Abstract:
Optimal estimation and detection of cell phone Radio Frequency (RF) - signal requires accurate techniques. Application of component improvement strategy through deployment of six-sigma tool in the design of device for the detection and location of mobile phone was studied. Multism 11.0 tools were used for the sizing of the detector components. The device circuitry comprises detector unit (signal receiver/transmitter); and display unit (showing Signal Quality (SQ)/Strength (SS)). The fabricated detector was tested using discrete and continuous techniques. Detector was analyzed using modified Neyman-Pearson (NP) approach and Artificial Intelligence (AI) tools; namely, Extreme Learning Machine (ELM) and Support Vector Machine (SVM). The models and the emerging algorithms were solved using simulation technique in MATLAB environment. The signal detection results showed improved Probability of Detection (PD) and signal Energy to Noise Ratio (ENR) with increased sigma quality. ENR improved steadily from initial level of 10 to 1000s. PD increased from 0.355 to 0.625 while Probability of
False Alarm (PFA) monotonically decreased, close to zero and the reliability of the detector improved from 0.827 per unit time to 0.987. The results of signal location when SVM and ELM were applied to SS and SQ data (using 70% for training and 30% for testing, respectively) showed high level of accuracy with correlation coefficients of 99.54% and 99.72% respectively. Results of validation using Multivariate Regression (MR) revealed better performance of the detector in terms of high correlation coefficient (0.997), coefficient of determination (0.994) and Root Mean Square Error (0.219). The findings revealed that accuracy of detection and location of mobile phone signal attained were attributed to the quality component choice of the detector circuitry. The established procedures could be tailored towards ascertaining accuracy of signal detectability of any given system to warrant prediction of failure and improvement strategy
