Multilateration is a fundamental technique used to determine the posi- tion of an object by analyzing distance measurements from multiple known points. Despite its widespread application, traditional Multilateration meth- ods are vulnerable to noisy data and can be computationally costly. This dissertation explores the DS method, which integrates Deep Set structures into neural network architectures to address these limitations. Deep Sets allow neural networks to handle a variable number of inputs without retrain- ing, by treating inputs as sets. Through experimentation with synthetic and real-world datasets, the study demonstrates that Deep Sets enhance accu- racy in extreme scenarios, showing its ability to manage missing data more effectively than conventional methods. Furthermore, our DS method showed a remarkable reduction in computational time compared to the baselines. The results indicate that DS is a highly suitable candidate for real-time and large-scale applications. The study also opens avenues for future research, including quantifying DS’s uncertainty.