An Improvised Benford Approach to Enhance Data Integrity and Cybersecurity for Residential Smart Meters

Smart meters provide a platform to facilitate bidirectional communication between customers and utility companies, increasing the functionality and efficiency of the grid and introducing vulnerabilities to cyber-attacks. These cyberattacks, such as false data injection into databases, cause inaccurate power demand forecasting, which further results in energy supply scheduling problems. This paper introduces a novel approach, the Square normalization law, to address the security issue and to enhance the security concern of residential electricity smart meters. The SNBL method utilizes a statistical phenomenon based on the principle of Benford law commonly employed in fraud detection. SNBL standardized the dataset by using squared normalization to intensify its robustness against outliers and noise. Subsequently, the expected frequency distribution of leading digits of energy profiles within the meter readings has been calculated. The integrity of the original data distribution is maintained by using SNBL, which makes it more sophisticated and challenging for unauthorized sources to draw sensitive information from the data. Various types of data attacks are simulated to evaluate Benford's law sensitivity using statistical performance indices to determine data anomaly detection. The statistical indices employed in this study to analyze the performance of SNBL are linear deviation index (LDI), relative squared deviation index (RSDI), D1I Digit 1 index, and chi-squared values. This paper aimed to safeguard the meter data against cyber attacks, artificial data manipulation, and false data injection. Furthermore, a comparative study has been conducted between Benford analysis with and without data normalization to assess its effectiveness in differentiating genuine consumption patterns from manipulated data.