WIRELESS SENSORNETWORK

Corrosive damage remains a critical issue across various industries, especially in remote oil and gas pipeline infrastructures.This study presents the design and implementation of anIoT-based wireless sensor network (WSN) integrated with machine learning Model (SVM) for corrosion monitoring and prediction. The system architecture involved deploying sensor nodes utilizing electromagnetic techniques for real-time corrosion data acquisition. These nodes communicated with an ESP32 microcontroller equipped with wireless transmission capabilities to relay data to the Thing Speak cloud platform for storage and visualization. Subsequently, MATLAB was used to preprocess the acquired data, enabling the training and validation of a supervised machine learning model for corrosion classification and prediction. With the help of the SVM model, corroded pipeline samples could be easily dif erentiated from a corrosion-free pipeline. 80% of the recorded data was used to train the algorithm, and the rest 20% was kept for testing the data without corrosion. The first graph displayed by the model shows that the resistance values from the corroded sample fluctuate only slightly over time Additionally, the chlorine level ranged between (1000–1500)ppm, showing emission of chlorine gas from the sample. There was a significant drop in resistance in the corrosion- free sample for the second graph, with values falling below 1000ohms and No chlorine data was indicated When the model was tested and validated, the model correctly classified 59 out of 60 test samples whileone incorrectly indicating an accuracy of 98.33%.. When unseen samples were used, the model was still able to predict the presence of corrosion with almost the same amount of precision and gave results showing the state of the pipelines with a 50% chance of them being either corroded or not from a 40 sample prediction.. The results obtained af irm the ef ectiveness of both processes for corrosion monitoringinremote pipeline networks. The system’s autonomous operation, real-time data handling, and intelligent decision-making capabilities highlight its potential as a cost-ef ective and ef icient
alternative to traditional, labor-intensive methods. Moreover, its predictive capabilities enable proactive maintenance scheduling and safer operational planning, significantly reducing the risk of pipeline failure. This research thus lays a strong foundation for scalable, field-deployable corrosion monitoring systems leveraging modern IoT and AI tools

Wireless Sensor Networks (WSNs) play a crucial role in modern communication systems, particularly in environmental monitoring, industrial automation, and smart cities. However, amajor challenge in WSNs is optimizing energy consumption due to the limited power resourcesof sensor nodes. One of the most effective ways to enhance energy efficiency is throughdatacompression, which reduces the amount of transmitted data while preserving essential
information. This project explores the integration of machine learning-based data compression techniquestoimprove energy-efficient transmission in WSNs. A hybrid approach is proposed, combiningRun-Length Encoding (RLE) as a traditional lossless compression method with Principal
Component Analysis (PCA) as a machine learning algorithm to reduce data redundancywhilemaintaining accuracy. The study focuses on temperature sensor datasets collectedover aspecified period, ensuring real-world applicability. The methodology involves preprocessing raw temperature data, applying Run-Length Encoding(RLE) for initial redundancy reduction, and then leveraging PCA to extract principal components, further reducing data dimensions before transmission. The efficiency of the proposed model isevaluated based on key metrics such as compression ratio, reconstruction accuracy, andenergysavings. Performance comparisons are made with conventional lossless compression algorithmslike Huffman Coding and Arithmetic Coding to assess improvements. Preliminary results indicate that the combined approach achieves a higher compressionratiowhile preserving critical temperature variations, leading to significant energy savings in wirelesstransmissions. This work contributes to advancing energy-efficient data handling inWSNs, making it highly relevant for resource-constrained environments. Future research directionsinclude expanding the model to handle multi-sensor data streams and implementing real-timeadaptive compression strategies.