Design and evaluation of an energy-efficient IoT sensor data compression scheme

Thesis Abstract

The proliferation of Internet of Things (IoT) devices in diverse application domains has underscored the critical need for efficient data transmission and energy management to ensure sustainable operation and prolong device lifespan. Sensor nodes in IoT networks generate vast volumes of data, often characterized by redundancy and irrelevancy, leading to increased energy consumption during data transmission and processing. This study aims to design and empirically evaluate an energy-efficient data compression scheme tailored for IoT sensor networks, with the primary objective of reducing energy expenditure while maintaining data integrity and fidelity. To achieve this, the research employs a mixed-methods approach, integrating algorithm development with experimental evaluation. The study adopts a quantitative research design to facilitate precise measurement of compression efficiency and energy savings. The target population comprises sensor nodes in an urban environmental monitoring system, with a sample of 150 sensor nodes selected through stratified random sampling across three different urban districts. Data collection involves deploying a custom-developed data compression algorithm implemented on sensor nodes equipped with open-source IoT hardware platforms, such as Arduino and Raspberry Pi, fitted with environmental sensors (temperature, humidity, air quality). The validity and reliability of the compression algorithm are established through benchmarking against state-of-the-art schemes like Huffman coding and compressive sensing, using simulation environments and real-world deployment. Data analysis encompasses the application of descriptive statistics to assess compression ratios and energy consumption metrics, followed by inferential tests including ANOVA and paired t-tests to compare the performance of the proposed scheme against existing methods. The analytical framework is grounded in the Information Theory, specifically Shannon’s source coding theorem, and the Energy-Efficient Data Transmission Model, which guide the algorithm development with an emphasis on minimizing entropy and transmission costs. Expected findings include a significant increase in compression ratios—anticipated to be between 40% and 60%—alongside reductions in energy consumption per node of approximately 30%, thereby extending sensor node operational lifetime. It is also foreseen that the scheme preserves essential data features necessary for accurate environmental monitoring, with minimal loss of data fidelity, validated through correlation analyses and domain-specific accuracy metrics. The study contributes to the body of knowledge by providing a novel, practically implementable data compression scheme explicitly optimized for energy efficiency in resource-constrained IoT environments. It advances existing theoretical frameworks by integrating principles from information theory with energy-aware network models, offering a scalable solution adaptable to various IoT contexts. Additionally, the research bridges the gap between theoretical algorithm design and real-world deployment, demonstrating tangible benefits in energy savings and data management. Concluding, the research advocates for the adoption of the proposed compression scheme in large-scale IoT deployments to enhance network sustainability and operational efficiency. Recommendations include further refinement of the algorithm to accommodate dynamic network topologies, the integration of machine learning techniques for adaptive compression, and conducting longitudinal studies to assess long-term performance. Future research avenues suggested encompass exploring compression schemes tailored for different sensor modalities and heterogeneous network environments, as well as investigating security implications associated with data compression processes. Overall, this study underscores the importance of energy-conscious data management strategies in expanding the practical viability of IoT systems and paves the way for sustainable sensor network architectures.