in the measurement is less compared to conventional methods, which is equal to the processing time of the equations. To provide accurate results proposed wireless sensor node is calibrated against the standard calibrated instruments. The proposed system has advantages over conventional methods such as less costly, automated, portable, less time consuming and having higher temporal and spatial resolution.Item Open Access WSN Network Analysis and Prediction using ML(2020) Patel, Charmy Bharatbhai; Sasidhar, P S KalyanWireless sensor networks are a group of sensors that monitor and record the physical changes of the environment that change rapidly over time. This ability of WSNs help in various fields ranging from the engineering industry to immediate home environments. A sensor node is capable of performing some processing, gathering sensory information, and communicating with other connected nodes in the Network. To make some decisions in this Network, sensors adopt machine learning algorithms. The main aim of this project is to find out the parameters which can increase throughput and decrease the Delay in our Network. Different network sizes have been taken into consideration to find the parameter changes required to meet our above objective. This research project mainly includes the data collection phase observing a network and learning phase. It is simulated for different network scenarios. Three types of machine learning algorithms have been applied: Linear Regression, Neural Network, and Random Forest. By applying these algorithms, we get to know that RandomForest overfits the model, whereas Neural-Network underfits the model because they are non-linear algorithms. Hence we can say that it is showing linear behavior as non-linear algorithms like Neural-Network and RandomForest didn’t help us to estimate the throughput and delay in our Network, and hence they are not suitable.Item Open Access Log based method for faster IoT queries(Dhirubhai Ambani Institute of Information and Communication Technology, 2017) Jain, Anubha; Bhise, MinalWith increase in the size of Internet of Things IoT device networks and applications, tremendous increase is witnessed in the size of IoT data. To build smart applications using IoT data, its efficient storage is important to facilitate faster queries. IoT data is represented in Resource Description Framework RDF and stored in relational format. This thesis addresses the issue of faster query processing of this data. It resents a Log Based Method LBM to partition IoT data. IoT systems exhibit skewedness in data access patterns as some records are accessed more frequently than the other. LBM exploits this skewedness in access patterns of records. It incorporates Forward Algorithm FA and Backward Algorithm BA to analyse the query workload and partition the basic triple table into hot and cold data tables. For our experiment, 8% of hot data table is found to solve 78.6% queries. The query execution is found to be 67.5% faster on partitioned data than triple table. To further accelerate FA and BA, we have executed them in parallel as well which is found to be 42% faster than its serial execution.